• Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 11471167, https://doi.org/10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Babaeian, E., M. Sadeghi, S. B. Jones, C. Montzka, H. Vereecken, and M. Tuller, 2019: Ground, proximal, and satellite remote sensing of soil moisture. Rev. Geophys., 57, 530616, https://doi.org/10.1029/2018RG000618.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Balsamo, G., and Coauthors, 2018: Satellite and in situ observations for Advancing Global Earth Surface Modelling: A review. Remote Sens., 10, 2038, https://doi.org/10.3390/rs10122038.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bauer-Marschallinger, B., and Coauthors, 2019: Toward global soil moisture monitoring with Sentinel-1: Harnessing assets and overcoming obstacles. IEEE Trans. Geosci. Remote Sens., 57, 520539, https://doi.org/10.1109/TGRS.2018.2858004.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beck, H. E., E. F. Wood, M. Pan, C. K. Fisher, D. G. Miralles, A. I. J. M. van Dijk, T. R. McVicar, and R. F. Adler, 2019: MSWEP V2 global 3-hourly 0.1° precipitation: Methodology and quantitative assessment. Bull. Amer. Meteor. Soc., 100, 473500, https://doi.org/10.1175/BAMS-D-17-0138.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beck, H. E., and Coauthors, 2020: Evaluation of 18 satellite- and model-based soil moisture products using in situ measurements from 826 sensors. Hydrol. Earth Syst. Sci., https://doi.org/10.5194/HESS-2020-184

    • Search Google Scholar
    • Export Citation
  • Bircher, S., N. Skou, K. H. Jensen, J. P. Walker, and L. Rasmussen, 2012: A soil moisture and temperature network for SMOS validation in western Denmark. Hydrol. Earth Syst. Sci., 16, 14451463, https://doi.org/10.5194/hess-16-1445-2012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bosch, D. D., J. M. Sheridan, and L. K. Marshall, 2007: Precipitation, soil moisture, and climate database, Little River Experimental Watershed, Georgia, United States. Water Resour. Res., 43, W09472, https://doi.org/10.1029/2006WR005834.

    • Search Google Scholar
    • Export Citation
  • Brodzik, M. J., B. Billingsley, T. Haran, B. Raup, and M. H. Savoie, 2012: EASE-Grid 2.0: Incremental but significant improvements for Earth-gridded data sets. ISPRS Int. J. Geoinf., 1, 3245, https://doi.org/10.3390/ijgi1010032.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Caldwell, T. G., T. Bongiovanni, M. H. Cosh, C. Halley, and M. H. Young, 2018: Field and laboratory evaluation of the CS655 soil water content sensor. Vadose Zone J., 17, 170214, https://doi.org/10.2136/vzj2017.12.0214.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chan, S. K., and Coauthors, 2016: Assessment of the SMAP passive soil moisture product. IEEE Trans. Geosci. Remote Sens., 54, 49945007, https://doi.org/10.1109/TGRS.2016.2561938.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chan, S. K., E. G. Njoku, and A. Colliander, 2018: SMAP L1C radiometer half-orbit 36 km EASE-grid brightness temperatures, version 4 NSIDC DAAC, accessed 4 March 2020, https://doi.org/10.5067/ZVILG0PS6CTI.

    • Crossref
    • Export Citation
  • Chen, F., and Coauthors, 2017: Application of triple collocation in ground-based validation of Soil Moisture Active/Passive (SMAP) level 2 data products. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 10, 489502, https://doi.org/10.1109/JSTARS.2016.2569998.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, F., W. T. Crow, R. Bindlish, A. Colliander, M. S. Burgin, J. Asanuma, and K. Aida, 2018: Global-scale evaluation of SMAP, SMOS and ASCAT soil moisture products using triple collocation. Remote Sens. Environ., 214, 113, https://doi.org/10.1016/j.rse.2018.05.008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, M., W. Shi, P. Xie, V. B. S. Silva, V. E. Kousky, R. Wayne Higgins, and J. E. Janowiak, 2008: Assessing objective techniques for gauge-based analyses of global daily precipitation. J. Geophys. Res., 113, D04110, https://doi.org/10.1029/2007JD009132.

    • Search Google Scholar
    • Export Citation
  • Clewley, D., and Coauthors, 2017: A method for upscaling in situ soil moisture measurements to satellite footprint scale using random forests. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 10, 26632673, https://doi.org/10.1109/JSTARS.2017.2690220.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Colliander, A., and Coauthors, 2017a: SMAP/in situ core validation site land surface parameters match-up data, version 1. NSIDC DAAC, accessed 25 June 2020, https://doi.org/10.5067/DXAVIXLY18KM.

    • Crossref
    • Export Citation
  • Colliander, A., and Coauthors, 2017b: Validation of SMAP surface soil moisture products with core validation sites. Remote Sens. Environ., 191, 215231, https://doi.org/10.1016/j.rse.2017.01.021.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Colliander, A., Z. Yang, R. Mueller, A. Sandborn, R. Reichle, W. Crow, D. Entekhabi, and S. Yueh, 2019: Consistency between NASS surveyed soil moisture conditions and SMAP soil moisture observations. Water Resour. Res., 55, 76827693, https://doi.org/10.1029/2018WR024475.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coopersmith, E. J., M. H. Cosh, W. A. Petersen, J. Prueger, and J. J. Niemeier, 2015: Soil moisture model calibration and validation: An ARS watershed on the South Fork Iowa River. J. Hydrometeor., 16, 10871101, https://doi.org/10.1175/JHM-D-14-0145.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cosh, M. H., T. J. Jackson, P. Starks, and G. Heathman, 2006: Temporal stability of surface soil moisture in the Little Washita River watershed and its applications in satellite soil moisture product validation. J. Hydrol., 323, 168177, https://doi.org/10.1016/j.jhydrol.2005.08.020.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cosh, M. H., P. J. Starks, J. A. Guzman, and D. N. Moriasi, 2014: Upper Washita river experimental watersheds: Multiyear stability of soil water content profiles. J. Environ. Qual., 43, 13281333, https://doi.org/10.2134/jeq2013.08.0318.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., and M. J. van den Berg, 2010: An improved approach for estimating observation and model error parameters in soil moisture data assimilation. Water Resour. Res., 46, W12519, https://doi.org/10.1029/2010WR009402.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., and Coauthors, 2012: Upscaling sparse ground-based soil moisture observations for the validation of coarse-resolution satellite soil moisture products. Rev. Geophys., 50, RG2002, https://doi.org/10.1029/2011RG000372.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., F. Chen, R. H. Reichle, and Q. Liu, 2017: L band microwave remote sensing and land data assimilation improve the representation of prestorm soil moisture conditions for hydrologic forecasting. Geophys. Res. Lett., 44, 54955503, https://doi.org/10.1002/2017GL073642.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., F. Chen, R. H. Reichle, Y. Xia, and Q. Liu, 2018: Exploiting soil moisture, precipitation, and streamflow observations to evaluate soil moisture/runoff coupling in land surface models. Geophys. Res. Lett., 45, 48694878, https://doi.org/10.1029/2018GL077193.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., F. Chen, R. H. Reichle, and Y. Xia, 2019: Diagnosing bias in modeled soil moisture/runoff coefficient correlation using the SMAP level 4 soil moisture product. Water Resour. Res., 55, 70107026, https://doi.org/10.1029/2019WR025245.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Das, N. N., and Coauthors, 2019: The SMAP and Copernicus Sentinel 1A/B microwave active-passive high resolution surface soil moisture product. Remote Sens. Environ., 233, 111380, https://doi.org/10.1016/j.rse.2019.111380.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • De Lannoy, G. J. M., R. H. Reichle, and V. R. N. Pauwels, 2013: Global calibration of the GEOS-5 L-band microwave radiative transfer model over nonfrozen land using SMOS observations. J. Hydrometeor., 14, 765785, https://doi.org/10.1175/JHM-D-12-092.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • De Lannoy, G. J. M., R. H. Reichle, and J. A. Vrugt, 2014: Uncertainty quantification of GEOS-5 L-band radiative transfer model parameters using Bayesian inference and SMOS observations. Remote Sens. Environ., 148, 146157, https://doi.org/10.1016/j.rse.2014.03.030.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., and W. T. Crow, 2017: An improved triple collocation analysis algorithm for decomposing autocorrelated and white soil moisture retrieval errors. J. Geophys. Res. Atmos., 122, 13 08113 094, https://doi.org/10.1002/2017JD027387.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., W. T. Crow, R. Reichle, Q. Liu, F. Lei, and M. H. Cosh, 2019a: A global assessment of added value in the SMAP level 4 soil moisture product relative to its baseline land surface model. Geophys. Res. Lett., 46, 66046613, https://doi.org/10.1029/2019GL083398.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., W. T. Crow, Z. Duan, L. Wei, and Y. Lu, 2019b: A double instrumental variable method for geophysical product error estimation. Remote Sens. Environ., 225, 217228, https://doi.org/10.1016/j.rse.2019.03.003.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., F. Lei, and L. Wei, 2020a: Triple collocation based multi-source precipitation merging. Front.Water, 2, 1, https://doi.org/10.3389/frwa.2020.00001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., W. T. Crow, and R. Reichle, 2020b: Improving rain/no-rain detection skill by merging precipitation estimates from different sources. J. Hydrometeor., 21, 24192429, https://doi.org/10.1175/JHM-D-20-0097.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dorigo, W. A., K. Scipal, R. M. Parinussa, Y. Y. Liu, W. Wagner, R. A. M. de Jeu, and V. Naeimi, 2010: Error characterisation of global active and passive microwave soil moisture datasets. Hydrol. Earth Syst. Sci., 14, 26052616, https://doi.org/10.5194/hess-14-2605-2010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Draper, C., R. Reichle, R. de Jeu, V. Naeimi, R. Parinussa, and W. Wagner, 2013: Estimating root mean square errors in remotely sensed soil moisture over continental scale domains. Remote Sens. Environ., 137, 288298, https://doi.org/10.1016/j.rse.2013.06.013.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ducharne, A., R. D. Koster, M. J. Suarez, M. Stieglitz, and P. Kumar, 2000: A catchment-based approach to modeling land surface processes in a general circulation model: 2. Parameter estimation and model demonstration. J. Geophys. Res., 105, 24 82324 838, https://doi.org/10.1029/2000JD900328.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Entekhabi, D., and Coauthors, 2010: The Soil Moisture Active Passive (SMAP) mission. Proc. IEEE, 98, 704716, https://doi.org/10.1109/JPROC.2010.2043918.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • EUMETSAT, 2019: Product User Manual (PUM) Metop ASCAT Soil Moisture Climate Data Record v5 12.5 km (H115) and extension (H116). EUMETSAT Satellite Application Facility on Support to Operational Hydrology and Water Management, Doc. SAF/HSAF/CDOP3/PUM/, Revision: 0.1, 32 pp., https://hsaf.meteoam.it/documents/H115_ASCAT_SSM_CDR_v5_PUM_v0.1.pdf.

  • Fang, K., M. Pan, and C. Shen, 2019: The value of SMAP for long-term soil moisture estimation with the help of deep learning. IEEE Trans. Geosci. Remote Sens., 57, 22212233, https://doi.org/10.1109/TGRS.2018.2872131.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fernandez-Moran, R., and Coauthors, 2017: SMOS-IC: An alternative SMOS soil moisture and vegetation optical depth product. Remote Sens., 9, 457, https://doi.org/10.3390/rs9050457.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ford, T. W., and S. M. Quiring, 2019: Comparison of contemporary in situ, model, and satellite remote sensing soil moisture with a focus on drought monitoring. Water Resour. Res., 55, 15651582, https://doi.org/10.1029/2018WR024039.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Galle, S., and Coauthors, 2018: AMMA-CATCH, a critical zone observatory in West Africa monitoring a region in transition. Vadose Zone J., 17, 180062, https://doi.org/10.2136/vzj2018.03.0062.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • González-Zamora, Á., N. Sánchez, J. Martínez-Fernández, Á. Gumuzzio, M. Piles, and E. Olmedo, 2015: Long-term SMOS soil moisture products: A comprehensive evaluation across scales and methods in the Duero Basin (Spain). Phys. Chem. Earth Parts ABC, 83–84, 123136, https://doi.org/10.1016/j.pce.2015.05.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gruber, A., C.-H. Su, S. Zwieback, W. Crow, W. Dorigo, and W. Wagner, 2016: Recent advances in (soil moisture) triple collocation analysis. Int. J. Appl. Earth Obs. Geoinf., 45, 200211, https://doi.org/10.1016/j.jag.2015.09.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gruber, A., and Coauthors, 2020: Validation practices for satellite soil moisture retrievals: What are (the) errors? Remote Sens. Environ., 244, 111806, https://doi.org/10.1016/j.rse.2020.111806.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Heathman, G. C., M. H. Cosh, V. Merwade, and E. Han, 2012: Multi-scale temporal stability analysis of surface and subsurface soil moisture within the Upper Cedar Creek Watershed, Indiana. Catena, 95, 91103, https://doi.org/10.1016/j.catena.2012.03.008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., R. F. Adler, D. T. Bolvin, and G. Gu, 2009: Improving the global precipitation record: GPCP version 2.1. Geophys. Res. Lett., 36, L17808, https://doi.org/10.1029/2009GL040000.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jensen, K. H., and J. C. Refsgaard, 2018: HOBE: The Danish hydrological observatory. Vadose Zone J., 17, 180059, https://doi.org/10.2136/vzj2018.03.0059.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Juglea, S., and Coauthors, 2010: Modelling soil moisture at SMOS scale by use of a SVAT model over the Valencia Anchor Station. Hydrol. Earth Syst. Sci., 14, 831846, https://doi.org/10.5194/hess-14-831-2010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Keefer, T. O., M. S. Moran, and G. B. Paige, 2008: Long-term meteorological and soil hydrology database, Walnut Gulch experimental watershed, Arizona, United States. Water Resour. Res., 44, W05S07, https://doi.org/10.1029/2006WR005702.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kerr, Y. H., and Coauthors, 2010: The SMOS mission: New tool for monitoring key elements of the global water cycle. Proc. IEEE, 98, 666687, https://doi.org/10.1109/JPROC.2010.2043032.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kerr, Y. H., and Coauthors, 2016: Overview of SMOS performance in terms of global soil moisture monitoring after six years in operation. Remote Sens. Environ., 180, 4063, https://doi.org/10.1016/j.rse.2016.02.042.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khodayar, S., A. Coll, and E. Lopez-Baeza, 2019: An improved perspective in the spatial representation of soil moisture: Potential added value of SMOS disaggregated 1 km resolution “all weather” product. Hydrol. Earth Syst. Sci., 23, 255275, https://doi.org/10.5194/hess-23-255-2019.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kimball, J. S., L. Jones, K. Jensco, M. He, M. Maneta, and R. Reichle, 2019: SMAP L4 assessment of the US northern plains 2017 flash drought. 2019 IEEE Int. Geoscience and Remote Sensing Symp., Yokohama, Japan, IEEE, 5366–5369, https://doi.org/10.1109/igarss.2019.8898354.

    • Crossref
    • Export Citation
  • Koster, R. D., M. J. Suarez, A. Ducharne, M. Stieglitz, and P. Kumar, 2000: A catchment-based approach to modeling land surface processes in a general circulation model: 1. Model structure. J. Geophys. Res., 105, 24 80924 822, https://doi.org/10.1029/2000JD900327.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 11381140, https://doi.org/10.1126/science.1100217.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2011: The second phase of the global land–atmosphere coupling experiment: Soil moisture contributions to subseasonal forecast skill. J. Hydrometeor., 12, 805822, https://doi.org/10.1175/2011JHM1365.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, X., J. Xiao, J. S. Kimball, R. H. Reichle, R. L. Scott, M. E. Litvak, G. Bohrer, and C. Frankenberg, 2020: Synergistic use of SMAP and OCO-2 data in assessing the responses of ecosystem productivity to the 2018 U.S. drought. Remote Sens. Environ., 251, 112062, https://doi.org/10.1016/j.rse.2020.112062.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lievens, H., and Coauthors, 2016: Assimilation of SMOS soil moisture and brightness temperature products into a land surface model. Remote Sens. Environ., 180, 292304, https://doi.org/10.1016/j.rse.2015.10.033.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, Q., and Coauthors, 2011: The contributions of precipitation and soil moisture observations to the skill of soil moisture estimates in a land data assimilation system. J. Hydrometeor., 12, 750765, https://doi.org/10.1175/JHM-D-10-05000.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lucchesi, R., 2018. File specification for GEOS-5 FP (Forward Processing). GMAO Office Note 4, version 1.2, 56 pp., https://gmao.gsfc.nasa.gov/pubs/docs/Lucchesi1203.pdf.

  • Matzler, C., 1998: Microwave permittivity of dry sand. IEEE Trans. Geosci. Remote Sens., 36, 317319, https://doi.org/10.1109/36.655342.

  • McNairn, H., and Coauthors, 2015: The Soil Moisture Active Passive Validation Experiment 2012 (SMAPVEX12): Prelaunch calibration and validation of the SMAP soil moisture algorithms. IEEE Trans. Geosci. Remote Sens., 53, 27842801, https://doi.org/10.1109/TGRS.2014.2364913.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Moghaddam, M., and Coauthors, 2016: Soil moisture profiles and temperature data from SoilSCAPE sites, USA. ORNL DAAC, accessed 25 June 2020, https://doi.org/10.3334/ORNLDAAC/1339.

    • Crossref
    • Export Citation
  • Pablos, M., Á. González-Zamora, N. Sánchez, and J. Martínez-Fernández, 2018: Assessment of root zone soil moisture estimations from SMAP, SMOS and MODIS observations. Remote Sens., 10, 981, https://doi.org/10.3390/rs10070981.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Panciera, R., and Coauthors, 2014: The Soil Moisture Active Passive Experiments (SMAPEx): Toward soil moisture retrieval from the SMAP mission. IEEE Trans. Geosci. Remote Sens., 52, 490507, https://doi.org/10.1109/TGRS.2013.2241774.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Parinussa, R. M., T. R. H. Holmes, N. Wanders, W. A. Dorigo, and R. A. M. de Jeu, 2015: A preliminary study toward consistent soil moisture from AMSR2. J. Hydrometeor., 16, 932947, https://doi.org/10.1175/JHM-D-13-0200.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Piepmeier, J. R., and Coauthors, 2017: SMAP L-band microwave radiometer: Instrument design and first year on orbit. IEEE Trans. Geosci. Remote Sens., 55, 19541966, https://doi.org/10.1109/TGRS.2016.2631978.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Qiu, J., J. Dong, W. T. Crow, X. Zhang, R. H. Reichle, and G. J. M. De Lannoy, 2020: The added value of brightness temperature assimilation for the SMAP level-4 surface and root-zone soil moisture analysis over mainland China. Hydrol. Earth Syst. Sci. Discuss., https://doi.org/10.5194/hess-2020-407

    • Search Google Scholar
    • Export Citation
  • Rahman, M., L. Di, E. Yu, L. Lin, C. Zhang, and J. Tang, 2019: Rapid flood progress monitoring in cropland with NASA SMAP. Remote Sens., 11, 191, https://doi.org/10.3390/rs11020191.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reichle, R. H., and R. D. Koster, 2003: Assessing the impact of horizontal error correlations in background fields on soil moisture estimation. J. Hydrometeor., 4, 12291242, https://doi.org/10.1175/1525-7541(2003)004<1229:ATIOHE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reichle, R. H., and Q. Liu, 2014: Observation-corrected precipitation estimates in GEOS-5. NASA/TM-2014-104606, Vol. 35, 18 pp., https://gmao.gsfc.nasa.gov/pubs/docs/Reichle734.pdf.

  • Reichle, R. H., J. P. Walker, R. D. Koster, and P. R. Houser, 2002: Extended vs. ensemble Kalman filtering for land data assimilation. J. Hydrometeor., 3, 728740, https://doi.org/10.1175/1525-7541(2002)003<0728:EVEKFF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reichle, R. H., and Coauthors, 2017a: Assessment of the SMAP level-4 surface and root-zone soil moisture product using in situ measurements. J. Hydrometeor., 18, 26212645, https://doi.org/10.1175/JHM-D-17-0063.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reichle, R. H., and Coauthors, 2017b: Global assessment of the SMAP level-4 surface and root-zone soil moisture product using assimilation diagnostics. J. Hydrometeor., 18, 32173237, https://doi.org/10.1175/JHM-D-17-0130.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reichle, R. H., and Coauthors, 2018a: SMAP L4 global 3-hourly 9 km EASE-grid surface and root zone soil moisture analysis update, version 4. NSIDC DAAC, accessed 6 April 2020, https://doi.org/10.5067/60HB8VIP2T8W.

    • Crossref
    • Export Citation
  • Reichle, R. H., and Coauthors, 2018b: SMAP L4 global 3-hourly 9 km EASE-grid surface and root zone soil moisture geophysical data, version 4. NSIDC DAAC, accessed 6 April 2020, https://doi.org/10.5067/KPJNN2GI1DQR.

    • Crossref
    • Export Citation
  • Reichle, R. H., Q. Liu, R. D. Koster, C. S. Draper, S. P. P. Mahanama, and G. S. Partyka, 2017c: Land surface precipitation in MERRA-2. J. Climate, 30, 16431664, https://doi.org/10.1175/JCLI-D-16-0570.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reichle, R. H., and Coauthors, 2019: Version 4 of the SMAP level-4 soil moisture algorithm and data product. J. Adv. Model. Earth Syst., 11, 31063130, https://doi.org/10.1029/2019MS001729.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rowlandson, T., S. Impera, J. Belanger, A. A. Berg, B. Toth, and R. Magagi, 2015: Use of in situ soil moisture network for estimating regional-scale soil moisture during high soil moisture conditions. Canadian Water Res. J., 40, 343351, https://doi.org/10.1080/07011784.2015.1061948.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sadri, S., E. F. Wood, and M. Pan, 2018: Developing a drought-monitoring index for the contiguous US using SMAP. Hydrol. Earth Syst. Sci., 22, 66116626, https://doi.org/10.5194/hess-22-6611-2018.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sanchez, N., J. Martinez-Fernandez, A. Scaini, and C. Perez-Gutierrez, 2012: Validation of the SMOS L2 soil moisture data in the REMEDHUS network (Spain). IEEE Trans. Geosci. Remote Sens., 50, 16021611, https://doi.org/10.1109/TGRS.2012.2186971.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Santanello, J. A., and Coauthors, 2018: Land–atmosphere interactions: The LoCo perspective. Bull. Amer. Meteor. Soc., 99, 12531272, https://doi.org/10.1175/BAMS-D-17-0001.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Scipal, K., T. Holmes, R. de Jeu, V. Naeimi, and W. Wagner, 2008: A possible solution for the problem of estimating the error structure of global soil moisture data sets. Geophys. Res. Lett., 35, L24403, https://doi.org/10.1029/2008GL035599.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Seneviratne, S. I., T. Corti, E. L. Davin, M. Hirschi, E. B. Jaeger, I. Lehner, B. Orlowsky, and A. J. Teuling, 2010: Investigating soil moisture–climate interactions in a changing climate: A review. Earth-Sci. Rev., 99, 125161, https://doi.org/10.1016/j.earscirev.2010.02.004.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Seyfried, M. S., M. D. Murdock, C. L. Hanson, G. N. Flerchinger, and S. Van Vactor, 2001: Long-term soil water content database, Reynolds Creek experimental watershed, Idaho, United States. Water Resour. Res., 37, 28472851, https://doi.org/10.1029/2001WR000419.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stillman, S., and X. Zeng, 2018: Evaluation of SMAP soil moisture relative to five other satellite products using the climate reference network measurements over USA. IEEE Trans. Geosci. Remote Sens., 56, 62966305, https://doi.org/10.1109/TGRS.2018.2835316.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Su, C.-H., D. Ryu, W. T. Crow, and A. W. Western, 2014: Beyond triple collocation: Applications to soil moisture monitoring. J. Geophys. Res., 119, 64196439, https://doi.org/10.1002/2013JD021043.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tan, J., G. J. Huffman, D. T. Bolvin, and E. J. Nelkin, 2019: IMERG V06: Changes to the morphing algorithm. J. Atmos. Oceanic Technol., 36, 24712482, https://doi.org/10.1175/JTECH-D-19-0114.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tavakol, A., V. Rahmani, S. M. Quiring, and S. V. Kumar, 2019: Evaluation analysis of NASA SMAP L3 and L4 and SPoRT-LIS soil moisture data in the United States. Remote Sens. Environ., 229, 234246, https://doi.org/10.1016/j.rse.2019.05.006.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tetlock, E., B. Toth, A. Berg, T. Rowlandson, and J. Ambadan-Thomas, 2019: An 11-yr (2007–2017) soil moisture and precipitation dataset from the Kenaston Network in the Brightwater Creek basin, Saskatchewan, Canada. Earth Syst. Sci. Data, 11, 787796, https://doi.org/10.5194/essd-11-787-2019.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thibeault, M., and Coauthors, 2015: Spatial and temporal analysis of the Monte Buey SAOCOM and SMAP core site. 2015 IEEE Int. Geoscience and Remote Sensing Symp., Milan, Italy, IEEE, 969–971, https://doi.org/10.1109/igarss.2015.7325929.

    • Crossref
    • Export Citation
  • Thomas, M. A., B. D. Collins, and B. B. Mirus, 2019: Assessing the feasibility of satellite-based thresholds for hydrologically driven landsliding. Water Resour. Res., 55, 90069023, https://doi.org/10.1029/2019WR025577.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tobin, K. J., W. T. Crow, J. Dong, and M. E. Bennett, 2019: Validation of a new root-zone soil moisture product: Soil MERGE. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 12, 33513365, https://doi.org/10.1109/JSTARS.2019.2930946.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wagner, W., and Coauthors, 2013: The ASCAT soil moisture product: A review of its specifications, validation results, and emerging applications. Meteor. Z., 22, 533, https://doi.org/10.1127/0941-2948/2013/0399.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Xie, P., M. Chen, S. Yang, A. Yatagai, T. Hayasaka, Y. Fukushima, and C. Liu, 2007: A gauge-based analysis of daily precipitation over East Asia. J. Hydrometeor., 8, 607626, https://doi.org/10.1175/JHM583.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yi, L., W. Zhang, and X. Li, 2018: Assessing hydrological modelling driven by different precipitation datasets via the SMAP soil moisture product and gauged streamflow data. Remote Sens., 10, 1872, https://doi.org/10.3390/rs10121872.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yi, Y., J. S. Kimball, R. H. Chen, M. Moghaddam, R. H. Reichle, U. Mishra, D. Zona, and W. C. Oechel, 2018: Characterizing permafrost active layer dynamics and sensitivity to landscape spatial heterogeneity in Alaska. Cryosphere, 12, 145161, https://doi.org/10.5194/tc-12-145-2018.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yi, Y., J. S. Kimball, R. H. Chen, M. Moghaddam, and C. E. Miller, 2019: Sensitivity of active-layer freezing process to snow cover in Arctic Alaska. Cryosphere, 13, 197218, https://doi.org/10.5194/tc-13-197-2019.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery
    Fig. 1.

    (a) April 2015–March 2019 average precipitation gauge density for the CPCU precipitation product. (b) Weight given to the CPCU product in the L4_SM and CPCU_SIM corrected precipitation forcing. (c) Effective gauge density in the L4_SM and CPCU_SIM corrected precipitation forcing, computed as the product of the CPCU gauge density in (a) with the weight in (b). Gauge density is derived as the time-average number of gauges within each 1.5° grid cell (consisting of three-by-three 0.5° CPCU grid cells).

  • View in gallery
    Fig. 2.

    Overview of model-only (CTRL, CPCU_SIM) and data assimilation (SMAP_DA, L4_SM) experiments.

  • View in gallery
    Fig. 3.

    (a) ubRMSE (m3 m−3), (b) R (dimensionless), and (c) anomaly R (dimensionless) for CTRL, SMAP_DA, CPCU_SIM, and L4_SM estimates of surface and root-zone soil moisture vs in situ measurements, averaged across 18 core validation sites (Table 1). Error bars indicate 95% confidence intervals.

  • View in gallery
    Fig. 4.

    Anomaly R skill differences for surface soil moisture at the SMAP core validation site locations. Anomaly R values were computed (gray bars) vs in situ measurements from the core validation sites and (black bars) from the IV method using ASCAT surface soil moisture retrievals. Bars show anomaly R differences averaged across 17 of the 18 core sites listed in Table 1 (excluding Niger; see text for details).

  • View in gallery
    Fig. 5.

    Surface soil moisture anomaly R skill differences for (a) L4_SM minus CTRL, (b) L4_SM minus CPCU_SIM, and (c) L4_SM minus SMAP_DA. Skill is measured in terms of anomaly R (dimensionless). Panel (a) shows the total contribution of CPCU precipitation and SMAP Tb data to L4_SM skill. Panel (b) shows the additional contribution of SMAP Tb assimilation to L4_SM skill on top of CPCU-based precipitation corrections. Panel (c) shows the additional contribution of precipitation corrections on top of SMAP Tb assimilation. Across all panels, red colors indicate that L4_SM is better and blue colors indicate that L4_SM is worse than the respective reference experiment. White shading indicates glaciated land or land for which skill metrics could not be computed (section 3b). Spatial averages (avg) are computed across all nonwhite land areas.

  • View in gallery
    Fig. 6.

    Surface soil moisture anomaly R skill differences for (a) SMAP_DA minus CTRL and (b) CPCU_SIM minus CTRL. Skill is measured in terms of anomaly R (dimensionless). Panel (a) shows the contribution of SMAP Tb assimilation to L4_SM skill in the absence of CPCU-based precipitation corrections. Panel (b) shows the contribution of precipitation corrections in the absence of SMAP Tb assimilation. Red colors indicate that SMAP_DA (or CPCU_SIM) is better and blue colors indicate that SMAP_DA (or CPCU_SIM) is worse than CTRL. White shading and spatial averages (avg) are as in Fig. 5.

  • View in gallery
    Fig. 7.

    Impact of CPCU-based precipitation corrections on surface soil moisture skill improvement from SMAP Tb assimilation: skill difference from SMAP assimilation without precipitation corrections (SMAP_DA minus CTRL) minus skill difference from SMAP assimilation with precipitation corrections (L4_SM minus CPCU_SIM). Yellow and red colors indicate where SMAP Tb assimilation and CPCU precipitation contribute overlapping information to the soil moisture estimates. Blue colors indicate where SMAP Tb assimilation compensates for soil moisture skill degradation caused by CPCU-based precipitation corrections. Skill is the anomaly R for surface soil moisture. White shading and spatial average (avg) are as in Fig. 5.

  • View in gallery
    Fig. 8.

    (a) Time series standard deviation of observations-minus-forecast (O-F) Tb residuals from the L4_SM product. (b) Difference in time series standard deviation of O-F Tb residuals between L4_SM and SMAP_DA. Blue colors in (b) indicate that typical O-F residuals are smaller (better) for L4_SM than for SMAP_DA. White shading indicates glaciated land or land for which the diagnostic could not be computed because an insufficient number of SMAP observations was assimilated. Spatial averages (avg) are computed across all nonwhite land areas.

  • View in gallery
    Fig. 9.

    (a) Time series standard deviation of surface soil moisture (SFSM) increments from the L4_SM product. (b) Difference in time series standard deviation of SFSM increments between L4_SM and SMAP_DA. (c),(d) As in (a) and (b), respectively, but for root-zone soil moisture (RZSM) increments. Blue colors in (b) and (d) indicate that typical soil moisture increments are smaller (better) for L4_SM than for SMAP_DA. White shading and the spatial average (avg) are as in Fig. 8.

  • View in gallery
    Fig. 10.

    Streamflow skill differences for (left) absolute value of bias and (right) R values for (a),(b) L4_SM minus CTRL; (c),(d) L4_SM minus CPCU_SIM; and (e),(f) L4_SM minus SMAP_DA. Panels (a) and (b) show the total contribution of CPCU precipitation and SMAP Tb data to L4_SM skill. Panels (c) and (d) show the additional contribution of SMAP Tb assimilation to L4_SM skill on top of CPCU-based precipitation corrections. Panels (e) and (f) show the additional contribution of precipitation corrections on top of SMAP Tb assimilation. Across all panels, blue colors indicate that L4_SM is better and red colors indicate that L4_SM is worse than the respective reference experiment. Note the different color axis limits in (c) and (d).

All Time Past Year Past 30 Days
Abstract Views 792 0 0
Full Text Views 740 488 34
PDF Downloads 530 305 22

The Contributions of Gauge-Based Precipitation and SMAP Brightness Temperature Observations to the Skill of the SMAP Level-4 Soil Moisture Product

Rolf H. Reichle Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, Maryland

Search for other papers by Rolf H. Reichle in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0001-5513-0150
,
Qing Liu Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, Maryland
Science Systems and Applications, Inc., Lanham, Maryland

Search for other papers by Qing Liu in
Current site
Google Scholar
PubMed
Close
,
Joseph V. Ardizzone Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, Maryland
Science Systems and Applications, Inc., Lanham, Maryland

Search for other papers by Joseph V. Ardizzone in
Current site
Google Scholar
PubMed
Close
,
Wade T. Crow Hydrology and Remote Sensing Laboratory, Agricultural Research Service, USDA, Beltsville, Maryland

Search for other papers by Wade T. Crow in
Current site
Google Scholar
PubMed
Close
,
Gabrielle J. M. De Lannoy Department of Earth and Environmental Sciences, Katholieke Universiteit Leuven, Heverlee, Belgium

Search for other papers by Gabrielle J. M. De Lannoy in
Current site
Google Scholar
PubMed
Close
,
Jianzhi Dong Hydrology and Remote Sensing Laboratory, Agricultural Research Service, USDA, Beltsville, Maryland

Search for other papers by Jianzhi Dong in
Current site
Google Scholar
PubMed
Close
,
John S. Kimball Numerical Terradynamic Simulation Group, University of Montana, Missoula, Montana

Search for other papers by John S. Kimball in
Current site
Google Scholar
PubMed
Close
, and
Randal D. Koster Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, Maryland

Search for other papers by Randal D. Koster in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Soil Moisture Active Passive (SMAP) mission L-band brightness temperature (Tb) observations are routinely assimilated into the Catchment land surface model to generate Level-4 soil moisture (L4_SM) estimates of global surface and root-zone soil moisture at 9-km, 3-hourly resolution with ~2.5-day latency. The Catchment model in the L4_SM algorithm is driven with 1/4°, hourly surface meteorological forcing data from the Goddard Earth Observing System (GEOS). Outside of Africa and the high latitudes, GEOS precipitation is corrected using Climate Prediction Center Unified (CPCU) gauge-based, 1/2°, daily precipitation. L4_SM soil moisture was previously shown to improve over land model-only estimates that use CPCU precipitation but no Tb assimilation (CPCU_SIM). Here, we additionally examine the skill of model-only (CTRL) and Tb assimilation-only (SMAP_DA) estimates derived without CPCU precipitation. Soil moisture is assessed versus in situ measurements in well-instrumented regions and globally through the instrumental variable (IV) method using independent soil moisture retrievals from the Advanced Scatterometer. At the in situ locations, SMAP_DA and CPCU_SIM have comparable soil moisture skill improvements relative to CTRL for the unbiased root-mean-square error (surface and root-zone) and correlation metrics (root-zone only). In the global average, SMAP Tb assimilation increases the surface soil moisture anomaly correlation by 0.10–0.11 compared to an increase of 0.02–0.03 from the CPCU-based precipitation corrections. The contrast is particularly strong in central Australia, where CPCU is known to have errors and observation-minus-forecast Tb residuals are larger when CPCU precipitation is used. Validation versus streamflow measurements in the contiguous United States reveals that CPCU precipitation provides most of the skill gained in L4_SM runoff estimates over CTRL.

Significance Statement

Soil moisture provides an important connection between the land surface water, energy, and carbon cycles. By routinely merging Soil Moisture Active Passive (SMAP) satellite observations and gauge-based precipitation data into a numerical model of land surface water and energy balance processes, NASA generates the global, 9-km resolution, 3-hourly Level-4 soil moisture (L4_SM) data product, which is published with ~2.5-day latency to support Earth science and applications, e.g., drought monitoring. Using additional model simulations and validation against independent measurements, we find that the SMAP and precipitation data contribute similarly, and largely independently, to the accuracy of the L4_SM product. SMAP’s contribution to L4_SM accuracy is particularly large in poorly instrumented regions, including portions of South America, Africa, and central Australia.

Supplemental information related to this paper is available at the Journals Online website: https://doi.org/10.1175/JHM-D-20-0217.s1.

© 2021 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Rolf H. Reichle, rolf.reichle@nasa.gov

Abstract

Soil Moisture Active Passive (SMAP) mission L-band brightness temperature (Tb) observations are routinely assimilated into the Catchment land surface model to generate Level-4 soil moisture (L4_SM) estimates of global surface and root-zone soil moisture at 9-km, 3-hourly resolution with ~2.5-day latency. The Catchment model in the L4_SM algorithm is driven with 1/4°, hourly surface meteorological forcing data from the Goddard Earth Observing System (GEOS). Outside of Africa and the high latitudes, GEOS precipitation is corrected using Climate Prediction Center Unified (CPCU) gauge-based, 1/2°, daily precipitation. L4_SM soil moisture was previously shown to improve over land model-only estimates that use CPCU precipitation but no Tb assimilation (CPCU_SIM). Here, we additionally examine the skill of model-only (CTRL) and Tb assimilation-only (SMAP_DA) estimates derived without CPCU precipitation. Soil moisture is assessed versus in situ measurements in well-instrumented regions and globally through the instrumental variable (IV) method using independent soil moisture retrievals from the Advanced Scatterometer. At the in situ locations, SMAP_DA and CPCU_SIM have comparable soil moisture skill improvements relative to CTRL for the unbiased root-mean-square error (surface and root-zone) and correlation metrics (root-zone only). In the global average, SMAP Tb assimilation increases the surface soil moisture anomaly correlation by 0.10–0.11 compared to an increase of 0.02–0.03 from the CPCU-based precipitation corrections. The contrast is particularly strong in central Australia, where CPCU is known to have errors and observation-minus-forecast Tb residuals are larger when CPCU precipitation is used. Validation versus streamflow measurements in the contiguous United States reveals that CPCU precipitation provides most of the skill gained in L4_SM runoff estimates over CTRL.

Significance Statement

Soil moisture provides an important connection between the land surface water, energy, and carbon cycles. By routinely merging Soil Moisture Active Passive (SMAP) satellite observations and gauge-based precipitation data into a numerical model of land surface water and energy balance processes, NASA generates the global, 9-km resolution, 3-hourly Level-4 soil moisture (L4_SM) data product, which is published with ~2.5-day latency to support Earth science and applications, e.g., drought monitoring. Using additional model simulations and validation against independent measurements, we find that the SMAP and precipitation data contribute similarly, and largely independently, to the accuracy of the L4_SM product. SMAP’s contribution to L4_SM accuracy is particularly large in poorly instrumented regions, including portions of South America, Africa, and central Australia.

Supplemental information related to this paper is available at the Journals Online website: https://doi.org/10.1175/JHM-D-20-0217.s1.

© 2021 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Rolf H. Reichle, rolf.reichle@nasa.gov

1. Introduction

Soil moisture is important because it connects the land surface water, energy, and carbon cycles (Seneviratne et al. 2010). Accurate, long-term, global observations of soil moisture conditions are critical for a wide range of science investigations and applications (Balsamo et al. 2018; Santanello et al. 2018). A variety of global satellite soil moisture data products are available based on microwave observations, including from the Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E) and its successor, AMSR2; the Advanced Scatterometer (ASCAT); the Soil Moisture Ocean Salinity (SMOS) mission; the Soil Moisture Active Passive (SMAP) mission; and the Sentinel-1 mission (e.g., Wagner et al. 2013; Parinussa et al. 2015; Chan et al. 2016; Kerr et al. 2016; Fernandez-Moran et al. 2017; Babaeian et al. 2019; Bauer-Marschallinger et al. 2019; Das et al. 2019). The L-band (1.4 GHz) passive microwave brightness temperature (Tb) observations collected by SMOS (since 2010) and SMAP (since 2015) are particularly sensitive to near-surface soil moisture (Entekhabi et al. 2010; Kerr et al. 2010).

The SMAP mission routinely generates a suite of global soil moisture data products, including Level-2/3 soil moisture retrievals (Chan et al. 2016; Das et al. 2019) and the value-added Level-4 soil moisture (L4_SM) product (Reichle et al. 2019). The latter is based on the assimilation of SMAP Tb observations into the NASA Catchment land surface model using an ensemble Kalman filter (Reichle et al. 2017b). The resulting global estimates of surface (0–5 cm) and root-zone (0–100 cm) soil moisture, along with a variety of related land surface variables, are published at 9-km, 3-hourly resolution with ~2.5-day mean latency. In effect, the L4_SM product interpolates and extrapolates the information provided by the SMAP Tb observations in time and in space.

The SMAP L4_SM product has been widely used in research and applications. For instance, Crow et al. (2017) found that prestorm surface soil moisture estimates from the L4_SM product can better predict the event-scale runoff coefficient than do soil moisture retrievals from AMSR-E, SMOS, or SMAP. The L4_SM product therefore supports an improved observational benchmark for the evaluation of runoff processes in land surface models (Crow et al. 2018, 2019), and thus should ultimately support improved flood predictions. Colliander et al. (2019) demonstrated that L4_SM soil moisture estimates are consistent with routine reports from agricultural surveyors and can therefore be used to create value-added maps of cropland soil moisture conditions (https://cloud.csiss.gmu.edu/Crop-CASMA/) and state-level statistics at daily intervals. Other examples of potential L4_SM applications include drought monitoring (Sadri et al. 2018; Kimball et al. 2019; Li et al. 2020), modeling of permafrost and cold season soil respiration (Y. Yi et al. 2018, 2019), monitoring of agricultural flooding (Rahman et al. 2019), the evaluation of other soil moisture products (L. Yi et al. 2018; Tobin et al. 2019), landslide monitoring (Thomas et al. 2019), and extending the process-based improvements of the L4_SM system into periods for which SMAP data are not available (Fang et al. 2019). Generally favorable validation results corroborate the suitability of the L4_SM product for use in research and applications (Reichle et al. 2017a,b, 2019; Pablos et al. 2018; Stillman and Zeng 2018; Ford and Quiring 2019; Tavakol et al. 2019; Qiu et al. 2020).

Besides the expanded coverage in the vertical, horizontal, and temporal dimensions compared to Level-2 retrievals, the L4_SM product also benefits from additional observational information, primarily in the form of the surface meteorological forcing data used to drive the Catchment model. The near-surface meteorological forcing data used in the L4_SM algorithm are from the NASA Goddard Earth Observing System (GEOS) Forward Processing product, a global, near-real-time, atmospheric analysis that assimilates ~4 million conventional and satellite observations of the atmosphere every 6 h (Lucchesi 2018). Outside of Africa and the high latitudes, the L4_SM algorithm further uses gauge-based daily precipitation from the NOAA Climate Prediction Center Unified (CPCU) product (Xie et al. 2007; Chen et al. 2008) to correct the GEOS precipitation estimates.

Several studies documented the impact of SMAP Tb assimilation on the skill of the L4_SM soil moisture data compared to estimates from a model-only, “open loop” simulation with CPCU precipitation information but without Tb assimilation. Using high-quality in situ measurements from SMAP core validation sites, Reichle et al. (2017a, 2019) found that the L4_SM product meets its accuracy requirement only with the assimilation of SMAP Tb observations; the model-only estimates do not meet this requirement. The same studies also reported statistically significant improvements in the (temporal) correlation skill of the L4_SM product compared to that of model-only estimates. Moreover, Dong et al. (2019a) used independent ASCAT soil moisture retrievals to derive a global map of skill improvement relative to a model-only baseline and demonstrated that the L4_SM product provides the most value in data-poor and lightly vegetated regions, including much of Africa and central Australia. Consequently, L4_SM assessments based on in situ measurements, which are concentrated in data-rich regions such as the United States and Europe, likely underestimated the contribution of SMAP Tb observations to the skill of the L4_SM product.

The L4_SM skill improvements documented by Reichle et al. (2017a, 2019) and Dong et al. (2019a) were relative to a model-only baseline driven with the same surface meteorological forcing data, including the same CPCU precipitation information, that were also used to derive the L4_SM product. These earlier studies thus do not quantify the individual contributions of the assimilated Tb observations and the gauge-based precipitation to the skill of the L4_SM soil moisture estimates. Such a skill breakdown was investigated by Liu et al. (2011) for the assimilation of AMSR-E soil moisture retrievals using a predecessor of the L4_SM algorithm. Using in situ soil moisture measurements in the contiguous United States (CONUS), Liu et al. (2011) found that gauge-based precipitation corrections and the assimilation of AMSR-E soil moisture retrievals contribute similar and largely independent amounts of information to the anomaly correlation skill of the resulting soil moisture estimates. In fact, the use of CPCU precipitation in the L4_SM algorithm was motivated by the results of Liu et al. (2011). However, the model, the forcing data, the assimilated observations, and the analysis system in the L4_SM algorithm have all changed considerably since Liu et al. (2011), which, in any case, was limited to CONUS. It is therefore important to take a fresh look at the impact of gauge-based precipitation observations in the state-of-the-art, global L4_SM system.

The primary objective of the present study is to quantify the individual contributions of the SMAP Tb observations and CPCU precipitation to the skill of the L4_SM soil moisture and runoff estimates at the continental-to-global scale. To this end, we generated new soil moisture and runoff estimates without the gauge-based precipitation corrections, both without (model-only) and with SMAP Tb assimilation. Skill differences between the various soil moisture estimates are assessed versus in situ measurements and, at the global scale, through an instrumental variable (IV) method (Su et al. 2014) using independent ASCAT soil moisture retrievals. We further use diagnostics of the data assimilation system to assess the impact of the gauge-based precipitation corrections on the SMAP Tb analysis. Finally, runoff skill differences are assessed using streamflow observations from unregulated basins across CONUS. The manuscript is organized as follows: The L4_SM product and experimental datasets assessed here are described in section 2, the validation data and methods are outlined in section 3, and results are presented and discussed in section 4. Finally, section 5 summarizes the paper and provides directions for future study.

2. L4_SM data and experiments

a. L4_SM data product

The present paper uses version 4 of the L4_SM product, which features several improvements in the land surface modeling system and the ensemble-based Tb analysis (Reichle et al. 2019). This section briefly summarizes the detailed descriptions of the L4_SM algorithm and data product provided by Reichle et al. (2017a,b, 2019).

The L4_SM algorithm (Reichle et al. 2017a, their Fig. 1) assimilates horizontal- and vertical-polarization Tb observations from ascending and descending half-orbits of the 36-km resolution SMAP Level-1C product (Piepmeier et al. 2017; Chan et al. 2018) into the Catchment land surface model (Koster et al. 2000; Ducharne et al. 2000) using a spatially distributed ensemble Kalman filter (EnKF; Reichle et al. 2017b). The Catchment model is supplemented with a tau–omega radiative transfer model for L-band Tb (De Lannoy et al. 2013, 2014) and set up on the global, 9-km resolution Equal Area Scalable Earth, version 2 (EASEv2) grid (Brodzik et al. 2012). The L4_SM product provides a variety of land surface fields, including surface (0–5 cm) and root-zone (0–100 cm) soil moisture, soil temperature, and surface fluxes. The L4_SM product also provides important data assimilation diagnostics, including the assimilated Tb observations and corresponding model forecasts. Here, we use 3-hourly instantaneous surface and root-zone soil moisture and brightness temperature from the L4_SM “analysis-update” files (Reichle et al. 2018a). We further use 3-hourly time-average total runoff data (including surface runoff and baseflow) from the L4_SM “geophysical” files (Reichle et al. 2018b) with Science Version ID Vv4030.

1) Surface meteorological forcing and CPCU-based precipitation corrections

The Catchment model in the L4_SM algorithm is driven with hourly, 0.25° × 0.3125° (latitude × longitude) resolution surface meteorological forcing data from the GEOS Forward Processing system (Lucchesi 2018). The GEOS precipitation is determined through the application of complex model parameterizations to the abovementioned atmospheric analysis, including three-dimensional fields of pressure, temperature, and humidity; this model-generated precipitation is then corrected with gauge-based precipitation observations from the CPCU product (Xie et al. 2007; Chen et al. 2008), separately for each 0.5° CPCU grid cell and each 24-h period (as defined by the reporting time in the CPCU product). That is, the resulting total daily precipitation at the 0.5° scale matches that of the CPCU product, and the variations at the subdiurnal and sub-0.5° scales are determined by the relative variations in the GEOS product. Prior to this grid cell-wise, daily correction, both the GEOS and CPCU precipitation are scaled to the pentad, 2.5° climatology of the Global Precipitation Climatology Project, version 2.2 (GPCPv2.2; Adler et al. 2003; Huffman et al. 2009) data, because GPCPv2.2 was considered to be the best available long-term global precipitation product when the L4_SM algorithm was first developed.

The spatial density of precipitation gauges that contribute to the CPCU product varies tremendously across the globe, with few gauges available in Africa and the high latitudes (Fig. 1a). Consequently, in Africa and poleward of 62.5° latitude, the Catchment model in the L4_SM system is forced with the uncorrected GEOS precipitation (Fig. 1b). Between 42.5° and 62.5° latitude (in the Northern and Southern Hemispheres), the precipitation corrections are linearly tapered between full corrections (at 42.5° latitude) and no corrections (at 62.5° latitude). That is, the precipitation corrections are applied in full only within 42.5° latitude from the equator except in Africa. Finally, the degree to which CPCU precipitation is based on gauge observations and used in the L4_SM algorithm is illustrated in Fig. 1c. See Reichle and Liu (2014) and Reichle et al. (2017a, 2019) for further details on the precipitation correction algorithm.

Fig. 1.
Fig. 1.

(a) April 2015–March 2019 average precipitation gauge density for the CPCU precipitation product. (b) Weight given to the CPCU product in the L4_SM and CPCU_SIM corrected precipitation forcing. (c) Effective gauge density in the L4_SM and CPCU_SIM corrected precipitation forcing, computed as the product of the CPCU gauge density in (a) with the weight in (b). Gauge density is derived as the time-average number of gauges within each 1.5° grid cell (consisting of three-by-three 0.5° CPCU grid cells).

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

2) SMAP Tb analysis

The EnKF analysis in the L4_SM system is designed to make the modeled soil moisture and soil temperature fields consistent with the SMAP Tb observations within their respective uncertainties. The observation error standard deviation is set to a constant value of 4 K, which mostly accounts for representativeness error (Reichle et al. 2017a). Uncertainty in the Catchment model ensemble simulations is represented by the ensemble spread and maintained by repeatedly adding perturbations to the model forcing and prognostic variables as described in Reichle et al. (2017a). See also Reichle et al. (2017b, 2019) for details and results regarding the observation and modeling uncertainties.

Every 3 h, the SMAP Tb observations within a 3-h window centered on the analysis times (0000, 0300, …, 2100 UTC) are compared with the corresponding model forecast Tb estimates. More specifically, only Tb anomalies from the respective multiyear Tb climatologies are compared such that the analysis corrects only for errors in short-term and interannual soil moisture variations but not for errors in the climatological seasonal cycle (Reichle et al. 2017b). The resulting soil moisture and soil temperature analysis increments are then added to the model forecast estimates to reinitialize the model and continue the simulation until the next analysis time. Note that the analysis exploits error correlations between the surface and root-zone reservoirs to adjust the modeled root-zone soil moisture (Reichle et al. 2017b). By design, the EnKF analysis does not conserve water at any given analysis time and location. However, the careful calibration of the L4_SM system ensures that the temporally averaged analysis increments are acceptably small at any given location and vanish in the global average (Reichle et al. 2019, their Figs. 6c,e).

3) Pathways of observation impact

During the assimilation process, the L4_SM algorithm spatially and temporally interpolates and extrapolates information from the SMAP Tb observations and the model forecast estimates, which are themselves informed by CPCU precipitation; the resulting L4_SM data product represents the merged information. The distance (in space and in time) from the assimilated SMAP Tb observations determines their impact on the L4_SM soil moisture estimates at a given time and location (Reichle and Koster 2003). Moreover, CPCU precipitation is never used in Africa and in the high latitudes, and it has no impact on L4_SM estimates there (Fig. 1c). Additional factors govern the impact of SMAP Tb assimilation on the L4_SM estimates, including the sensitivity of the Tb observations to surface soil moisture and the coupling between the surface and root-zone layers. Under dense vegetation, for instance, or when the ground is frozen or snow-covered, SMAP Tb observations have little or no immediate impact on the L4_SM soil moisture (Reichle et al. 2017b). Likewise, if the surface and root-zone layers are decoupled, SMAP Tb observations have little immediate impact on L4_SM root-zone soil moisture (e.g., Reichle et al. 2002, their Fig. 7). Finally, surface conditions, including vegetation and soil properties, modulate the impact of the CPCU precipitation on the L4_SM soil moisture estimates.

b. Experiments

Besides the L4_SM data product, this paper investigates output from three additional experimental simulations (Fig. 2). Two of the experiments are model-only simulations, that is, without SMAP Tb assimilation. The first model-only simulation, CTRL, uses precipitation forcing from GEOS only, that is, without CPCU-based corrections. The second model-only simulation, CPCU_SIM, is identical to the CTRL simulation but with the addition of the CPCU-based precipitation corrections. Finally, the third experimental simulation, SMAP_DA, is identical to the CTRL simulation but with the addition of SMAP Tb assimilation. Put differently, CPCU_SIM does not assimilate SMAP Tb observations, and SMAP_DA does not use CPCU precipitation. The L4_SM product, of course, uses both CPCU precipitation and SMAP Tb observations. Using the experimental simulations as reference, it is straightforward to determine the individual contributions of the CPCU precipitation and SMAP Tb assimilation to the skill of the L4_SM product.

Fig. 2.
Fig. 2.

Overview of model-only (CTRL, CPCU_SIM) and data assimilation (SMAP_DA, L4_SM) experiments.

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

The three experiments (CTRL, CPCU_SIM, and SMAP_DA) cover the 4-yr period from April 2015 to March 2019 and perfectly match the configuration of the L4_SM algorithm unless noted otherwise above. That is, the experiments are all at 9-km resolution, and in all three cases the precipitation forcing was scaled to the 2.5°, pentad GPCPv2.2 climatology (section 2a), so that all experiments share the same, seasonally varying precipitation climatology. This implies that the assimilated Tb observations in SMAP_DA can be rescaled using the same climatology as in L4_SM (section 2a). Moreover, SMAP_DA uses the same observation and model error parameters as L4_SM. The latter two assumptions are reasonable but arguably somewhat suboptimal settings for SMAP_DA. Finally, the model-only experiments (CTRL, CPCU_SIM) are ensemble simulations with perturbations to the model forcing and prognostics according to the L4_SM settings. Note that this is a minor difference from the model-only “Nature Run” simulations used by Reichle et al. (2017a,b, 2019), which are single-member simulations without perturbations. That is, the CPCU_SIM experiment used here is the ensemble version of the model-only NR4030 simulation of Reichle et al. (2019). We used the ensemble-average soil moisture values from CTRL and CPCU_SIM to compute the model-only skill metrics. Using the unperturbed NR4030 simulation instead of CPCU_SIM yielded only minimal changes in the results (not shown).

3. Validation data and approach

a. In situ soil moisture measurements

We use independent in situ measurements from the same SMAP core validation sites that were used by Reichle et al. (2019) and only briefly summarize their discussion of the data and validation approach here. Table 1 provides an overview of the site characteristics, along with references for each site, and Fig. S1 in the online supplemental material shows a map of the locations. The sites span a variety of climate regimes and land cover conditions. Each site features a locally dense network of sensor profiles (Colliander et al. 2017a,b) to alleviate the upscaling errors (Crow et al. 2012) that typically affect the validation of model output (here representing average soil moisture conditions in the 81 km2 area of an EASEv2 grid cell). At each core validation site, the in situ measurements from the individual sensor profiles are horizontally averaged across one or more 9-km EASEv2 grid cells (or reference pixels).

Table 1.

Soil moisture core validation sites used in the present paper. A map and further details are provided in Fig. S1 and Table S1.

Table 1.

The present study uses 4 years of data (April 2015–March 2019) where available, as opposed to the 3-yr period ending in March 2018 used by Reichle et al. (2019). The present study further uses root-zone soil moisture measurements at the Tonzi site in California that had not been available in Reichle et al. (2019). That is, here we use surface (root-zone) measurements from 18 (7) different core validation sites. Across all sites, we use surface (root-zone) measurements for a total of 31 (13) 9-km grid cells (or reference pixels). We also validated the L4_SM and other model data using SMAP core validation site measurements for 33-km reference pixels and measurements from the sparse networks used by Reichle et al. (2017a). This separate validation (not shown) yielded qualitatively very similar results.

Validation metrics used here include the unbiased root-mean-square error (ubRMSE), time series correlation (R), and anomaly R, all computed as in Reichle et al. (2019). The ubRMSE measures the RMSE after removing the mean difference, and the anomaly R measures the time series correlation after removing the multiyear mean monthly climatological cycle. Averages of in situ–based metrics are computed by first averaging the metrics of all 9-km reference grid cells (weighted by the number of in situ measurements used) within a given site and then computing a nonweighted average across all sites. Statistical uncertainty in the ubRMSE, R, and anomaly R metrics is estimated with 95% confidence intervals as in Reichle et al. (2019), including a correction for temporal autocorrelation. Changes in metrics are considered statistically significant when there is no overlap in the 95% confidence intervals.

b. ASCAT soil moisture retrievals

ASCAT surface soil moisture retrievals are derived from observations of the C-band (5.3 GHz) backscatter coefficient using a semiempirical change detection approach (Wagner et al. 2013). Here, we used dataset versions H115 (April 2015–December 2018) and H116 (January–March 2019) (EUMETSAT 2019). The ASCAT retrievals are in relative saturation units, which range between 0% and 100%, with the extreme values representing the driest and wettest observations at each location, respectively. ASCAT data are only used to compute correlations (section 3d) and thus used in their original units. The ASCAT data have a resolution of ~25 km and were regridded using natural-neighbor interpolation to the 9-km EASEv2 grid for this analysis. ASCAT surface soil moisture retrievals typically represent soil moisture conditions in the ~0–1-cm surface layer. For this study, we averaged ASCAT soil moisture retrievals from morning and evening overpasses and used them together with daily-average L4_SM model data in the IV analysis (section 3d). Averaging the model data only over the ASCAT overpass times or computing metrics separately for morning and evening overpasses resulted in only very minor changes in the results (not shown).

Quality control of the ASCAT retrievals was based on (i) flags supplied with the ASCAT product, (ii) L4_SM estimates, and (iii) consistency checks. We used only ASCAT retrievals with a favorable surface state flag (SSF = 1), thereby excluding frozen surfaces, temporary melting, standing water, or permanent ice. Moreover, we used only ASCAT retrievals with a favorable confidence flag (CONF_FLAG = 0), thereby excluding data for which the topographic complexity is greater than 50%, the wetland fraction is greater than 50%, soil moisture noise is greater than 50%, or soil moisture sensitivity is less than 1 dB. We further excluded ASCAT retrievals when and where L4_SM indicated that the surface soil temperature was near or below freezing or the ground was partially or fully covered with snow. Finally, we excluded locations for which the IV-based estimate of the anomaly R skill of the ASCAT retrievals (section 3d) was less than 0.1 (for any of the ASCAT-experiment pairs) because the IV method requires at least some skill in the independent dataset. For the same reason, we also excluded locations where the cross-correlation of ASCAT and L4_SM (or experiment) anomalies or the 2-day lagged autocorrelation of ASCAT anomalies was negative.

c. Streamflow measurements

To examine the impact of SMAP Tb assimilation and CPCU precipitation on the skill of the modeled runoff estimates (section 4e), we use streamflow measurements published by the U.S. Geological Survey (USGS) for 237 unregulated hydrological basins of intermediate size (2000–10 000 km2) within CONUS. These measurements are extended series of those used by Reichle et al. (2019), except that two basins were dropped here because of irregularly formatted data files. We refer the reader to Reichle et al. (2019) for details and only briefly summarize their discussion here. The published streamflow measurements were converted to units of millimeters per day (mm day−1) through normalization with the basin area, and the 9-km model total runoff estimates were spatially aggregated for each the 237 basins. A 10-day running mean was applied to the USGS measurements and to the modeled runoff estimates to account for the lack of a routing model in the L4_SM system. As for soil moisture, the evaluation period here spans 4 years (2015–18), but only the warm season (June–September) of each year is included to avoid periods when runoff is dominated by snowmelt. The (raw) R metric computed for runoff is sensitive to climatological seasonal variations within the warm season as well as to intraseasonal and interannual variations.

d. Instrumental variable approach

We use the single IV approach of Su et al. (2014) to determine global skill differences between the L4_SM and experiment soil moisture estimates with the help of the imperfect yet independent ASCAT surface soil moisture retrievals (section 3b). The broader class of IV methods also includes triple collocation, which has been used extensively for the assessment of satellite soil moisture products (Scipal et al. 2008; Dorigo et al. 2010; Gruber et al. 2016, 2020; Chen et al. 2017, 2018). When applied to soil moisture, triple collocation provides robust metrics only after the climatological mean seasonal cycle has been removed from the time series (Draper et al. 2013). Our global skill assessment is therefore applied only to anomaly data and limited to differences in anomaly R values for surface soil moisture.

IV methods go beyond simple pairwise correlation and determine skill versus unknown true values from the information provided by multiple datasets that have uncorrelated errors. Standard triple collocation requires three such datasets; typically, soil moisture skill assessments use land model estimates, (passive) radiometer retrievals, and (active) radar retrievals. This does not work for L4_SM estimates, however, which are based on the assimilation of radiometer observations into a land surface model; L4_SM errors are therefore almost certainly correlated with those of other land model and radiometer-based soil moisture datasets.

Fortunately, the single IV approach of Su et al. (2014) offers a viable method for assessing L4_SM skill because it requires only one additional independent dataset with serially white errors. While errors in satellite soil moisture retrievals are generally autocorrelated (Crow and van den Berg 2010), errors in ASCAT retrievals and a 2-day lagged ASCAT error time series were found to be nearly uncorrelated except in deserts (Dong and Crow 2017), which we excluded from the analysis through our quality control of ASCAT (section 3b). The single IV approach used here is thus based on triplets comprised of (i) L4_SM (or CTRL, or CPCU_SIM, or SMAP_DA) data, (ii) ASCAT retrievals, and (iii) a 2-day lagged time series of the ASCAT retrievals. Additional tests (not shown) indicate that the results are insensitive to the choice of lag within a range of 0.5–3 days.

There are other IV variants that require only one additional independent dataset. Dong et al. (2019b) introduce a so-called double IV approach that uses two lagged time series instead of just one, constructed from each of the two independent datasets. When the two datasets under consideration have errors with similar autocorrelation, as is the case in the precipitation application investigated by Dong et al. (2019b), the double IV approach is less prone to bias in the estimated skill metrics and has reduced sampling error compared to the single IV method of Su et al. (2014). In our soil moisture application, however, errors in the model-based estimates (CTRL, CPCU_SIM, SMAP_DA, and L4_SM) likely have much longer autocorrelation time scales than do errors in the ASCAT retrievals, which makes the single IV method more suitable. We tested this by comparing the metrics obtained with the single and double IV approaches against reference metrics obtained from triple collocation at the SMAP core sites, where in situ measurements provide a third independent dataset. Results (not shown) indicate that the single IV approach is indeed more suitable for soil moisture and therefore used here. Also note that in the present paper we do not use absolute skill metrics and focus exclusively on skill differences, which mitigates potential issues with bias in the IV-based metrics.

Another IV variant, developed by Dong et al. (2019a), determines the relative skill of two soil moisture datasets, expressed as the ratio of the anomaly correlation coefficients versus the unknown true values, with the help of just one additional dataset that has uncorrelated errors. A disadvantage of the resulting skill ratios is that they can be difficult to interpret because small differences in anomaly R values can translate into very large ratios when the baseline skill is poor. On the other hand, the method does not require that the errors in the instrumental variable (here, ASCAT) are serially white. We repeated our analysis using the Dong et al. (2019a) approach and find that the resulting skill improvements from SMAP Tb assimilation and CPCU-based precipitation corrections (not shown) are generally consistent with the results obtained with the single IV method, which suggests that (outside desert regions) the error autocorrelation of ASCAT soil moisture anomalies is sufficiently small for our IV-based skill assessment.

4. Results

a. Surface and root-zone soil moisture skill differences based on in situ measurements

This section summarizes the average soil moisture skill metrics versus in situ measurements (section 3a). For reference, complete results for the individual 9-km reference pixels are provided in Tables S2–S5. The average ubRMSE of the L4_SM product across all 18 core validation sites, shown in Fig. 3a, is 0.040 m3 m−3 for surface and 0.026 m3 m−3 for root-zone soil moisture, which confirms that the L4_SM product meets its accuracy requirement (ubRMSE ≤ 0.04 m3 m−3) for the 4-yr period investigated here. [The metrics published by Reichle et al. (2019) are for 3 years of data.] As expected, L4_SM consistently performs best, and CTRL consistently performs worst across all metrics (Fig. 3). In every case, the combined impact from the assimilation of SMAP Tb observations and the use of CPCU precipitation in L4_SM results in statistically significant skill improvements over CTRL. Between CTRL and L4_SM surface (root-zone) soil moisture, the ubRMSE is reduced by 0.007 (0.009) m3 m−3, the R value is increased by 0.16 (0.21), and the anomaly R value is increased by 0.22 (0.25).

Fig. 3.
Fig. 3.

(a) ubRMSE (m3 m−3), (b) R (dimensionless), and (c) anomaly R (dimensionless) for CTRL, SMAP_DA, CPCU_SIM, and L4_SM estimates of surface and root-zone soil moisture vs in situ measurements, averaged across 18 core validation sites (Table 1). Error bars indicate 95% confidence intervals.

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

The performance of the SMAP_DA and CPCU_SIM experiments falls between the performance of CTRL and that of L4_SM (Fig. 3). SMAP_DA and CPCU_SIM both capture about 60% of the surface and root-zone soil moisture ubRMSE reduction and about 70% of the root-zone R and anomaly R improvements achieved by L4_SM relative to CTRL. For the surface soil moisture correlation metrics, however, the assimilation of SMAP Tb observations has a significantly greater impact than the use of CPCU precipitation. The R values (anomaly R values) are improved by 0.14 (0.18) for SMAP_DA and by only 0.07 (0.08) for CPCU_SIM.

Figure 3 suggests that the assimilation of SMAP Tb observations and the use of CPCU precipitation each contribute independent information. However, the contributions of the two observational datasets to the total improvement from CTRL to L4_SM are not fully independent. For example, the root-zone soil moisture anomaly R is increased by ~0.17 for SMAP_DA and CPCU_SIM relative to CTRL. But the total improvement in the root-zone anomaly R (relative to CTRL) is only 0.25, which is less than the sum of the individual contributions in the absence of the other data source. Another way of expressing the same result is to compare the root-zone anomaly R improvement of 0.18 for SMAP_DA over CTRL with the improvement of 0.09 for L4_SM over CPCU_SIM, both of which reflect the additional information from SMAP Tb assimilation. The latter improvement is smaller because some of the skill contributed by the SMAP Tb assimilation in L4_SM is already reflected in the CPCU-informed simulation (CPCU_SIM). This result differs somewhat from that of Liu et al. (2011), who found that in their system—a precursor to the L4_SM system—the information contributed by precipitation corrections and AMSR-E soil moisture retrievals was largely independent.

An interesting aside is that the anomaly R value for surface soil moisture increased by 0.23 from CTRL to L4_SM when averaged over the four “CalVal” sites (Reynolds Creek, Walnut Gulch, Little Washita, and Little River; Table S3) used by Liu et al. (2011). Their corresponding skill increase from AMSR-E assimilation and (an older version of) CPCU-based precipitation corrections was only 0.16 (their Fig. 6), despite the fact that the baseline model skill has improved in the meantime (e.g., Reichle et al. 2019, their Fig. 2d). The larger surface soil moisture anomaly R increase seen here therefore underscores the considerably improved quality of the assimilated SMAP observations and the improvements in the (spatially distributed) analysis of the L4_SM algorithm compared to the assimilation of AMSR-E retrievals using a local (1D) EnKF in Liu et al. (2011).

b. IV-based surface soil moisture skill differences at core validation sites

In this section, we evaluate the skill improvements derived from the IV method using ASCAT data against those obtained versus in situ measurements. Recall that the IV method only affords the estimation of the anomaly R skill metric for surface soil moisture (section 3d), so for the remainder of this section we simply use the term “skill” as shorthand for surface soil moisture skill in terms of the anomaly R metric. After quality control (section 3b), IV-based skill estimates were available for 17 of the 18 core validation sites listed in Table 1 (excluding Niger).

Figure 4 shows the skill differences for five pairs of experiments averaged over the 17 core validation site locations. Skill differences were derived versus the in situ measurements (gray bars) and from the IV method using ASCAT retrievals (black bars). While there are minor discrepancies between the in situ–based and the IV-based skill differences for each pair of experiments, the key results derived from in situ–based metrics in the previous section are clearly reproduced by the IV-based skill differences (Fig. 4). The contribution of CPCU precipitation to the skill (seen in the first and third groups of bars) is clearly smaller than that of SMAP Tb assimilation (seen in the second and fourth groups of bars). Moreover, the skill contribution of the CPCU precipitation is greater in the absence of SMAP Tb assimilation (first group of bars) than in the presence of SMAP Tb assimilation (third groups of bars). Similarly, the skill contribution of SMAP Tb assimilation is greater in the absence of CPCU precipitation (second group of bars) than in their presence (fourth group of bars).

Fig. 4.
Fig. 4.

Anomaly R skill differences for surface soil moisture at the SMAP core validation site locations. Anomaly R values were computed (gray bars) vs in situ measurements from the core validation sites and (black bars) from the IV method using ASCAT surface soil moisture retrievals. Bars show anomaly R differences averaged across 17 of the 18 core sites listed in Table 1 (excluding Niger; see text for details).

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

The IV-based skill differences overestimate the in situ–based skill differences by just 0.005 on average across the five pairs of experiments (Fig. 4). Moreover, scatterplots (not shown) confirm that the skill differences at the individual core validation sites are highly consistent between the in situ–derived and IV-based estimates, with typical R2 values of ~0.69. We also repeated the analysis with the skill differences at the individual 9-km core site reference pixels. In this case, the scatter (not shown) is a bit larger, with typical R2 values of ~0.56, but the results, especially the relative magnitude of the skill differences across the five pairs of experiments, remain virtually unchanged from those shown in Fig. 4.

c. Global surface soil moisture skill differences

Most of the SMAP core validation sites used for the in situ validation in section 4a are in CONUS and western Europe (Fig. S1). In these regions, the CTRL simulation should generally perform better than elsewhere because there are relatively dense networks of conventional and aircraft observations of the atmosphere that inform the surface meteorological forcing data from GEOS (section 2a). Moreover, in these regions we should have better knowledge of land model parameters that are ultimately based on in situ measurements, such as the soil hydraulic parameters derived from soil texture data. It is therefore unclear if the skill improvements from CPCU precipitation and SMAP Tb assimilation seen at the core validation site locations (Figs. 3 and 4) are similar elsewhere. Based on the successful verification (section 4b) of the IV-based skill differences at the core validation site locations, this section expands the performance assessment to most of the global land surface. As above, for the remainder of this section we simply use the term “skill” as shorthand for surface soil moisture skill in terms of the anomaly R metric.

Figure 5a shows a global map of the skill difference between the L4_SM and CTRL surface soil moisture, representing the combined impact of CPCU precipitation and SMAP Tb assimilation. Overall, there is widespread and strong improvement in the L4_SM anomaly R values relative to CTRL, with an increase of 0.15 in the global average. Skill differences are generally not available in tropical forests and mountainous regions, owing to the lack of quality-controlled ASCAT retrievals there (section 3b). Outside of these regions, the L4_SM skill exceeds that of CTRL by ~0.2–0.3 in South America, subtropical Africa, and most of North America and Eurasia except high-latitude regions (north of 55°N). The L4_SM skill is still generally better than that of CTRL in high-latitude regions, albeit with typically smaller improvements and a few regions that show a slight skill degradation. Small pockets of degradation in L4_SM skill relative to that of CTRL are seen in the patches of tropical forests and deserts where ASCAT data were not screened by our quality control. In deserts, volume scattering is known to adversely impact the quality of the ASCAT retrievals (Matzler 1998; Wagner et al. 2013), which may in turn result in questionable skill estimates from the IV analysis. In tropical forests, ASCAT soil moisture retrievals are also of poor quality owing to the dense vegetation cover.

Fig. 5.
Fig. 5.

Surface soil moisture anomaly R skill differences for (a) L4_SM minus CTRL, (b) L4_SM minus CPCU_SIM, and (c) L4_SM minus SMAP_DA. Skill is measured in terms of anomaly R (dimensionless). Panel (a) shows the total contribution of CPCU precipitation and SMAP Tb data to L4_SM skill. Panel (b) shows the additional contribution of SMAP Tb assimilation to L4_SM skill on top of CPCU-based precipitation corrections. Panel (c) shows the additional contribution of precipitation corrections on top of SMAP Tb assimilation. Across all panels, red colors indicate that L4_SM is better and blue colors indicate that L4_SM is worse than the respective reference experiment. White shading indicates glaciated land or land for which skill metrics could not be computed (section 3b). Spatial averages (avg) are computed across all nonwhite land areas.

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

The individual contributions of SMAP Tb assimilation and CPCU precipitation to the L4_SM skill are shown in Figs. 5b and 5c, respectively. The contributions shown here are the “additional” contributions of each source of information on top of that of the other data source. That is, Fig. 5b measures the impact of the SMAP Tb observations as the skill difference between L4_SM and the model-only CPCU_SIM simulation, both of which use CPCU-based precipitation corrections. (Note that Fig. 5b measures the skill improvement in terms of an anomaly R difference across 4 years, whereas Dong et al. (2019a, their Fig. 2a) present the anomaly R skill ratio for the same two datasets across 3 years.) Similarly, Fig. 5c measures the impact of CPCU observations as the skill difference between the L4_SM and SMAP_DA estimates, both of which assimilate SMAP Tb observations.

A comparison of Figs. 5b and 5c makes it immediately clear that SMAP Tb observations have a much larger impact on the L4_SM skill than does CPCU precipitation across much of the globe. In the global average, the L4_SM anomaly R value is increased by 0.11 from SMAP Tb assimilation and by only 0.03 from CPCU-based precipitation corrections. SMAP Tb observations have a strong positive impact in central North America, much of South America, subtropical Africa, India and portions of Eurasia. The strongest positive impact of the SMAP Tb observations is seen in central Australia, where L4_SM anomaly R values exceed those of the model-only CPCU_SIM estimates by up to 0.5. SMAP Tb assimilation leads to some skill degradation in other dry regions (e.g., southern Africa). As mentioned above, though, it is possible that the low signal-to-noise ratio of ASCAT retrievals in these regions distorts the skill difference results.

By construction, there is no impact from CPCU precipitation in Africa and in the high latitudes (Fig. 5c) because the GEOS precipitation is not corrected in these regions (Fig. 1c). Outside of the high latitudes, CPCU precipitation contributes the most to L4_SM skill in North America, Europe, and East Asia, with typical increases of up to 0.1, consistent with the generally dense gauge networks in these regions. In South Asia, improvements in L4_SM skill from CPCU precipitation are particularly strong in Laos and Thailand, but with noticeable skill degradation in neighboring Myanmar and Vietnam. CPCU precipitation also generally improves L4_SM skill across South America, but some degradation is seen along the edges of the tropical forests and the Andes. The strongest skill degradation from CPCU precipitation is seen in central Australia, the Arabian Peninsula, and Tibet, suggesting that in these regions the GEOS precipitation rates are superior to those estimated based on a few sparsely distributed gauges in the CPCU dataset (Fig. 1c).

Next, Fig. 6 assesses the skill contributions of the SMAP Tb observations and CPCU precipitation in the absence of information from the other data source. That is, Fig. 6a measures the impact of SMAP Tb observations as the skill difference between the SMAP_DA assimilation estimates and the model-only CTRL estimates, both of which are forced with GEOS precipitation and do not use CPCU precipitation. Similarly, Fig. 6b measures the impact of CPCU precipitation as the skill difference between the model-only CPCU_SIM and CTRL simulations, neither of which assimilates SMAP Tb observations.

Fig. 6.
Fig. 6.

Surface soil moisture anomaly R skill differences for (a) SMAP_DA minus CTRL and (b) CPCU_SIM minus CTRL. Skill is measured in terms of anomaly R (dimensionless). Panel (a) shows the contribution of SMAP Tb assimilation to L4_SM skill in the absence of CPCU-based precipitation corrections. Panel (b) shows the contribution of precipitation corrections in the absence of SMAP Tb assimilation. Red colors indicate that SMAP_DA (or CPCU_SIM) is better and blue colors indicate that SMAP_DA (or CPCU_SIM) is worse than CTRL. White shading and spatial averages (avg) are as in Fig. 5.

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

The impact of SMAP Tb assimilation in the absence of CPCU-based precipitation corrections (Fig. 6a) is similar overall to that in the presence of CPCU-based corrections (Fig. 5b), with anomaly R differences of 0.13 and 0.11, respectively, in the global average. There are notable differences in the patterns, though, as illustrated in the double difference map of Fig. 7, which shows the difference between the skill contribution from the SMAP Tb observations in the absence of CPCU precipitation (i.e., Fig. 6a) minus that in the presence of CPCU precipitation (i.e., Fig. 5b). Most strikingly, the strong L4_SM improvement seen in central Australia from SMAP Tb assimilation in the presence of CPCU-based precipitation corrections (Fig. 5b) is reduced considerably to a mostly neutral impact when CPCU-based corrections are absent (Fig. 6a). That is, the strong positive impact in central Australia from SMAP Tb assimilation in L4_SM (Fig. 5b) is primarily facilitated by the dramatic skill degradation of ~0.3–0.5 in the model-only CPCU_SIM simulation relative to the CTRL estimates (Fig. 6b). CPCU-based precipitation corrections also reduce the model-only skill in parts of South America, the Arabian Peninsula, Myanmar, Vietnam, and Tibet (Fig. 6b), where SMAP Tb assimilation then compensates for the loss of skill by contributing more to the L4_SM skill in the presence of precipitation corrections than it would in their absence (blue-colored regions in Fig. 7). These findings about the poor quality of the CPCU precipitation in central Australia and Myanmar are consistent with the earlier, more anecdotal reports by Reichle et al. (2017b,c). Conversely, where the precipitation corrections increase the skill of the model-only CPCU_SIM simulation (relative to CTRL; Fig. 6b), including most of North America and Europe but also portions of South America and Asia, SMAP Tb assimilation contributes less to the L4_SM skill in the presence of precipitation corrections than it would in their absence (yellow- and red-colored regions in Fig. 7).

Fig. 7.
Fig. 7.

Impact of CPCU-based precipitation corrections on surface soil moisture skill improvement from SMAP Tb assimilation: skill difference from SMAP assimilation without precipitation corrections (SMAP_DA minus CTRL) minus skill difference from SMAP assimilation with precipitation corrections (L4_SM minus CPCU_SIM). Yellow and red colors indicate where SMAP Tb assimilation and CPCU precipitation contribute overlapping information to the soil moisture estimates. Blue colors indicate where SMAP Tb assimilation compensates for soil moisture skill degradation caused by CPCU-based precipitation corrections. Skill is the anomaly R for surface soil moisture. White shading and spatial average (avg) are as in Fig. 5.

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

For the most part, however, the skill contributions from the SMAP Tb observations and CPCU precipitation are additive, as indicated by the generally much lower double-differences values of Fig. 7 compared to the combined skill impact of Fig. 5a. (Note that simple arithmetic manipulation reveals that the double difference shown in Fig. 7 is also equal to the sum of the individual contributions from SMAP Tb observations (Fig. 6a) and CPCU precipitation (Fig. 6b) minus their combined contribution (Fig. 5a) relative to CTRL).

In the global average, the skill improvement is 0.13 from SMAP Tb assimilation (Fig. 6a) and 0.04 from CPCU precipitation (Fig. 6b). At the core validation site locations, the corresponding, IV-based skill improvements are 0.19 for SMAP Tb assimilation and 0.10 for CPCU precipitation (Fig. 4), suggesting that in both cases SMAP Tb observations contribute about 2–3 times as much skill as does CPCU precipitation. In absolute terms, the skill improvements are larger at the core validation site locations than in the global average, most likely because the core validation sites are preferentially situated in grasslands and croplands, where the SMAP Tb observations are most sensitive to surface soil moisture, whereas the global average includes a substantial fraction of more densely vegetated regions, where SMAP Tb observations are less sensitive to surface soil moisture. Likewise, core validation sites are preferentially located in regions with good precipitation gauge coverage, whereas the global average includes vast regions with low or no gauge coverage, including Africa and the high latitudes, where CPCU data are not used at all.

The individual skill contributions in the presence of the other dataset are similarly smaller in the global average (Fig. 5) than at the core validation site locations only (Fig. 4). This suggests that (absolute) skill improvements diagnosed with in situ measurements overestimate those obtained in the global average. However, in many regions, particularly in the transition zones between wet and dry climates where soil moisture plays a large role in land–atmosphere interactions (Koster et al. 2004, 2011), the (absolute) skill improvement values can easily exceed the numbers typically seen at the core validation site locations, as illustrated by the darker red colors in Figs. 5 and 6.

d. Impact of CPCU precipitation on assimilation diagnostics

In this section, we investigate the impact of the CPCU-based precipitation corrections on the observation-minus-forecast (O-F) Tb residuals and soil moisture increments produced by the SMAP Tb assimilation in the SMAP_DA and L4_SM experiments. Figure 8a shows the time series standard deviation of the O-F Tb residuals for the L4_SM estimates. This metric measures the typical misfit of the model Tb forecast and the subsequently assimilated SMAP observation (after rescaling). In the global average, the typical misfit is 5.8 K, with values ranging from ~3 K in deserts and forested regions to ~12 K in regions with modest vegetation cover and relatively strong short-term and interannual variations in soil moisture, including in central North America, southern South America, southern Africa, the Sahel, portions of central Asia, India, and most of Australia. This map is based on 4 years of data but otherwise identical to Fig. 8a of Reichle et al. (2019), who provide further discussion.

Fig. 8.
Fig. 8.

(a) Time series standard deviation of observations-minus-forecast (O-F) Tb residuals from the L4_SM product. (b) Difference in time series standard deviation of O-F Tb residuals between L4_SM and SMAP_DA. Blue colors in (b) indicate that typical O-F residuals are smaller (better) for L4_SM than for SMAP_DA. White shading indicates glaciated land or land for which the diagnostic could not be computed because an insufficient number of SMAP observations was assimilated. Spatial averages (avg) are computed across all nonwhite land areas.

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

The primary focus here is on the difference between the standard deviations of the O-F Tb residuals in the L4_SM and SMAP_DA assimilation estimates, which reveals the impact of CPCU precipitation on the SMAP Tb analysis in the L4_SM system. In the global average, the standard deviation (or typical magnitude) of the O-F Tb residuals is reduced by 0.1 K when CPCU precipitation is used (Fig. 8b). The smaller typical magnitude of the O-F Tb residuals in L4_SM indicates a slightly better fit of the model forecast Tb values to the subsequently assimilated SMAP Tb observations. The impact of CPCU precipitation is not uniform, though. Again, by construction there is no impact in Africa and the high latitudes. In the central United States, large portions of South America, and India, the use of CPCU precipitation reduces the typical magnitude of the O-F Tb residuals by up to 2 K. Conversely, the use of CPCU precipitation increases the typical magnitude of the O-F Tb residuals by up to 3 K in parts of the Andes, the Arabian Peninsula, Tibet, and, perhaps most prominently, in central Australia, suggesting that the use of CPCU precipitation degrades the skill of the simulated Tbs in these regions. This geographic pattern is consistent with, and thus provides confirmation of, the impact of the CPCU precipitation seen in the IV-based global skill analysis (Figs. 5c and 6b).

The time series standard deviations of the L4_SM surface and root-zone soil moisture increments are shown for reference in Figs. 9a and 9c, respectively. As for the O-F Tb residuals, the soil moisture increments graphics shown here are 4-yr versions of those published by Reichle et al. (2019, their Figs. 7c,e), who provide further discussion. Because the soil moisture increments are derived from the O-F Tb residuals, it comes as no surprise that the spatial patterns of the typical magnitude of the increments (Figs. 9a,c) largely match that of the O-F Tb residuals (Fig. 8a), with relatively larger typical increments in regions featuring modest vegetation cover and relatively strong short-term and interannual variations in soil moisture. Moreover, the differences between the L4_SM and SMAP_DA systems in the typical magnitude of the surface soil moisture increments (Fig. 9b) exhibit a pattern that is similar to that of the corresponding differences in the typical magnitude of the O-F Tb residuals (Fig. 8b). That is, the use of CPCU precipitation generally results in typical L4_SM surface soil moisture increments that are slightly smaller than corresponding SMAP_DA increments in the central United States, portions of South America, and India. Typical L4_SM surface soil moisture increments are larger than those of SMAP_DA in parts of the Andes, the Arabian Peninsula, Tibet, and central Australia. The larger typical surface soil moisture increments in these regions again confirm that the L4_SM model forecasts are worse than those of the SMAP_DA experiment, that is, using CPCU precipitation degrades the model skill in these regions, and SMAP Tb assimilation compensates for this skill degradation by applying larger soil moisture increments when CPCU precipitation is used.

Fig. 9.
Fig. 9.

(a) Time series standard deviation of surface soil moisture (SFSM) increments from the L4_SM product. (b) Difference in time series standard deviation of SFSM increments between L4_SM and SMAP_DA. (c),(d) As in (a) and (b), respectively, but for root-zone soil moisture (RZSM) increments. Blue colors in (b) and (d) indicate that typical soil moisture increments are smaller (better) for L4_SM than for SMAP_DA. White shading and the spatial average (avg) are as in Fig. 8.

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

The same is not true for root-zone soil moisture increments in central Australia (Fig. 9d). Here, the typical magnitude of root-zone soil moisture increments is smaller with the use of CPCU precipitation (i.e., L4_SM) than it is without (i.e., SMAP_DA). One possible explanation for this somewhat counterintuitive and surprising result is that the very poor quality of the CPCU precipitation in this region adversely impacts the quality of the modeled error correlations between the surface and root-zone soil moisture variables in the ensemble assimilation algorithm. A more in-depth investigation of this result is left for future work.

e. Runoff skill differences in CONUS

In this section, we assess the impact of SMAP Tb assimilation and CPCU precipitation on the skill of the L4_SM runoff estimates. On average across the 237 basins in CONUS and the 2015–18 warm seasons (section 3c), the L4_SM runoff underestimates the observed streamflow by −0.085 mm day−1 (31 mm yr−1), with a mean absolute bias of 0.290 mm day−1 (106 mm yr−1) (Fig. S2). The underestimation is largest in the western and central CONUS, with values up to −0.8 mm day−1. In contrast, streamflow is overestimated by up to 0.8 mm day−1 in Florida and along the East coast. The average runoff R skill is 0.56, with generally larger skill values (>0.5) in eastern CONUS (except in Florida and along the Carolina coast) and lower values (<0.5) in much of western CONUS. We refer the reader to Reichle et al. (2019, their section 4.3) for a more detailed discussion of the L4_SM runoff skill. Here, we focus instead on the skill differences seen between L4_SM and the CTRL, CPCU_SIM, and SMAP_DA experiments.

On average across the 237 basins, the mean absolute runoff bias in L4_SM is smaller by 0.377 mm day−1 (138 mm yr−1) than that of CTRL, with most of the reductions concentrated in the central CONUS, Florida and along the Carolina coast (Fig. 10a), where CTRL considerably overestimated the USGS streamflow measurements (Fig. S2). Similar improvements are seen in the L4_SM R values, which are, on average, larger by 0.18 and with improvements concentrated in the central CONUS (Fig. 10b), where the R skill of CTRL was particularly poor (Fig. S2). As done in Fig. 5 for soil moisture, the individual contribution of SMAP Tb assimilation to the runoff skill is determined by subtracting the CPCU_SIM skill from that of L4_SM. By design, SMAP Tb assimilation barely changes the runoff bias (Fig. 10c; note the different color bar). There is a small average improvement of 0.02 in the L4_SM R over that of CPCU_SIM, with improvements of ~0.05 in central CONUS and Florida (Fig. 10d; note the different color scale). These results are consistent with earlier findings that soil moisture assimilation yields only slight improvements in runoff skill (Lievens et al. 2016; Reichle et al. 2019). Here, we find that much more of the skill in the L4_SM runoff estimates is contributed by the CPCU precipitation; the skill difference maps between L4_SM and SMAP_DA for the absolute bias (Fig. 10e) and the R values (Fig. 10f) are very similar to the corresponding skill differences between L4_SM and CTRL (Figs. 10a,b). Because in all experiments the precipitation from GEOS and CPCU (if used) is rescaled to the (seasonally varying) GPCPv2.2 climatology (section 2a), the considerable improvement in the mean absolute bias and correlation of runoff estimates from the precipitation corrections reflects the improved timing, frequency, and intensity of the gauge-based CPCU precipitation compared to that of GEOS.

Fig. 10.
Fig. 10.

Streamflow skill differences for (left) absolute value of bias and (right) R values for (a),(b) L4_SM minus CTRL; (c),(d) L4_SM minus CPCU_SIM; and (e),(f) L4_SM minus SMAP_DA. Panels (a) and (b) show the total contribution of CPCU precipitation and SMAP Tb data to L4_SM skill. Panels (c) and (d) show the additional contribution of SMAP Tb assimilation to L4_SM skill on top of CPCU-based precipitation corrections. Panels (e) and (f) show the additional contribution of precipitation corrections on top of SMAP Tb assimilation. Across all panels, blue colors indicate that L4_SM is better and red colors indicate that L4_SM is worse than the respective reference experiment. Note the different color axis limits in (c) and (d).

Citation: Journal of Hydrometeorology 22, 2; 10.1175/JHM-D-20-0217.1

5. Summary and conclusion

In this paper, we examined the skill of the global SMAP L4_SM data product, which is generated through the assimilation of SMAP L-band Tb observations into the Catchment land surface model driven with precipitation forcing that is based on the CPCU gauge product (outside of Africa and the high latitudes). Our analysis compared the skill of the L4_SM soil moisture and runoff data to corresponding estimates from three additional simulations that use only SMAP Tb assimilation (SMAP_DA), only CPCU precipitation (CPCU_SIM), and neither (CTRL; Fig. 2).

Previously, Liu et al. (2011) similarly examined a precursor of the L4_SM system over CONUS and found that gauge-based precipitation corrections and the assimilation of AMSR-E satellite soil moisture retrievals contribute similar and largely independent amounts of information to the anomaly R skill of soil moisture estimates. Here, based on validation against in situ measurements from 18 SMAP core validation sites, we find that this result continues to hold for the L4_SM system in terms of the ubRMSE skill of surface and root-zone soil moisture estimates as well as the correlation and anomaly correlation skill of root-zone soil moisture (Fig. 3). SMAP Tb assimilation, however, contributes around 3 times as much to the anomaly R skill of surface soil moisture than do gauge-based precipitation corrections (Figs. 3c and 4). At the global scale, the contribution of SMAP Tb assimilation to the surface soil moisture anomaly R skill is also about 3 times as large as that of CPCU precipitation (Figs. 5 and 6), based on the use of independent ASCAT soil moisture retrievals in the single IV method of Su et al. (2014). The contribution to the surface soil moisture anomaly R skill from SMAP Tb assimilation is particularly large in parts of Africa (Fig. 5b), where CPCU data are not used owing the sparsity of the gauge network (Fig. 1c), and in central Australia (Fig. 5b), where the use of CPCU data that are based on very poor gauge coverage (Fig. 1c) adversely impacts the skill of the model-based soil moisture estimates (Fig. 6b). SMAP Tb assimilation also compensates for the poor CPCU precipitation in Myanmar and Vietnam. In most other regions of the globe, the contributions to the surface soil moisture anomaly R skill from SMAP Tb assimilation and CPCU-based precipitation corrections are largely independent, as can be seen in the muted colors of the double-difference map of Fig. 7.

There is also a positive impact from the CPCU precipitation on the analysis of SMAP Tb observations. In the global average, the typical magnitude of the O-F Tb residuals is smaller by 0.1 K when CPCU precipitation data are used in the modeling system versus when they are not. In large parts of North America, southern South America, and India, typical O-F Tb values are reduced by ~1–2 K (Fig. 8b), indicating that the modeling system can better forecast SMAP Tb observations prior to their assimilation. In central Australia, typical O-F Tb values increase by ~1–3 K (Fig. 8b), which reflects the added work done by the SMAP Tb analysis in order to compensate for the degradation of the modeled soil moisture owing to the poor quality of the CPCU precipitation there. A similar difference pattern is seen in the typical magnitude of the surface soil moisture increments (Fig. 8b). Surprisingly, however, typical root-zone soil moisture increments are smaller in central Australia when CPCU-based precipitation corrections are used (Fig. 9d), despite the obvious deficiencies in the quality of the CPCU precipitation there.

It is important to keep in mind that the skill contribution from SMAP Tb assimilation dominates that from the use of CPCU precipitation only for the correlation metrics and only for surface soil moisture. The in situ–based skill analysis suggests that for root-zone soil moisture ubRMSE and correlation as well as for the surface soil moisture ubRMSE, SMAP Tb assimilation and CPCU precipitation contribute similar amounts of information to the L4_SM skill. Moreover, the R skill for L4_SM runoff estimates relies almost exclusively on the contribution from the CPCU precipitation (Fig. 10). Although there are obvious downsides to the use of CPCU precipitation, particularly the poor accuracy of CPCU precipitation in central Australia, on balance the gauge-based precipitation corrections have a net positive contribution to the soil moisture and runoff skill of the L4_SM product.

It is straightforward to not use the gauge-based precipitation corrections for entire continents such as Africa, as implemented in the L4_SM system. The same approach could be applied to Australia, which would almost certainly improve the L4_SM product skill in central Australia; however, at the same time, the L4_SM product would lose the valuable information provided by the dense network of precipitation gauges in southeastern Australia. It would, in fact, be difficult to construct a precipitation correction algorithm that applies weights with the short length scales imposed by the spatial density of gauges or, equivalently, by precipitation skill metrics derived from triple collocation (Dong et al. 2020a,b). Such an algorithm would unavoidably impose length scales on the very complex, spatiotemporal dynamics of individual precipitation events and is beyond the scope of this paper and the broader SMAP Level-4 algorithm development.

Precipitation products that are optimal for the generation of operational, global soil moisture products are still lacking. Including precipitation gauge observations in operations naturally depends on the availability of gauge measurements within the desired latency, which imposes a trade-off between quality and latency that varies across the globe. For instance, the “Early” and “Late” run products of the Integrated Multisatellite Retrievals for the Global Precipitation Measurement (IMERG) mission (Tan et al. 2019) offer short-latency precipitation estimates that do not include gauge measurements. The higher-quality “Final” IMERG estimates include gauge measurements but have a much longer latency of 3–4 months. Important progress is being made toward precipitation products that can support operational soil moisture simulations through, e.g., the development of the Multi-Source Weighted-Ensemble Precipitation product (Beck et al. 2019, 2020). There is, however, still an urgent need for improved operational, global, gauge-based precipitation products with short latency (~2–3 days) that can support the operational generation of high-quality soil moisture products for applications users. Suitable precipitation products must also provide a corresponding multidecadal historic record that can support the calibration of the soil moisture algorithm.

Acknowledgments

Funding for this work was provided by the NASA SMAP mission and the SMAP Science Team. Computational resources were provided by the NASA High-End Computing program through the NASA Center for Climate Simulation. We are grateful for those who make the generation and dissemination of the SMAP data products possible, including staff at JPL, GSFC, NSIDC, USGS, NOAA CPC, NOAA NCEI, and USDA-NRCS. For the validation data we thank A. Berg, S. Bircher, D. Bosch, T. G. Caldwell, A. Colliander, M. Cosh, Á. González-Zamora, C. D. Holifield Collins, T. Jackson, K. H. Jensen, S. Livingston, E. Lopez-Baeza, J. Martínez-Fernández, H. McNairn, M. Moghaddam, A. Pacheco, T. Pellarin, J. Prueger, T. Rowlandson, M. Seyfried, P. Starks, Z. Su, E. Tetlock, M. Thibeault, R. van der Velde, J. P. Walker, X. Wu, Y. Zeng, and staff at USGS, USDA-ARS, and the University of Valencia. Funding for the Kenaston network was provided by the Canadian Space Agency and by Environment and Climate Change Canada. USDA is an equal opportunity provider and employer. We thank three anonymous reviewers for their helpful comments.

Data availability statement

The SMAP L4_SM product, ancillary inputs, and the in situ soil moisture measurements used for validation are available from http://nsidc.org/data/smap. Output from the CTRL, SMAP_DA, and CPCU_SIM simulations can be obtained from https://gmao.gsfc.nasa.gov/gmaoftp/reichle/datasets/jhm-d-20-0217. GEOS forcing data are available from https://fluid.nccs.nasa.gov/weather. CPCU precipitation is available from ftp://ftp.cpc.ncep.noaa.gov/precip/CPC_UNI_PRCP/GAUGE_CONUS. ASCAT soil moisture data are available from the EUMETSAT Hydrology Satellite Application Facility at http://hsaf.meteoam.it/soil-moisture.php. Streamflow data are available from http://nwis.waterdata.usgs.gov/nwis. The GEOS source code is available under the NASA Open-Source Agreement at http://opensource.gsfc.nasa.gov/projects/GEOS-5.

REFERENCES

  • Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 11471167, https://doi.org/10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Babaeian, E., M. Sadeghi, S. B. Jones, C. Montzka, H. Vereecken, and M. Tuller, 2019: Ground, proximal, and satellite remote sensing of soil moisture. Rev. Geophys., 57, 530616, https://doi.org/10.1029/2018RG000618.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Balsamo, G., and Coauthors, 2018: Satellite and in situ observations for Advancing Global Earth Surface Modelling: A review. Remote Sens., 10, 2038, https://doi.org/10.3390/rs10122038.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bauer-Marschallinger, B., and Coauthors, 2019: Toward global soil moisture monitoring with Sentinel-1: Harnessing assets and overcoming obstacles. IEEE Trans. Geosci. Remote Sens., 57, 520539, https://doi.org/10.1109/TGRS.2018.2858004.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beck, H. E., E. F. Wood, M. Pan, C. K. Fisher, D. G. Miralles, A. I. J. M. van Dijk, T. R. McVicar, and R. F. Adler, 2019: MSWEP V2 global 3-hourly 0.1° precipitation: Methodology and quantitative assessment. Bull. Amer. Meteor. Soc., 100, 473500, https://doi.org/10.1175/BAMS-D-17-0138.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beck, H. E., and Coauthors, 2020: Evaluation of 18 satellite- and model-based soil moisture products using in situ measurements from 826 sensors. Hydrol. Earth Syst. Sci., https://doi.org/10.5194/HESS-2020-184

    • Search Google Scholar
    • Export Citation
  • Bircher, S., N. Skou, K. H. Jensen, J. P. Walker, and L. Rasmussen, 2012: A soil moisture and temperature network for SMOS validation in western Denmark. Hydrol. Earth Syst. Sci., 16, 14451463, https://doi.org/10.5194/hess-16-1445-2012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bosch, D. D., J. M. Sheridan, and L. K. Marshall, 2007: Precipitation, soil moisture, and climate database, Little River Experimental Watershed, Georgia, United States. Water Resour. Res., 43, W09472, https://doi.org/10.1029/2006WR005834.

    • Search Google Scholar
    • Export Citation
  • Brodzik, M. J., B. Billingsley, T. Haran, B. Raup, and M. H. Savoie, 2012: EASE-Grid 2.0: Incremental but significant improvements for Earth-gridded data sets. ISPRS Int. J. Geoinf., 1, 3245, https://doi.org/10.3390/ijgi1010032.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Caldwell, T. G., T. Bongiovanni, M. H. Cosh, C. Halley, and M. H. Young, 2018: Field and laboratory evaluation of the CS655 soil water content sensor. Vadose Zone J., 17, 170214, https://doi.org/10.2136/vzj2017.12.0214.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chan, S. K., and Coauthors, 2016: Assessment of the SMAP passive soil moisture product. IEEE Trans. Geosci. Remote Sens., 54, 49945007, https://doi.org/10.1109/TGRS.2016.2561938.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chan, S. K., E. G. Njoku, and A. Colliander, 2018: SMAP L1C radiometer half-orbit 36 km EASE-grid brightness temperatures, version 4 NSIDC DAAC, accessed 4 March 2020, https://doi.org/10.5067/ZVILG0PS6CTI.

    • Crossref
    • Export Citation
  • Chen, F., and Coauthors, 2017: Application of triple collocation in ground-based validation of Soil Moisture Active/Passive (SMAP) level 2 data products. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 10, 489502, https://doi.org/10.1109/JSTARS.2016.2569998.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, F., W. T. Crow, R. Bindlish, A. Colliander, M. S. Burgin, J. Asanuma, and K. Aida, 2018: Global-scale evaluation of SMAP, SMOS and ASCAT soil moisture products using triple collocation. Remote Sens. Environ., 214, 113, https://doi.org/10.1016/j.rse.2018.05.008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, M., W. Shi, P. Xie, V. B. S. Silva, V. E. Kousky, R. Wayne Higgins, and J. E. Janowiak, 2008: Assessing objective techniques for gauge-based analyses of global daily precipitation. J. Geophys. Res., 113, D04110, https://doi.org/10.1029/2007JD009132.

    • Search Google Scholar
    • Export Citation
  • Clewley, D., and Coauthors, 2017: A method for upscaling in situ soil moisture measurements to satellite footprint scale using random forests. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 10, 26632673, https://doi.org/10.1109/JSTARS.2017.2690220.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Colliander, A., and Coauthors, 2017a: SMAP/in situ core validation site land surface parameters match-up data, version 1. NSIDC DAAC, accessed 25 June 2020, https://doi.org/10.5067/DXAVIXLY18KM.

    • Crossref
    • Export Citation
  • Colliander, A., and Coauthors, 2017b: Validation of SMAP surface soil moisture products with core validation sites. Remote Sens. Environ., 191, 215231, https://doi.org/10.1016/j.rse.2017.01.021.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Colliander, A., Z. Yang, R. Mueller, A. Sandborn, R. Reichle, W. Crow, D. Entekhabi, and S. Yueh, 2019: Consistency between NASS surveyed soil moisture conditions and SMAP soil moisture observations. Water Resour. Res., 55, 76827693, https://doi.org/10.1029/2018WR024475.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coopersmith, E. J., M. H. Cosh, W. A. Petersen, J. Prueger, and J. J. Niemeier, 2015: Soil moisture model calibration and validation: An ARS watershed on the South Fork Iowa River. J. Hydrometeor., 16, 10871101, https://doi.org/10.1175/JHM-D-14-0145.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cosh, M. H., T. J. Jackson, P. Starks, and G. Heathman, 2006: Temporal stability of surface soil moisture in the Little Washita River watershed and its applications in satellite soil moisture product validation. J. Hydrol., 323, 168177, https://doi.org/10.1016/j.jhydrol.2005.08.020.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cosh, M. H., P. J. Starks, J. A. Guzman, and D. N. Moriasi, 2014: Upper Washita river experimental watersheds: Multiyear stability of soil water content profiles. J. Environ. Qual., 43, 13281333, https://doi.org/10.2134/jeq2013.08.0318.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., and M. J. van den Berg, 2010: An improved approach for estimating observation and model error parameters in soil moisture data assimilation. Water Resour. Res., 46, W12519, https://doi.org/10.1029/2010WR009402.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., and Coauthors, 2012: Upscaling sparse ground-based soil moisture observations for the validation of coarse-resolution satellite soil moisture products. Rev. Geophys., 50, RG2002, https://doi.org/10.1029/2011RG000372.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., F. Chen, R. H. Reichle, and Q. Liu, 2017: L band microwave remote sensing and land data assimilation improve the representation of prestorm soil moisture conditions for hydrologic forecasting. Geophys. Res. Lett., 44, 54955503, https://doi.org/10.1002/2017GL073642.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., F. Chen, R. H. Reichle, Y. Xia, and Q. Liu, 2018: Exploiting soil moisture, precipitation, and streamflow observations to evaluate soil moisture/runoff coupling in land surface models. Geophys. Res. Lett., 45, 48694878, https://doi.org/10.1029/2018GL077193.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crow, W. T., F. Chen, R. H. Reichle, and Y. Xia, 2019: Diagnosing bias in modeled soil moisture/runoff coefficient correlation using the SMAP level 4 soil moisture product. Water Resour. Res., 55, 70107026, https://doi.org/10.1029/2019WR025245.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Das, N. N., and Coauthors, 2019: The SMAP and Copernicus Sentinel 1A/B microwave active-passive high resolution surface soil moisture product. Remote Sens. Environ., 233, 111380, https://doi.org/10.1016/j.rse.2019.111380.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • De Lannoy, G. J. M., R. H. Reichle, and V. R. N. Pauwels, 2013: Global calibration of the GEOS-5 L-band microwave radiative transfer model over nonfrozen land using SMOS observations. J. Hydrometeor., 14, 765785, https://doi.org/10.1175/JHM-D-12-092.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • De Lannoy, G. J. M., R. H. Reichle, and J. A. Vrugt, 2014: Uncertainty quantification of GEOS-5 L-band radiative transfer model parameters using Bayesian inference and SMOS observations. Remote Sens. Environ., 148, 146157, https://doi.org/10.1016/j.rse.2014.03.030.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., and W. T. Crow, 2017: An improved triple collocation analysis algorithm for decomposing autocorrelated and white soil moisture retrieval errors. J. Geophys. Res. Atmos., 122, 13 08113 094, https://doi.org/10.1002/2017JD027387.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., W. T. Crow, R. Reichle, Q. Liu, F. Lei, and M. H. Cosh, 2019a: A global assessment of added value in the SMAP level 4 soil moisture product relative to its baseline land surface model. Geophys. Res. Lett., 46, 66046613, https://doi.org/10.1029/2019GL083398.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., W. T. Crow, Z. Duan, L. Wei, and Y. Lu, 2019b: A double instrumental variable method for geophysical product error estimation. Remote Sens. Environ., 225, 217228, https://doi.org/10.1016/j.rse.2019.03.003.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., F. Lei, and L. Wei, 2020a: Triple collocation based multi-source precipitation merging. Front.Water, 2, 1, https://doi.org/10.3389/frwa.2020.00001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, J., W. T. Crow, and R. Reichle, 2020b: Improving rain/no-rain detection skill by merging precipitation estimates from different sources. J. Hydrometeor., 21, 24192429, https://doi.org/10.1175/JHM-D-20-0097.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dorigo, W. A., K. Scipal, R. M. Parinussa, Y. Y. Liu, W. Wagner, R. A. M. de Jeu, and V. Naeimi, 2010: Error characterisation of global active and passive microwave soil moisture datasets. Hydrol. Earth Syst. Sci., 14, 26052616, https://doi.org/10.5194/hess-14-2605-2010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Draper, C., R. Reichle, R. de Jeu, V. Naeimi, R. Parinussa, and W. Wagner, 2013: Estimating root mean square errors in remotely sensed soil moisture over continental scale domains. Remote Sens. Environ., 137, 288298, https://doi.org/10.1016/j.rse.2013.06.013.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ducharne, A., R. D. Koster, M. J. Suarez, M. Stieglitz, and P. Kumar, 2000: A catchment-based approach to modeling land surface processes in a general circulation model: 2. Parameter estimation and model demonstration. J. Geophys. Res., 105, 24 82324 838, https://doi.org/10.1029/2000JD900328.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Entekhabi, D., and Coauthors, 2010: The Soil Moisture Active Passive (SMAP) mission. Proc. IEEE, 98, 704716, https://doi.org/10.1109/JPROC.2010.2043918.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • EUMETSAT, 2019: Product User Manual (PUM) Metop ASCAT Soil Moisture Climate Data Record v5 12.5 km (H115) and extension (H116). EUMETSAT Satellite Application Facility on Support to Operational Hydrology and Water Management, Doc. SAF/HSAF/CDOP3/PUM/, Revision: 0.1, 32 pp., https://hsaf.meteoam.it/documents/H115_ASCAT_SSM_CDR_v5_PUM_v0.1.pdf.

  • Fang, K., M. Pan, and C. Shen, 2019: The value of SMAP for long-term soil moisture estimation with the help of deep learning. IEEE Trans. Geosci. Remote Sens., 57, 22212233, https://doi.org/10.1109/TGRS.2018.2872131.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fernandez-Moran, R., and Coauthors, 2017: SMOS-IC: An alternative SMOS soil moisture and vegetation optical depth product. Remote Sens., 9, 457, https://doi.org/10.3390/rs9050457.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ford, T. W., and S. M. Quiring, 2019: Comparison of contemporary in situ, model, and satellite remote sensing soil moisture with a focus on drought monitoring. Water Resour. Res., 55, 15651582, https://doi.org/10.1029/2018WR024039.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Galle, S., and Coauthors, 2018: AMMA-CATCH, a critical zone observatory in West Africa monitoring a region in transition. Vadose Zone J., 17, 180062, https://doi.org/10.2136/vzj2018.03.0062.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • González-Zamora, Á., N. Sánchez, J. Martínez-Fernández, Á. Gumuzzio, M. Piles, and E. Olmedo, 2015: Long-term SMOS soil moisture products: A comprehensive evaluation across scales and methods in the Duero Basin (Spain). Phys. Chem. Earth Parts ABC, 83–84, 123136, https://doi.org/10.1016/j.pce.2015.05.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gruber, A., C.-H. Su, S. Zwieback, W. Crow, W. Dorigo, and W. Wagner, 2016: Recent advances in (soil moisture) triple collocation analysis. Int. J. Appl. Earth Obs. Geoinf., 45, 200211, https://doi.org/10.1016/j.jag.2015.09.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gruber, A., and Coauthors, 2020: Validation practices for satellite soil moisture retrievals: What are (the) errors? Remote Sens. Environ., 244, 111806, https://doi.org/10.1016/j.rse.2020.111806.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Heathman, G. C., M. H. Cosh, V. Merwade, and E. Han, 2012: Multi-scale temporal stability analysis of surface and subsurface soil moisture within the Upper Cedar Creek Watershed, Indiana. Catena, 95, 91103, https://doi.org/10.1016/j.catena.2012.03.008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., R. F. Adler, D. T. Bolvin, and G. Gu, 2009: Improving the global precipitation record: GPCP version 2.1. Geophys. Res. Lett., 36, L17808, https://doi.org/10.1029/2009GL040000.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jensen, K. H., and J. C. Refsgaard, 2018: HOBE: The Danish hydrological observatory. Vadose Zone J., 17, 180059, https://doi.org/10.2136/vzj2018.03.0059.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Juglea, S., and Coauthors, 2010: Modelling soil moisture at SMOS scale by use of a SVAT model over the Valencia Anchor Station. Hydrol. Earth Syst. Sci., 14, 831846, https://doi.org/10.5194/hess-14-831-2010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Keefer, T. O., M. S. Moran, and G. B. Paige, 2008: Long-term meteorological and soil hydrology database, Walnut Gulch experimental watershed, Arizona, United States. Water Resour. Res., 44, W05S07, https://doi.org/10.1029/2006WR005702.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kerr, Y. H., and Coauthors, 2010: The SMOS mission: New tool for monitoring key elements of the global water cycle. Proc. IEEE, 98, 666687, https://doi.org/10.1109/JPROC.2010.2043032.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kerr, Y. H., and Coauthors, 2016: Overview of SMOS performance in terms of global soil moisture monitoring after six years in operation. Remote Sens. Environ., 180, 4063, https://doi.org/10.1016/j.rse.2016.02.042.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khodayar, S., A. Coll, and E. Lopez-Baeza, 2019: An improved perspective in the spatial representation of soil moisture: Potential added value of SMOS disaggregated 1 km resolution “all weather” product. Hydrol. Earth Syst. Sci., 23, 255275, https://doi.org/10.5194/hess-23-255-2019.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kimball, J. S., L. Jones, K. Jensco, M. He, M. Maneta, and R. Reichle, 2019: SMAP L4 assessment of the US northern plains 2017 flash drought. 2019 IEEE Int. Geoscience and Remote Sensing Symp., Yokohama, Japan, IEEE, 5366–5369, https://doi.org/10.1109/igarss.2019.8898354.

    • Crossref
    • Export Citation
  • Koster, R. D., M. J. Suarez, A. Ducharne, M. Stieglitz, and P. Kumar, 2000: A catchment-based approach to modeling land surface processes in a general circulation model: 1. Model structure. J. Geophys. Res., 105, 24 80924 822, https://doi.org/10.1029/2000JD900327.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 11381140, https://doi.org/10.1126/science.1100217.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2011: The second phase of the global land–atmosphere coupling experiment: Soil moisture contributions to subseasonal forecast skill. J. Hydrometeor., 12, 805822, https://doi.org/10.1175/2011JHM1365.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, X., J. Xiao, J. S. Kimball, R. H. Reichle, R. L. Scott, M. E. Litvak, G. Bohrer, and C. Frankenberg, 2020: Synergistic use of SMAP and OCO-2 data in assessing the responses of ecosystem productivity to the 2018 U.S. drought. Remote Sens. Environ., 251, 112062, https://doi.org/10.1016/j.rse.2020.112062.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lievens, H., and Coauthors, 2016: Assimilation of SMOS soil moisture and brightness temperature products into a land surface model. Remote Sens. Environ., 180, 292304, https://doi.org/10.1016/j.rse.2015.10.033.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, Q., and Coauthors, 2011: The contributions of precipitation and soil moisture observations to the skill of soil moisture estimates in a land data assimilation system. J. Hydrometeor., 12, 750765, https://doi.org/10.1175/JHM-D-10-05000.1.

    • Crossref