• Betts, A. K., , Ball J. H. , , Beljaars A. C. M. , , Miller M. J. , , and Viterbo P. , 1994: Coupling between land-surface, boundary-layer parameterizations and rainfall on local and regional scales: Lessons from the wet summer of 1993. Preprints, Fifth Symp. on Global Change Studies, Nashville, TN, Amer. Meteor. Soc., 174–181.

    • Search Google Scholar
    • Export Citation
  • Bosilovich, M. G., , and Sun W. Y. , 1999: Numerical simulation of the 1993 Midwestern flood: Land–atmosphere interactions. J. Climate, 12, 14901505.

    • Search Google Scholar
    • Export Citation
  • Collins, W. D., and Coauthors, 2004: Description of the NCAR Community Atmosphere Model (CAM 3.0). NCAR Tech. Rep. NCAR/TN-464+STR, 226 pp.

  • Cook, B. I., , Bonan G. B. , , and Levis S. , 2006: Soil moisture feedbacks to precipitation in southern Africa. J. Climate, 19, 41984206.

  • Dirmeyer, P. A., 2000: Using a global soil wetness dataset to improve seasonal climate simulation. J. Climate, 13, 29002922.

  • Dirmeyer, P. A., 2011: The terrestrial segment of soil moisture–climate coupling. Geophys. Res. Lett., 38, L16702, doi:10.1029/2011GL048268.

    • Search Google Scholar
    • Export Citation
  • Dirmeyer, P. A., , Schlosser C. A. , , and Brubaker K. L. , 2009: Precipitation, recycling, and land memory: An integrated analysis. J. Hydrometeor., 10, 278288.

    • Search Google Scholar
    • Export Citation
  • D’Odorico, P., , and Porporato A. , 2004: Preferential states in soil moisture and climate dynamics. Proc. Natl. Acad. Sci. USA, 101, 88488851.

    • Search Google Scholar
    • Export Citation
  • Dominguez, F., , Kumar P. , , Liang X. Z. , , and Ting M. , 2006: Impact of atmospheric moisture storage on precipitation recycling. J. Climate, 19, 15131530.

    • Search Google Scholar
    • Export Citation
  • Eltahir, E. A. B., 1998: A soil moisture–rainfall feedback mechanism: 1. Theory and observations. Water Resour. Res., 34, 765776.

  • Eltahir, E. A. B., , and Bras R. L. , 1996: Precipitation recycling. Rev. Geophys., 34, 367378.

  • Findell, K. L., , and Eltahir E. A. B. , 1997: An analysis of the soil moisture-rainfall feedback, based on direct observations from Illinois. Water Resour. Res., 33, 725735.

    • Search Google Scholar
    • Export Citation
  • Findell, K. L., , and Eltahir E. A. B. , 2003a: Atmospheric controls on soil moisture–boundary layer interactions. Part I: Framework development. J. Hydrometeor., 4, 552569.

    • Search Google Scholar
    • Export Citation
  • Findell, K. L., , and Eltahir E. A. B. , 2003b: Atmospheric controls on soil moisture–boundary layer interactions. Part II: Feedbacks within the continental United States. J. Hydrometeor., 4, 570572.

    • Search Google Scholar
    • Export Citation
  • Findell, K. L., , Gentine P. , , Linter B. R. , , and Kerr C. , 2011: Probability of afternoon precipitation in eastern United States and Mexico enhanced by high evaporation. Nat. Geosci., 4, 434439, doi:10.1038/NGEO1174.

    • Search Google Scholar
    • Export Citation
  • Frankignoul, C., , and Hasselmann K. , 1977: Stochastic climate models, part II. Application to sea-surface temperature anomalies and thermocline variability. Tellus, 29, 289305.

    • Search Google Scholar
    • Export Citation
  • Guo, Z., and Coauthors, 2006: GLACE: The Global Land–Atmosphere Coupling Experiment. Part II: Analysis. J. Hydrometeor., 7, 611625.

    • Search Google Scholar
    • Export Citation
  • Higgins, R. W., , Shi W. , , Yarosh E. , , and Joyce R. , 2000: Improved United States precipitation quality control system and analysis. NCEP/Climate Prediction Center ATLAS 7, Camp Springs, MD, 40 pp.

  • Kim, Y. J., , and Wang G. L. , 2007: Impact of initial soil moisture anomalies on subsequent precipitation over North America in the coupled land–atmosphere model CAM3–CLM3. J. Hydrometeor., 8, 513533.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., , Suarez M. J. , , and Heiser M. , 2000: Variance and predictability of precipitation at seasonal-to-interannual time scales. J. Hydrometeor., 1, 2646.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 11381140.

  • Koster, R. D., and Coauthors, 2006: GLACE: The Global Land–Atmosphere Coupling Experiment. Part I: Overview. J. Hydrometeor., 7, 590610.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., , Schubert S. D. , , and Suarez M. J. , 2009: Analyzing the concurrence of meteorological droughts and warm periods, with implications for the determination of evaporative regime. J. Climate, 22, 33313341.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2010: The contribution of land surface initialization to subseasonal forecast skill: First results from a multi-model experiment. Geophys. Res. Lett., 37, L02402, doi:10.1029/2009GL041677.

    • 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.

    • Search Google Scholar
    • Export Citation
  • Liang, X., , Lettenmaier D. P. , , and Wood E. F. , 1996a: One-dimensional statistical dynamic representation of subgrid spatial variability of precipitation in the two-layer variable infiltration capacity model. J. Geophys. Res., 101 (D16), 21 40321 422.

    • Search Google Scholar
    • Export Citation
  • Liang, X., , Wood E. F. , , and Lettenmaier D. P. , 1996b: Surface soil moisture parameterization of the VIC-2L model: Evaluation and modifications. Global Planet. Change, 13, 195206.

    • Search Google Scholar
    • Export Citation
  • Loveland, T. R., , Reed B. C. , , Brown J. F. , , Ohlen D. O. , , Zhu J. , , Yang L. , , and Merchant J. W. , 2001: Development of a global land cover characteristics database and IGBP DISCover from 1-km AVHRR data. Int. J. Remote Sens., 21, 13031330.

    • Search Google Scholar
    • Export Citation
  • Maurer, E. P., , Wood A. W. , , Adam J. C. , , Lettenmaier D. P. , , and Nijssen B. , 2002: A long-term hydrologically based dataset of land surface fluxes and states for the conterminous United States. J. Climate, 15, 32373251.

    • Search Google Scholar
    • Export Citation
  • Mei, R., , and Wang G. L. , 2011: Impact of sea surface temperature and soil moisture on summer precipitation in the United States based on observational data. J. Hydrometeor., 12, 10861099.

    • Search Google Scholar
    • Export Citation
  • Meng, L., , and Quiring S. M. , 2008: A comparison of soil moisture models using Soil Climate Analysis Network observations. J. Hydrometeor., 9, 641659.

    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360.

  • Neale, R. B., and Coauthors, 2010: Description of the NCAR Community Atmosphere Model (CAM 4.0). NCAR Tech. Rep. NCAR/TN -485+STR, 212 pp.

  • New, M., , Washington R. , , Jack C. , , and Hewitson B. , 2003: Sensitivity of southern African climate to soil moisture. CLIVAR Exchanges, No. 8, International CLIVAR Project Office, Southampton, United Kingdom, 45–47.

  • Notaro, M., 2008: Statistical identification of global hot spots in soil moisture feedbacks among IPCC AR4 models. J. Geophys. Res., 113, D09101, doi:10.1029/2007JD009199.

    • Search Google Scholar
    • Export Citation
  • Notaro, M., , Liu Z. , , and Williams J. W. , 2006: Observed vegetation–climate feedbacks in the United States. J. Climate, 19, 763786.

  • Oglesby, R. J., , Marshall S. , , Erickson D. J. III, , Roads J. O. , , and Robertson F. R. , 2002: Thresholds in atmosphere–soil moisture interactions: Results from climate model studies. J. Geophys. Res., 107, 4224, doi:10.1029/2001JD001045.

    • Search Google Scholar
    • Export Citation
  • Oleson, K. W., and Coauthors, 2004: Technical description of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-461+STR, 174 pp.

  • Oleson, K. W., and Coauthors, 2010: Technical description of version 4.0 of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-478+STR, 257 pp.

  • Orlowsky, B., , and Seneviratne S. I. , 2010: Statistical analyses of land–atmosphere feedbacks and their possible pitfalls. J. Climate, 23, 39183932.

    • Search Google Scholar
    • Export Citation
  • Pal, J. S., , and Eltahir E. A. B. , 2001: Pathways relating soil moisture conditions to future summer precipitation within a model of the land–atmosphere system. J. Climate, 14, 12271242.

    • Search Google Scholar
    • Export Citation
  • Pal, J. S., , and Eltahir E. A. B. , 2002: Teleconnections of soil moisture and precipitation during the 1993 midwest summer flood. Geophys. Res. Lett., 29, 1865, doi:10.1029/2002GL014815.

    • Search Google Scholar
    • Export Citation
  • Rayner, N. A., , Parker D. E. , , Horton E. B. , , Folland C. K. , , Alexander L. V. , , Rowell D. P. , , Kent E. C. , , and Kaplan A. , 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.

    • Search Google Scholar
    • Export Citation
  • Ruiz-Barradas, A., , and Nigam S. , 2005: Warm season rainfall variability over the U.S. Great Plains in observations, NCEP and ERA-40 reanalyses, and NCAR and NASA atmospheric model simulations. J. Climate, 18, 18081830.

    • Search Google Scholar
    • Export Citation
  • Ruiz-Barradas, A., , and Nigam S. , 2006: Great Plains hydroclimate variability: The view from North American Regional Reanalysis. J. Climate, 19, 30043010.

    • Search Google Scholar
    • Export Citation
  • Saha S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 10151057.

  • Salvucci, G. D., , Saleem J. A. , , and Kaufmann R. , 2002: Investigating soil moisture feedbacks on precipitation with tests of Granger causality. Adv. Water Resour., 25, 13051312.

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

    • Search Google Scholar
    • Export Citation
  • Wang, G. L., , Kim Y. J. , , and Wang D. G. , 2007: Quantifying the strength of soil moisture–precipitation coupling and its sensitivity to changes in surface water budget. J. Hydrometeor., 8, 551570.

    • Search Google Scholar
    • Export Citation
  • Zeng, X. B., , Barlage M. , , Castro C. , , and Fling K. , 2010: Comparison of land–precipitation coupling strength using observations and models. J. Hydrometeor., 11, 979994.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Biome distribution for the United States from EROS Data Center’s global land cover classification dataset (adopted from Notaro et al. 2006, their Fig. 3). The classified subregions include northern Great Plains (NGp: 34.4°–49°N, 105°–96°W), southern Great Plains (SGp: 25°–34.4°N, 105°–96°W), northern shrubland (NSh: 40°–49°N, 119.4°–105°W), southern shrubland (SSh: 30.8°–40°N, 119.4°–105°W), Midwest (MW: 38°–45°N, 96°–80°W), Southeast (SE: 30°–34.5°N, 92.5°–75°W), Northwest (NW: 40°–49°N, 124°–119.4°W), and Northeast (NE: 38°–47.5°N, 80°–67°W).

  • View in gallery

    Probability distribution function of summer correlation between 1-day soil moisture and subsequent 21-day precipitation, derived with CPC–VIC data through sampling the pool of the 48-yr whole group and each of the two 24-yr subgroups (outer and inner quartiles) categorized according to the amount of summer precipitation in each region. The sample size for all correlation calculation is 24.

  • View in gallery

    Distribution of gridded PC value for pdf of correlation between 1-day soil moisture and 21-day precipitation over the United States for CPC–VIC (1950–97), NARR (1978–2007), and CFSR (1978–2007) in outer quartiles.

  • View in gallery

    Distribution of gridded PC value for pdf of correlation between 1-day soil moisture and 21-day precipitation over the United States for CAM3–CLM3 and CAM4–CLM4 during 1950–97 for outer quartiles.

  • View in gallery

    Distribution of ΔΩp (land–atmosphere coupling strength for precipitation) and ΔΩE × σE (product of land–atmosphere coupling strength for evaporation and its standard deviation) for the United States. Areas of extreme seasonal aridity (i.e., precipitation less than 0.25 mm day−1) or with negative coupling strength (considered as noise) are not shaded.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 66 66 11
PDF Downloads 43 43 7

Summer Land–Atmosphere Coupling Strength in the United States: Comparison among Observations, Reanalysis Data, and Numerical Models

View More View Less
  • 1 Department of Civil and Environmental Engineering, and Center for Environmental Sciences and Engineering, University of Connecticut, Storrs, Connecticut
© Get Permissions
Full access

Abstract

This study examines the land–atmosphere coupling strength during summer over subregions of the United States based on observations [Climate Prediction Center (CPC)–Variable Infiltration Capacity (VIC)], reanalysis data [North American Regional Reanalysis (NARR) and NCEP Climate Forecast System Reanalysis (CFSR)], and models [Community Atmosphere Model, version 3 (CAM3)–Community Land Model, version 3 (CLM3) and CAM4–CLM4]. The probability density function of conditioned correlation between soil moisture and subsequent precipitation or surface temperature during the years of large precipitation anomalies is used as a measure for the coupling strength. There are three major findings: 1) among the eight subregions (classified by land cover types), the transition zone Great Plains (and, to a lesser extent, the Midwest and Southeast) are identified as hot spots for strong land–atmosphere coupling; 2) soil moisture–precipitation coupling is weaker than soil moisture–surface temperature coupling; and 3) the coupling strength is stronger in observational and reanalysis products than in the models examined, especially in CAM4–CLM4. The conditioned correlation analysis also indicates that the coupling strength in CAM4–CLM4 is weaker than in CAM3–CLM3, which is further supported by Global Land–Atmosphere Coupling Experiments1 (GLACE1)-type experiments and attributed to changes in CAM rather than modifications in CLM. Contrary to suggestions in previous studies, CAM–CLM models do not seem to overestimate the land–atmosphere coupling strength.

Corresponding author address: Guiling Wang, Department of Civil and Environmental Engineering, University of Connecticut, 261 Glenbrook Rd., Storrs, CT 06269-2037. E-mail: gwang@engr.uconn.edu

Abstract

This study examines the land–atmosphere coupling strength during summer over subregions of the United States based on observations [Climate Prediction Center (CPC)–Variable Infiltration Capacity (VIC)], reanalysis data [North American Regional Reanalysis (NARR) and NCEP Climate Forecast System Reanalysis (CFSR)], and models [Community Atmosphere Model, version 3 (CAM3)–Community Land Model, version 3 (CLM3) and CAM4–CLM4]. The probability density function of conditioned correlation between soil moisture and subsequent precipitation or surface temperature during the years of large precipitation anomalies is used as a measure for the coupling strength. There are three major findings: 1) among the eight subregions (classified by land cover types), the transition zone Great Plains (and, to a lesser extent, the Midwest and Southeast) are identified as hot spots for strong land–atmosphere coupling; 2) soil moisture–precipitation coupling is weaker than soil moisture–surface temperature coupling; and 3) the coupling strength is stronger in observational and reanalysis products than in the models examined, especially in CAM4–CLM4. The conditioned correlation analysis also indicates that the coupling strength in CAM4–CLM4 is weaker than in CAM3–CLM3, which is further supported by Global Land–Atmosphere Coupling Experiments1 (GLACE1)-type experiments and attributed to changes in CAM rather than modifications in CLM. Contrary to suggestions in previous studies, CAM–CLM models do not seem to overestimate the land–atmosphere coupling strength.

Corresponding author address: Guiling Wang, Department of Civil and Environmental Engineering, University of Connecticut, 261 Glenbrook Rd., Storrs, CT 06269-2037. E-mail: gwang@engr.uconn.edu

1. Introduction

Land–atmosphere coupling strength, in terms of the degree of consistent impact of soil moisture on precipitation and surface temperature, has been extensively studied in the literature (e.g., Dirmeyer 2000; Koster et al. 2004, 2006, 2009, 2010, 2011; to name a few recent among many others). Soil moisture’s impact on surface temperature is well understood and well explained using models and observation (Koster et al. 2009, 2010, 2011). Specifically, higher soil moisture results in higher evaporation and therefore larger evaporative cooling, which leads to a negative correlation between soil moisture and surface temperature. However, the mechanism for the impact of soil moisture on precipitation is much more complicated (Seneviratne et al. 2010). Based on the prevailing view of positive feedback between soil moisture and precipitation, wet soil will lead to a chain of reactions including decreasing albedo and increasing net radiation, increasing latent heat flux and decreasing sensible heat flux and therefore boundary layer height, and increasing moist static energy density and eventually precipitation (e.g., Betts et al. 1994; Eltahir 1998; Findell and Eltahir 2003a,b). Under certain circumstances, however, the shift of surface fluxes from sensible to latent heat and the resultant decrease of the boundary layer depth will inhibit precipitation by increasing atmospheric stability and reducing moisture convergence (New et al. 2003; Cook et al. 2006).

Take the Midwest region for example. Numerical modeling studies have demonstrated evidences for a positive soil moisture–precipitation feedback (Bosilovich and Sun 1999; Pal and Eltahir 2001, 2002; Oglesby et al. 2002; Kim and Wang 2007). Among many other numerical studies, Kim and Wang (2007), using the coupled Community Atmosphere Model, version 3 (CAM3)–Community Land Model, version 3 (CLM3) model, found that soil moisture anomalies could affect summer precipitation over the upper Mississippi River basin (Midwest) with positive feedback, and the impact depends on the characteristics of such anomalies (e.g., their timing and magnitude). However, observational studies are still inconclusive (Findell and Eltahir 1997; Salvucci et al. 2002; D’Odorico and Porporato 2004). The study of Findell and Eltahir (1997)—an observation-based analysis supporting the positive feedback between soil moisture and precipitation in the Midwest—was brought into question by Salvucci et al. (2002), who did not find any significant correlation between soil moisture and subsequent precipitation using the same raw data. D’Odorico and Porporato (2004), also using the same data as Findell and Eltahir (1997), found that soil moisture appears to influence the frequency of subsequent precipitation, but not the amount.

A similar contrast between numerical modeling studies and observational analysis exists for the Great Plains (Koster et al. 2004, 2006, 2010, 2011; Wang et al. 2007; Dirmeyer et al. 2009; Ruiz-Barradas and Nigam 2005, 2006). The Great Plains region is identified by Koster et al. (2004, 2006) as a major region of strong soil moisture–precipitation coupling in the Global Land–Atmosphere Coupling Experiments1 (GLACE1) based on 12 general circulation models (GCMs). In the follow-up multimodel experiment GLACE2, Koster et al. (2010) found moderate contribution of land surface initialization to subseasonal precipitation prediction over limited areas in the United States including part of the Great Plains. Dirmeyer et al. (2009) analyzed the correlations among soil moisture, length of soil memory, evaporation, and recycled precipitation to assess the season dependence of land–atmosphere feedback and found that such feedback is strong through most of the year in the Great Plains region. In contrast, studies based on reanalysis data (Ruiz-Barradas and Nigam 2005, 2006) found very weak local precipitation recycling in the Great Plains and suggested that coupling strength in numerical models might have been overestimated.

This obvious dilemma—that is, overwhelming evidence from numerical models on one hand and lack of sufficient evidence from observations on the other—poses a severe obstacle for advancing our understanding of land–atmosphere coupling. One fundamental issue related to such dilemma is the incomparability between numerical modeling experiments and observational analysis. Specifically, most numerical studies examining the impact of soil moisture on subsequent precipitation were based on either an ensemble approach or one that is conceptually equivalent to an ensemble approach. That is, for a specific initial soil moisture anomaly applied in the model, results from multiple simulations (that differ either in other initial conditions or boundary conditions) are averaged to derive the representative effect on precipitation (e.g., Oglesby et al. 2002; Kim and Wang 2007), which partially eliminates the noise caused by the atmospheric internal variability. However, short observational record technically represents one member of a potential ensemble of experiments using planet Earth as the model. Therefore, addressing this mismatch might be the crux to reconcile the dilemma presented above.

Several recent studies have presented methodologies or approaches to quantify the land–atmosphere coupling strength that can be applied similarly to both observations and models. Notaro (2008) adopted a statistical approach that originated from the field of ocean–atmosphere interactions (Frankignoul and Hasselmann 1977), analyzed the global soil moisture–precipitation feedbacks for 19 models participating in the Intergovernmental Panel on Climate Change’s (IPCC) Fourth Assessment Report (AR4) using the statistical feedback parameter λ, and confirmed the location of land–atmosphere coupling hot spots such as the Great Plains found in GLACE1. However, this feedback parameter approach is cautioned by Orlowsky and Seneviratne (2010) who illustrated the potential pitfalls where high feedback parameter values can be attributed to the influence of sea surface temperature (SST). Zeng et al. (2010) proposed another statistical index Γ to quantify the land–precipitation coupling strength, and they corroborated the identified hot spots in the Great Plains as well and found that the coupling strength in models is higher than in observations. They also pointed out that a relatively high Γ is a necessary, but not sufficient, condition for a relatively strong land–precipitation coupling. Recently, Dirmeyer (2011) examined the soil moisture–evaporation link using a sensitivity index, and corroborated the “hot spots” of land–atmosphere coupling strength found in the GLACE work (Koster et al. 2006, 2011). Focusing on the Midwest and the Great Plains and using the probability density function (pdf) of conditioned correlation between soil moisture and precipitation, Mei and Wang (2011) found that soil moisture–precipitation feedback is more likely to be positive and significant during years of large precipitation anomalies and when the skill of precipitation prediction based on SST alone is low.

In this study, we adopt the approach of Mei and Wang (2011) and use the pdf of the conditioned, lagged correlation to gauge and compare the summer land–atmosphere coupling strength between observations and models over the United States. Section 2 briefly describes the regions, data, and methodology. Results are documented in section 3. Section 4 presents the summary and discussions.

2. Regions, data description, and methodology

a. Regions of interest

In this study, we divide the United States into eight subregions (Fig. 1, adopted from Notaro et al. 2006), including northern Great Plains (NGp: 34.4°–49°N, 105°–96°W), southern Great Plains (SGp: 25°–34.4°N, 105°–96°W), northern shrubland (NSh: 40°–49°N, 119.4°–105°W), southern shrubland (SSh: 30.8°–40°N, 119.4°–105°W), Midwest (MW: 38°–45°N, 96°–80°W), Southeast (SE: 30°–34.5°N, 92.5°–75°W), Northwest (NW: 40°–49°N, 124°–119.4°W), and Northeast (NE: 38°–47.5°N, 80°–67°W). These eight regions are delineated in the same way as Notaro et al. (2006) based on the simplified seven land cover types derived from U.S. Geological Survey Earth Resources Observation and Science (EROS) Data Center’s global land cover characteristic database (Loveland et al. 2001). Note that the definition of the Great Plains and the Midwest is consistent with previous studies (e.g., Koster et al. 2004; Kim and Wang 2007).

Fig. 1.
Fig. 1.

Biome distribution for the United States from EROS Data Center’s global land cover classification dataset (adopted from Notaro et al. 2006, their Fig. 3). The classified subregions include northern Great Plains (NGp: 34.4°–49°N, 105°–96°W), southern Great Plains (SGp: 25°–34.4°N, 105°–96°W), northern shrubland (NSh: 40°–49°N, 119.4°–105°W), southern shrubland (SSh: 30.8°–40°N, 119.4°–105°W), Midwest (MW: 38°–45°N, 96°–80°W), Southeast (SE: 30°–34.5°N, 92.5°–75°W), Northwest (NW: 40°–49°N, 124°–119.4°W), and Northeast (NE: 38°–47.5°N, 80°–67°W).

Citation: Journal of Hydrometeorology 13, 3; 10.1175/JHM-D-11-075.1

b. Data and model output

We will use three sets of observational or reanalysis data and output from two coupled land–atmosphere models. The observational or reanalysis data include 1) Climate Prediction Center (CPC) U.S. Unified precipitation paired with soil moisture data simulated by the Variable Infiltration Capacity (VIC) model (CPC–VIC) during the overlapping period 1950–97, 2) the National Centers for Environmental Prediction (NCEP) North American Regional Reanalysis (NARR) data during 1979–2008, and 3) the NCEP Climate Forecast System Reanalysis (CFSR) data during 1979–2008.

In the first set of observation-based data CPC–VIC, the daily CPC precipitation data—spanning the period 1948–97—is provided by the National Oceanic and Atmospheric Administration (NOAA)/Office of Oceanic and Atmospheric Research (OAR)/Earth System Research Laboratory (ESRL) Physical Sciences Division (PSD), Boulder, Colorado, from their website at http://www.esrl.noaa.gov/psd/, and covers the land area over 20°–60°N, 140°–60°W with a spatial resolution of 0.25° × 0.25°. The data is developed from multiple sources of rain gauge data and has passed through standard quality control conducted by CPC. Further details of this dataset can be found in Higgins et al. (2000). The daily VIC soil moisture data, at 0.125° × 0.125° spatial resolution over the domain of 25°–52°N, 124°–67°W, is derived from the simulation for the United States from 1950 to 2000 (Maurer et al. 2002). The VIC model, a 1D macroscale distributed hydrological model developed at the University of Washington and Princeton University (Liang et al. 1996a,b), was driven by observed meteorological forcing (with precipitation derived from NOAA cooperative observer stations, which is one of the CPC precipitation data sources) to produce the soil moisture data used here. The VIC daily soil moisture includes data of three soil layers with depth of each layer specified for each grid cell as derived during calibration. Evaluation of simulated VIC soil moisture against observations from the Soil Climate Analysis Network (SCAN) over several U.S. sites (Meng and Quiring 2008) shows that VIC generally performs well in simulating soil moisture conditions. The sum of moisture over all three soil layers (up to 40–100 cm in soil depth) will be used here.

The second set is a regional daily dataset from NARR with 32-km spatial resolution during 1979–2008. The NARR project (Mesinger et al. 2006) directly assimilates precipitation from rain gauges and radiances from satellite observations in addition to improvements to the land surface model, providing improved accuracy over its predecessor NCEP–National Center for Atmospheric Research (NCAR) Global Reanalysis. The third dataset CFSR is a 6-hourly global dataset at 0.5° spatial resolution covering the same period as NARR (1979–2008). It is preferred over previous NCEP–NCAR Global Reanalysis data because of several improvements, including the coupling to the ocean during generation of the 6-hourly guess field, an interactive sea ice model, and the assimilation of satellite radiances (Saha et al. 2010). Note that the 6-hourly CFSR data will also be aggregated to daily at first in this study.

In addition to observational and reanalysis data, we will also evaluate the land–atmosphere coupling strength based on output from the coupled CAM3–CLM3 model, which consists of CAM3 (Collins et al. 2004) and CLM3 (Oleson et al. 2004) developed at NCAR, and its updated version CAM4 (Neale et al. 2010)–CLM4 (Oleson et al. 2010). CAM simulates atmospheric dynamics and physics. The model has three different dynamic schemes: Eulerian spectral scheme, semi-Lagrangian scheme, and finite volume scheme. We choose the default Eulerian spectral dynamical core with a T42 (2.8125° × 2.8125°) horizontal resolution for CAM3 and finite volume core with f19 (1.875° × 2.5°) horizontal resolution for CAM4. One major improvement to CAM4 from CAM3 is the enhancements to the deep convection parameterization (Neale et al. 2010). CLM simulates the energy, moisture, and momentum exchanges between soil, vegetation, and the overlying atmosphere. There are comprehensive improvements from CLM3 to CLM4, including more sophisticated representations of soil hydrology and vegetation canopy hydrology with treatments of soil column–groundwater interactions and snow processes, a carbon–nitrogen (CN) model to simulate the carbon–nitrogen cycles and transient land cover change, an urban canopy model to simulate urban energy balance and climate, and an improved surface dataset with refinement of global land units and plant functional type (PFT) distributions, among others (Oleson et al. 2010). In the application to this study, the CN model in CLM4 is turned off, and vegetation conditions (with respect to PFTs and their fractional coverage, leaf area index, and seasonal variations) in both CLM3 and CLM4 are prescribed according to the observed climatology (which is MODIS based). Both CAM3–CLM3 and CAM4–CLM4 simulations are driven by interannually varying SST forcing over the period 1940–2007, which are derived from the Hadley Centre Global Sea Ice and Sea Surface Temperature dataset that gives monthly sea surface temperature (SST) in 1° grids (HadISST_1.1_SST; Rayner et al. 2003). Daily outputs from the models are used. The first 10 years will be discarded as a spinup and only data from 1950–2007 will be used in results analysis.

The CPC–VIC data analyzed include precipitation and soil moisture only. The variables analyzed from NARR, CFSR, and model outputs include precipitation, evaporative fraction (i.e., latent heat flux divided by the sum of latent and sensible heat), surface temperature (at 2 m), and soil moisture (summed up for top three layers to 100 cm for NARR and CFSR, and for top seven layers to ~83 cm for model output). For all the datasets, only soil moisture will be normalized at the daily scale and all other variables will be normalized at a 21-day scale in summer as described in the following subsection. Table 1 summarizes the datasets described above.

Table 1.

Summary of datasets used.

Table 1.

c. Methodology

For any specific dataset, we first divide the whole dataset into two categories: outer quartiles (with summer precipitation amount in the first and fourth quartiles) and inner quartiles (with summer precipitation amount in the second and third quartiles). The correlation between 1-day soil moisture and the subsequent 21-day precipitation (or surface temperature, evaporative fraction) is then examined based on the pdf of correlation coefficient in each category and the whole dataset. The total number of data pairs (e.g., 1-day soil moisture paired with subsequent 21-day precipitation) in each pool will be 92 × n for each category individually and 92 × 2 × n for the data as a whole (where 92 is the number of days in summer, and n is the number of years in each category). The pdf is derived from 10 000 repeated calculations of correlation, each based on 24 data pairs randomly drawn without replacement from the pool of each category or from the whole data, with a constraint to exclude temporal overlapping of precipitation between any two data pairs. Further details can be found in Mei and Wang (2011). Note that 1-day soil moisture and the subsequent 21-day precipitation (or surface temperature, evaporative fraction) is chosen for analysis because it accommodates a comparison between our study and previous studies (Findell and Eltahir 1997; D’Odorico and Porporato 2004), and is consistent with the time scale of interest here (i.e., subseasonal scale). Using a different time interval for precipitation aggregation or inserting a longer time lag between soil moisture and precipitation do not qualitatively change the results of this study.

Our previous study (Mei and Wang 2011) found that the summer soil moisture–precipitation feedback is more likely to be positive and significant in the years of outer quartiles than in the years of inner quartiles. This is consistent with findings from the GLACE2 study (Koster et al. 2010, 2011) that the contribution of realistic soil moisture initialization to subseasonal forecast is significant only when the magnitude of initial soil moisture anomaly is large. To further illustrate this concept, here we apply the Mei and Wang (2011) approach to the eight subregions of the United States using the CPC–VIC dataset. Figure 2 presents the probability distribution function (pdf) of correlation between 1-day soil moisture and subsequent 21-day precipitation for years of outer quartiles, inner quartiles, and the whole dataset. It is clear from Fig. 2 that the summer soil moisture–precipitation correlation for the whole dataset tends to be positive in most of the regions; such positive feedback signal is obviously amplified in outer quartiles, which is consistent with the soil moisture threshold behavior (i.e., the impact of imposed soil moisture anomalies on subsequent precipitation is only significant when the anomalies exceed a certain threshold) found in Oglesby et al. (2002). In inner quartiles, the correlations are mostly negative, indicating a negative soil moisture–precipitation feedback, although the signal is weak compared with the outer quartiles.

Fig. 2.
Fig. 2.

Probability distribution function of summer correlation between 1-day soil moisture and subsequent 21-day precipitation, derived with CPC–VIC data through sampling the pool of the 48-yr whole group and each of the two 24-yr subgroups (outer and inner quartiles) categorized according to the amount of summer precipitation in each region. The sample size for all correlation calculation is 24.

Citation: Journal of Hydrometeorology 13, 3; 10.1175/JHM-D-11-075.1

In addition to the soil moisture–precipitation correlation, also analyzed are the correlation between soil moisture and surface temperature and the correlation between soil moisture and evaporative fraction for the outer quartiles and inner quartiles of summer precipitation. Here we use the evaporative fraction instead of evaporation alone to account for the effects of net surface radiation on soil moisture–evaporation coupling (Koster et al. 2009). Note that the daily evaporative fraction is calculated based on daily latent and sensible heat flux instead of average of subdaily evaporative fraction.

To facilitate the comparison between different products, we define two indicators with respect to the pdf of lagged conditioned correlation: one is the correlation coefficient value with peak probability density (PC) to represent the magnitude of correlation; the other is the fraction of correlation values smaller (larger) than zero as the indicator for the significance of positive (negative) correlation (SI). PC values with a SI less than 0.1 will be considered significant in this study. In the following text, for convenience, both observational and reanalysis products will be referred to as “observations.” Note that the focus of this study is on comparing results from observations and those from models, recognizing that inconsistency among the three observations or between the two models does exist.

3. Results

a. Soil moisture correlation with precipitation

Table 2 presents values of PC and SI for pdf of correlation between 1-day soil moisture and 21-day precipitation in outer quartiles using different datasets spanning different periods, and the multidataset average of PC values are listed in the last column. Common to the three observational datasets (in the upper panel of Table 2), soil moisture–precipitation correlations in outer quartiles are positive and significant in the Great Plains (NGp and SGp), Midwest, Southeast, and Northeast, and insignificant in the Northwest and shrubland (NSh and SSh). The identified hot spots of soil moisture–precipitation feedback and the comparison among different subregions are mostly confirmed by model output during the two different periods presented in the lower panel of Table 2. However, the models are unable to produce the significant correlation in the Northeast found in observations. Note that the use of 0.1 as a cutoff value for SI to assess the significance of PC value in the above analysis is arbitrary. However, using a different cutoff value will not change the comparison among different regions and products (which is the focus of this study), even though the significance level of an individual product or region will change. For example, if 0.05 is used, only the Great Plains stand out with significant soil moisture–precipitation correlation in both observations and model outputs, indicating that the identification of the Great Plains as a hot spot is more robust.

Table 2.

Values of PC and SI (in parentheses) for pdf of correlation between 1-day soil moisture and 21-day precipitation for different datasets over the eight regions in outer quartiles: values with SI less than 0.1 are in bold and italic; dataset-averaged PC values presented in the last column are also in bold and italic when consensus view shows significance.

Table 2.

Similar to Table 2, Table 3 presents values of PC and SI for pdf of correlation between 1-day soil moisture and 21-day precipitation for inner quartiles. Consistent with Fig. 2 (which presents results for the CPC–VIC dataset only), correlations in inner quartiles are negative but insignificant, and their magnitudes for observational datasets and model outputs are at a similar level. While this negative correlation may be related to the hypothesized negative soil moisture–precipitation feedback (Cook et al. 2006) discussed in the introduction, the physical mechanisms underlying this negative feedback remains a topic to further explore in follow-up studies and is beyond the scope of this paper. Because of the insignificance of negative correlation in inner quartiles and the strong evidence for the widely held notion of positive soil moisture–precipitation feedback in outer quartiles, the rest of the results analysis will focus on comparison of coupling strength among different products in outer quartiles only.

Table 3.

Values of PC and SI (in parentheses) for pdf of correlation between 1-day soil moisture and 21-day precipitation for different datasets over the eight regions in inner quartiles: values with SI less than 0.1 are in bold and italic; dataset-averaged PC values presented in the last column are also in bold and italic when consensus view shows significance.

Table 3.

Based on the averaged PC values (last column in Table 2), the correlations are stronger in observations than in models for all regions except for the north Great Plains where the two are about the same. This contradicts findings from previous studies (Ruiz-Barradas and Nigam 2005, 2006; Zeng et al. 2010) that suggested overestimation of coupling strength by the models. This statement also holds true if the comparison is done between observations and models for each of the two different time periods (i.e., CPC–VIC versus CAM–CLM during 1950–97, and NARR and CFSR during 1979–2008 versus CAM–CLM during 1978–2007).

According to the theoretical explanation given in Koster et al. (2004), strong soil moisture–precipitation coupling can only occur in regions where evaporation is suitably high but still sensitive enough to soil moisture variation, and these regions are in the climate transition zones. Consistent with the results in Table 2, both our observational and model results identified the transition zone between dry and wet climates (the Great Plains) as a hot spot of coupling, which also agrees with previous studies including the GLACE1 project involving 12 GCMs (Koster et al. 2006); the statistical feedback parameter analysis of output from 19 IPCC AR4 models (Notaro 2008); the composite analysis of coupling strength with soil moisture memory, soil moisture–evaporation connection, and precipitation recycling (Dirmeyer et al. 2009); and Γ index statistical analysis of observations and model output (Zeng et al. 2010). Figure 3 shows the grid-based PC values for outer quartiles in each of the three observational datasets. All three datasets have been gridded to T42 resolution from their native resolution before calculation of gridded PC. Note that unlike the correlations based on averages of precipitation and soil moisture over each subregion, the point-to-point correlation in Fig. 3 does not account for the impact from neighboring grid points (thus potentially leading to underestimation). Although gridded PC values show more inconsistency among all the three datasets (in Fig. 3) than that captured in Table 2, all of them have shown signatures over part of the Great Plains, Midwest, and Southeast. As presented in Fig. 3, maximum PC values are located more in Mexico (i.e., part of southern Great Plains as defined in this study) for NARR and in the eastern United States (i.e., Midwest and Southeast as defined in this study) for CFSR. These appear to be the locations found by Findell et al. (2011) where probability of afternoon precipitation is enhanced by before-noon high evaporation. However, the emphasis of the Findell et al. (2011) study is on precipitation sensitivity to latent and sensible heat fluxes (as opposed to precipitation correlation with soil moisture changes in our present study).

Fig. 3.
Fig. 3.

Distribution of gridded PC value for pdf of correlation between 1-day soil moisture and 21-day precipitation over the United States for CPC–VIC (1950–97), NARR (1978–2007), and CFSR (1978–2007) in outer quartiles.

Citation: Journal of Hydrometeorology 13, 3; 10.1175/JHM-D-11-075.1

Another important result is the strong contrast between CAM3–CLM3 and CAM4–CLM4 in the lagged correlation between soil moisture and precipitation shown in Table 2, with the correlation coefficient in CAM4–CLM4 much lower than that in CAM3–CLM3. Such contrast is spatially extensive, and stands out in the spatial map of the PC value for each grid cell over the United States during 1950–97 (Fig. 4). Results for the period 1978–2007 support a similar contrast (not shown here). The decrease of coupling strength from CAM3–CLM3 to CAM4–CLM4 will be further examined in section 3c.

Fig. 4.
Fig. 4.

Distribution of gridded PC value for pdf of correlation between 1-day soil moisture and 21-day precipitation over the United States for CAM3–CLM3 and CAM4–CLM4 during 1950–97 for outer quartiles.

Citation: Journal of Hydrometeorology 13, 3; 10.1175/JHM-D-11-075.1

b. Soil moisture correlation with temperature and evaporative fraction

Table 4 is similar to Table 2, but without CPC–VIC, and for correlation between 1-day soil moisture and 21-day surface temperature in outer quartiles. In observations, most of the regions excluding the shrubland present significant negative correlations, indicating strong coupling between soil moisture and surface temperature. In contrast, the model cannot detect such observed signal in the Northwest and Northeast. It instead presents a strong and significant correlation in the northern shrubland that is not found in observations. Similar to soil moisture–precipitation correlations, the averaged correlations (last column in Table 3) between soil moisture and surface temperature in observations are stronger than those in models for all regions except the northern Great Plains and northern shrubland.

Table 4.

Values of PC and SI (in parentheses) for pdf of correlation between 1-day soil moisture and 21-day surface temperature for different datasets over the eight regions in outer quartiles: values with SI less than 0.1 are in bold and italic; dataset-averaged PC values presented in the last column are also in bold and italic when consensus view shows significance.

Table 4.

Consistent with the more straightforward effects of soil moisture on temperature than on precipitation (as explained in the introduction section), our results show that the soil moisture–surface temperature correlation (Table 4) is generally stronger and more significant than correlation between soil moisture and precipitation (Table 2). This confirms the key results of the recent GLACE2 study (Koster et al. 2010, 2011)—that is, the contribution of realistic initialization of soil moisture to subseasonal forecast is more promising for surface temperature than precipitation in the United States.

Analogous to Tables 2 and 4, Table 5 documents the PC and SI values of pdf of correlation between 1-day soil moisture and 21-day evaporative fraction in outer quartiles. It is obvious from Table 5 that the soil moisture–evaporation correlations are significant for all regions in observations, and for all regions but the Northeast in models. They are also stronger compared with the correlations between soil moisture and precipitation or surface temperature. This may be an indication that soil moisture–evaporation relationship is a potentially important pathway for land–atmosphere coupling to take place, particularly for the coupling between soil moisture and surface temperature (Koster et al. 2009, 2010, 2011). For all eight regions considered, the correlations between soil moisture and evaporative fraction are stronger in observations than in models.

Table 5.

Values of PC and SI (in parentheses) for pdf of correlation between 1-day soil moisture and 21-day evaporative fraction for different datasets over the eight regions in outer quartiles: values with SI less than 0.1 are in bold and italic; dataset-averaged PC values presented in the last column are also in bold and italic when consensus view shows significance.

Table 5.

The soil moisture–temperature correlation in inner quartiles in all regions (results not shown) has the same sign (negative) as, but is much weaker than, the correlation in outer quartiles. Similarly, the correlation between soil moisture and evaporative fraction in inner quartiles for all regions (results not shown) has the same sign (positive) as, but is much weaker than, the correlation in outer quartiles. In the inner quartiles, the correlation between soil moisture and evaporative fraction is stronger in observations than in models, while the correlation between soil moisture and temperature in observations and models are at a similar level.

Koster et al. (2011) found that the effects of soil moisture initializations are asymmetric between wet and dry cases, with the forecast skills for surface temperature being higher in wet climate zones when initial soil moisture is drier and in dry climate zones when initial soil moisture is wetter—a behavior that is consistent with our theoretical understanding of the role of soil moisture in different evaporative regimes (Koster et al. 2009). As a comparison with the contrast between inner and outer quartiles, here we also conduct our correlation pdf analysis based on wet and dry categorization of the data for both the soil moisture–temperature relationship and soil moisture–evaporative fraction relationship (results not shown). The contrast between inner and outer quartiles is much stronger than the contrast between dry and wet categories. The results for the wet years and dry years (categorized according to summer total precipitation) show a similar level of correlation in terms of the spatial pattern and magnitude. When the wet and dry categorization is based on soil moisture of each day (as in Koster et al. 2011), the correlation in the wet eastern half of the United States is slightly stronger in the dry category than in the wet category, and in the dry western half slightly stronger in the wet category than in the dry category for NARR, which is generally consistent with the Koster et al. (2011) finding. However, this is not demonstrated in CAM–CLM models and CFSR. This lack of consistency might be related to the limitation of our correlation approach (i.e., linear correlation cannot capture the potential nonlinearity of the land–atmosphere coupling), while this is not an issue for the modeling approach in Koster et al. (2011). This further highlights the challenges facing observational data analysis and the critical need for developing a creative approach to better understand and characterize the land–atmosphere coupling.

c. Global Land–Atmosphere Coupling Experiment

As mentioned in section 3a and demonstrated in Tables 2 and 4, the correlation between soil moisture and precipitation (or surface temperature) in CAM4–CLM4 is generally weaker than in CAM3–CLM3. But this is not the case for the correlation of evaporative fraction (Table 5). To further investigate and understand this issue, we adopt the GLACE1 approach and conducted experiments (Koster et al. 2006) composed of two ensembles: a W ensemble of 16 simulations that differ in their initial conditions with soil moisture allowed to evolve freely, and an S ensemble that is otherwise the same as W except that soil moisture of all 10 layers across all 16 simulations is forced to have the same values derived from one member of the W ensemble. This is done for both CAM3–CLM3 and CAM4–CLM4, respectively. A statistic Ω (Koster et al. 2000) is then derived to quantify the similarity of variables of our interest (e.g., precipitation) across all members of each ensemble, and the difference between the two ensembles ΔΩ = Ωs − Ωw is used to quantify the soil moisture–precipitation coupling strength. More details of the experimental design and coupling strength index calculations can be found in Koster et al. (2006).

Figure 5 plots the soil moisture–precipitation coupling strength ΔΩp for CAM3–CLM3 and CAM4–CLM4, respectively. From version 3 to 4 of CAM–CLM, there is a spatially significant decrease of soil moisture–precipitation coupling strength in the Great Plains (left panels in Fig. 5), which confirms our findings based on soil moisture–precipitation correlation in section 3a. Note that the decrease of coupling strength in terms of spatial extent is similar to the decrease of correlation in Fig. 4. A smaller but still significant decrease also occurs to the soil moisture–surface temperature coupling strength index (not shown here) that is consistent with the decrease of correlations between soil moisture and surface temperature from CAM3–CLM3 to CAM4–CLM4 (Table 4). Based on theoretical understanding of regional characteristics favoring strong land–atmosphere coupling strength (Koster et al. 2004), Guo et al. (2006) proposed to use ΔΩE × σE, the product of coupling strength for evaporation ΔΩE and the standard deviation of evaporation σE, to quantify the soil moisture–evaporation link. Despite the substantial difference in the soil moisture–precipitation coupling strength, the change in soil moisture–evaporation coupling (i.e., ΔΩE × σE; right column in Fig. 5) from version 3 to 4 are less obvious, although there is some degree of spatial shift in the strength of soil moisture–evaporation coupling. Collectively, these results indicate that the strong decrease of soil moisture–precipitation coupling strength from version 3 to 4 should have originated from changes in CAM, which may be caused by the modification of deep convection parameterization in CAM4 (Neale et al. 2010), since convection in CAM3 is diagnosed to be oversensitive to surface heat flux exchanges (Guo et al. 2006).

Fig. 5.
Fig. 5.

Distribution of ΔΩp (land–atmosphere coupling strength for precipitation) and ΔΩE × σE (product of land–atmosphere coupling strength for evaporation and its standard deviation) for the United States. Areas of extreme seasonal aridity (i.e., precipitation less than 0.25 mm day−1) or with negative coupling strength (considered as noise) are not shaded.

Citation: Journal of Hydrometeorology 13, 3; 10.1175/JHM-D-11-075.1

4. Summary and discussions

In this study, we have examined land–atmosphere coupling strength during summer over subregions of the United States based on observations (CPC–VIC), reanalysis data (NARR and CFSR), and models (CAM3–CLM3 and CAM4–CLM4). The probability density function of conditioned correlation between soil moisture and subsequent precipitation or surface temperature during precipitation outer- and inner-quartile years are treated separately, and the correlations for outer quartiles (which are stronger and more significant than those in inner quartiles) are used as a measure for the coupling strength to facilitate the comparison among different products. Further, GLACE1-type experiments are conducted and analyzed to complement the correlation analysis. The main results are summarized in the following.

  1. The correlation analysis from both observations and models identify the Great Plains (i.e., NGp and SGp in this study) as hot spots for strong land–atmosphere coupling, which is consistent with previous studies using different methodologies (e.g., Koster et al. 2006; Notaro 2008; Dirmeyer et al. 2009; Zeng et al. 2010). In addition, Midwest and Southeast also stand out with rather strong correlations.
  2. The soil moisture–precipitation coupling is weaker than soil moisture–surface temperature coupling, as supported by both correlation analysis and GLACE1-type experiments. These may be a reflection of the rather straightforward effects of soil moisture–induced evaporative cooling and the more complicated soil moisture–precipitation relationship that involves competing mechanisms.
  3. The CAM–CLM models underestimate the land–atmosphere coupling strength as compared with both observational data and reanalysis data. This may be related to the underestimation of soil moisture–evaporation correlation in the model (relative to reanalysis; Table 5), which would indicate the need for improving the parameterization of evapotranspiration response to soil water stress in CLM. Further scrutiny is needed in follow-up studies.
  4. From CAM3–CLM3 to CAM4–CLM4, the coupling strength decreases according to both correlation analysis and GLACE1-type experiments. This decrease most likely results from modifications to the convection scheme in CAM (Neale et al. 2010) rather than changes in CLM (Oleson et al. 2010). The underestimation of land–atmosphere coupling strength found in the CAM–CLM models contradicts previous studies (e.g., Ruiz-Barradas and Nigam 2005, 2006; Zeng et al. 2010) that suggest an overestimation of coupling strength in climate models.

Ruiz-Barradas and Nigam (2005, 2006) investigated the precipitation recycling over the Great Plains by regressing evaporation and stationary (or transient) horizontal moisture flux, respectively, over normalized precipitation with no time lag in between, and found that the regression coefficient for evaporation is much larger than that for moisture flux in models and the opposite occurs in observations and reanalysis data; Zeng et al. (2010) developed the index Γ—a dimensionless term that is conceptually equivalent to the regression coefficient from regressing concurrent evaporation over precipitation—to quantify the land–atmosphere coupling strength, and found that the index is stronger in models. Although the equation used to estimate the feedback parameter of Notaro (2008) involves precipitation at different times, the feedback parameter represents the immediate, not lagged, impact of soil moisture on precipitation. Because of the “no time-lag” nature of their approaches in these studies, their results cannot exclude the dominant impact of atmosphere on land within the coupled land–atmosphere system. In contrast, our pdf method focuses specifically on the impact of land on atmosphere.

The sensitivity index in Dirmeyer (2011), accounting for not only the sensitivity of evaporation to soil moisture changes but also the variation of soil moisture to imprint a signal on evaporation, is conceptually similar to the statistic of Guo et al. (2006) and aims to quantify the soil moisture–evaporation link (i.e., terrestrial segment); the statistic in Findell et al. (2011), in contrast, focuses on the atmospheric segment: the evaporation–precipitation link. Our pdf method examines the soil moisture–precipitation link, but treating years of larger precipitation anomalies separately from years of close-to-normal conditions. Mei and Wang (2011) found that SST performs better in predicting precipitation during years with small precipitation anomalies, and soil moisture plays a significant complementary role to SST in predicting precipitation during the years of large precipitation anomalies. This, together with the finding that accurate soil moisture initialization contributes more to improving precipitation prediction in extreme years (Koster et al. 2010, 2011), supports our primary focus on outer quartiles in the present study. These also suggest that developing statistical forecast models in outer quartiles may be more useful in operational forecasts, and our study can serve as a contribution to that direction.

It has to be recognized that this study is limited by the intrinsic limitation of statistical analysis. Correlation, even lagged correlation in our study, is not evidence for causality. Therefore, further research using other approaches, such as precipitation recycling ratio analysis (e.g., Eltahir and Bras 1996; Dominguez et al. 2006), focusing on comparison between extreme years and nonextreme years and between flood years and drought years, might be helpful for the comparison of land–atmosphere interaction between observations and models.

Acknowledgments

This research is supported by funding from NOAA CPPA (Grant NA08OAR4310871). We thank the two anonymous reviewers for their constructive comments on an earlier version of the manuscript.

REFERENCES

  • Betts, A. K., , Ball J. H. , , Beljaars A. C. M. , , Miller M. J. , , and Viterbo P. , 1994: Coupling between land-surface, boundary-layer parameterizations and rainfall on local and regional scales: Lessons from the wet summer of 1993. Preprints, Fifth Symp. on Global Change Studies, Nashville, TN, Amer. Meteor. Soc., 174–181.

    • Search Google Scholar
    • Export Citation
  • Bosilovich, M. G., , and Sun W. Y. , 1999: Numerical simulation of the 1993 Midwestern flood: Land–atmosphere interactions. J. Climate, 12, 14901505.

    • Search Google Scholar
    • Export Citation
  • Collins, W. D., and Coauthors, 2004: Description of the NCAR Community Atmosphere Model (CAM 3.0). NCAR Tech. Rep. NCAR/TN-464+STR, 226 pp.

  • Cook, B. I., , Bonan G. B. , , and Levis S. , 2006: Soil moisture feedbacks to precipitation in southern Africa. J. Climate, 19, 41984206.

  • Dirmeyer, P. A., 2000: Using a global soil wetness dataset to improve seasonal climate simulation. J. Climate, 13, 29002922.

  • Dirmeyer, P. A., 2011: The terrestrial segment of soil moisture–climate coupling. Geophys. Res. Lett., 38, L16702, doi:10.1029/2011GL048268.

    • Search Google Scholar
    • Export Citation
  • Dirmeyer, P. A., , Schlosser C. A. , , and Brubaker K. L. , 2009: Precipitation, recycling, and land memory: An integrated analysis. J. Hydrometeor., 10, 278288.

    • Search Google Scholar
    • Export Citation
  • D’Odorico, P., , and Porporato A. , 2004: Preferential states in soil moisture and climate dynamics. Proc. Natl. Acad. Sci. USA, 101, 88488851.

    • Search Google Scholar
    • Export Citation
  • Dominguez, F., , Kumar P. , , Liang X. Z. , , and Ting M. , 2006: Impact of atmospheric moisture storage on precipitation recycling. J. Climate, 19, 15131530.

    • Search Google Scholar
    • Export Citation
  • Eltahir, E. A. B., 1998: A soil moisture–rainfall feedback mechanism: 1. Theory and observations. Water Resour. Res., 34, 765776.

  • Eltahir, E. A. B., , and Bras R. L. , 1996: Precipitation recycling. Rev. Geophys., 34, 367378.

  • Findell, K. L., , and Eltahir E. A. B. , 1997: An analysis of the soil moisture-rainfall feedback, based on direct observations from Illinois. Water Resour. Res., 33, 725735.

    • Search Google Scholar
    • Export Citation
  • Findell, K. L., , and Eltahir E. A. B. , 2003a: Atmospheric controls on soil moisture–boundary layer interactions. Part I: Framework development. J. Hydrometeor., 4, 552569.

    • Search Google Scholar
    • Export Citation
  • Findell, K. L., , and Eltahir E. A. B. , 2003b: Atmospheric controls on soil moisture–boundary layer interactions. Part II: Feedbacks within the continental United States. J. Hydrometeor., 4, 570572.

    • Search Google Scholar
    • Export Citation
  • Findell, K. L., , Gentine P. , , Linter B. R. , , and Kerr C. , 2011: Probability of afternoon precipitation in eastern United States and Mexico enhanced by high evaporation. Nat. Geosci., 4, 434439, doi:10.1038/NGEO1174.

    • Search Google Scholar
    • Export Citation
  • Frankignoul, C., , and Hasselmann K. , 1977: Stochastic climate models, part II. Application to sea-surface temperature anomalies and thermocline variability. Tellus, 29, 289305.

    • Search Google Scholar
    • Export Citation
  • Guo, Z., and Coauthors, 2006: GLACE: The Global Land–Atmosphere Coupling Experiment. Part II: Analysis. J. Hydrometeor., 7, 611625.

    • Search Google Scholar
    • Export Citation
  • Higgins, R. W., , Shi W. , , Yarosh E. , , and Joyce R. , 2000: Improved United States precipitation quality control system and analysis. NCEP/Climate Prediction Center ATLAS 7, Camp Springs, MD, 40 pp.

  • Kim, Y. J., , and Wang G. L. , 2007: Impact of initial soil moisture anomalies on subsequent precipitation over North America in the coupled land–atmosphere model CAM3–CLM3. J. Hydrometeor., 8, 513533.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., , Suarez M. J. , , and Heiser M. , 2000: Variance and predictability of precipitation at seasonal-to-interannual time scales. J. Hydrometeor., 1, 2646.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 11381140.

  • Koster, R. D., and Coauthors, 2006: GLACE: The Global Land–Atmosphere Coupling Experiment. Part I: Overview. J. Hydrometeor., 7, 590610.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., , Schubert S. D. , , and Suarez M. J. , 2009: Analyzing the concurrence of meteorological droughts and warm periods, with implications for the determination of evaporative regime. J. Climate, 22, 33313341.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2010: The contribution of land surface initialization to subseasonal forecast skill: First results from a multi-model experiment. Geophys. Res. Lett., 37, L02402, doi:10.1029/2009GL041677.

    • 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.

    • Search Google Scholar
    • Export Citation
  • Liang, X., , Lettenmaier D. P. , , and Wood E. F. , 1996a: One-dimensional statistical dynamic representation of subgrid spatial variability of precipitation in the two-layer variable infiltration capacity model. J. Geophys. Res., 101 (D16), 21 40321 422.

    • Search Google Scholar
    • Export Citation
  • Liang, X., , Wood E. F. , , and Lettenmaier D. P. , 1996b: Surface soil moisture parameterization of the VIC-2L model: Evaluation and modifications. Global Planet. Change, 13, 195206.

    • Search Google Scholar
    • Export Citation
  • Loveland, T. R., , Reed B. C. , , Brown J. F. , , Ohlen D. O. , , Zhu J. , , Yang L. , , and Merchant J. W. , 2001: Development of a global land cover characteristics database and IGBP DISCover from 1-km AVHRR data. Int. J. Remote Sens., 21, 13031330.

    • Search Google Scholar
    • Export Citation
  • Maurer, E. P., , Wood A. W. , , Adam J. C. , , Lettenmaier D. P. , , and Nijssen B. , 2002: A long-term hydrologically based dataset of land surface fluxes and states for the conterminous United States. J. Climate, 15, 32373251.

    • Search Google Scholar
    • Export Citation
  • Mei, R., , and Wang G. L. , 2011: Impact of sea surface temperature and soil moisture on summer precipitation in the United States based on observational data. J. Hydrometeor., 12, 10861099.

    • Search Google Scholar
    • Export Citation
  • Meng, L., , and Quiring S. M. , 2008: A comparison of soil moisture models using Soil Climate Analysis Network observations. J. Hydrometeor., 9, 641659.

    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360.

  • Neale, R. B., and Coauthors, 2010: Description of the NCAR Community Atmosphere Model (CAM 4.0). NCAR Tech. Rep. NCAR/TN -485+STR, 212 pp.

  • New, M., , Washington R. , , Jack C. , , and Hewitson B. , 2003: Sensitivity of southern African climate to soil moisture. CLIVAR Exchanges, No. 8, International CLIVAR Project Office, Southampton, United Kingdom, 45–47.

  • Notaro, M., 2008: Statistical identification of global hot spots in soil moisture feedbacks among IPCC AR4 models. J. Geophys. Res., 113, D09101, doi:10.1029/2007JD009199.

    • Search Google Scholar
    • Export Citation
  • Notaro, M., , Liu Z. , , and Williams J. W. , 2006: Observed vegetation–climate feedbacks in the United States. J. Climate, 19, 763786.

  • Oglesby, R. J., , Marshall S. , , Erickson D. J. III, , Roads J. O. , , and Robertson F. R. , 2002: Thresholds in atmosphere–soil moisture interactions: Results from climate model studies. J. Geophys. Res., 107, 4224, doi:10.1029/2001JD001045.

    • Search Google Scholar
    • Export Citation
  • Oleson, K. W., and Coauthors, 2004: Technical description of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-461+STR, 174 pp.

  • Oleson, K. W., and Coauthors, 2010: Technical description of version 4.0 of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-478+STR, 257 pp.

  • Orlowsky, B., , and Seneviratne S. I. , 2010: Statistical analyses of land–atmosphere feedbacks and their possible pitfalls. J. Climate, 23, 39183932.

    • Search Google Scholar
    • Export Citation
  • Pal, J. S., , and Eltahir E. A. B. , 2001: Pathways relating soil moisture conditions to future summer precipitation within a model of the land–atmosphere system. J. Climate, 14, 12271242.

    • Search Google Scholar
    • Export Citation
  • Pal, J. S., , and Eltahir E. A. B. , 2002: Teleconnections of soil moisture and precipitation during the 1993 midwest summer flood. Geophys. Res. Lett., 29, 1865, doi:10.1029/2002GL014815.

    • Search Google Scholar
    • Export Citation
  • Rayner, N. A., , Parker D. E. , , Horton E. B. , , Folland C. K. , , Alexander L. V. , , Rowell D. P. , , Kent E. C. , , and Kaplan A. , 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.

    • Search Google Scholar
    • Export Citation
  • Ruiz-Barradas, A., , and Nigam S. , 2005: Warm season rainfall variability over the U.S. Great Plains in observations, NCEP and ERA-40 reanalyses, and NCAR and NASA atmospheric model simulations. J. Climate, 18, 18081830.

    • Search Google Scholar
    • Export Citation
  • Ruiz-Barradas, A., , and Nigam S. , 2006: Great Plains hydroclimate variability: The view from North American Regional Reanalysis. J. Climate, 19, 30043010.

    • Search Google Scholar
    • Export Citation
  • Saha S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 10151057.

  • Salvucci, G. D., , Saleem J. A. , , and Kaufmann R. , 2002: Investigating soil moisture feedbacks on precipitation with tests of Granger causality. Adv. Water Resour., 25, 13051312.

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

    • Search Google Scholar
    • Export Citation
  • Wang, G. L., , Kim Y. J. , , and Wang D. G. , 2007: Quantifying the strength of soil moisture–precipitation coupling and its sensitivity to changes in surface water budget. J. Hydrometeor., 8, 551570.

    • Search Google Scholar
    • Export Citation
  • Zeng, X. B., , Barlage M. , , Castro C. , , and Fling K. , 2010: Comparison of land–precipitation coupling strength using observations and models. J. Hydrometeor., 11, 979994.

    • Search Google Scholar
    • Export Citation
Save