• Arkin, P. A., , Joyce R. J. , , and Janowiak J. E. , 1994: IR techniques: GOES precipitation index. Remote Sens. Rev., 11, 107124, doi:10.1080/02757259409532261.

    • Search Google Scholar
    • Export Citation
  • Atlas, D., 1990: Radar in Meteorology: Battan Memorial and 40th Anniversary Radar Meteorology Conference. Amer. Meteor. Soc., 806 pp.

  • Awaka, J., , Iguchi T. , , and Okamoto K. , 1998: Early results on rain type classification by the Tropical Rainfall Measuring Mission (TRMM) precipitation radar. Proc. 8th URSI Commission F Triennial Open Symp. on Wave Propagation and Remote Sensing, Aveiro, Portugal, URSI, 143146.

  • Awaka, J., , Iguchi T. , , and Okamoto K. , 2009: TRMM PR standard algorithm 2A23 and its performance on bright band detection. J. Meteor. Soc. Japan, 87A, 3152, doi:10.2151/jmsj.87A.31.

    • Search Google Scholar
    • Export Citation
  • Berg, W., , L’Ecuyer T. , , and Kummerow C. , 2006: Rainfall climate regimes: The relationship of regional TRMM rainfall biases to the environment. J. Appl. Meteor. Climatol., 45, 434454, doi:10.1175/JAM2331.1.

    • Search Google Scholar
    • Export Citation
  • Berg, W., , L’Ecuyer T. , , and Haynes J. M. , 2010: The distribution of rainfall over oceans from spaceborne radars. J. Appl. Meteor. Climatol., 49, 535543, doi:10.1175/2009JAMC2330.1.

    • Search Google Scholar
    • Export Citation
  • Bowman, K. P., 2005: Comparison of TRMM precipitation retrievals with rain gauge data from ocean buoys. J. Climate, 18, 178190, doi:10.1175/JCLI3259.1.

    • Search Google Scholar
    • Export Citation
  • Ferraro, R. R., 1997: Special Sensor Microwave Imager derived global rainfall estimates for climatological applications. J. Geophys. Res., 102, 16 71516 735, doi:10.1029/97JD01210.

    • Search Google Scholar
    • Export Citation
  • Ferraro, R. R., , Smith E. A. , , Berg W. , , and Huffman G. J. , 1998: A screening methodology for passive microwave precipitation retrieval algorithms. J. Atmos. Sci., 55, 15831600, doi:10.1175/1520-0469(1998)055<1583:ASMFPM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Ferraro, R. R., and et al. , 2012: An evaluation of microwave land surface emissivities over the continental United Stated to benefit GPM-era precipitation algorithms. IEEE Trans. Geosci. Remote Sens.,51, 378–398, doi:10.1109/TGRS.2012.2199121.

  • Funk, A., , and Schumacher C. , 2013: Analysis of rain classification over tropics by version 7 of TRMM 2A23 algorithm. J. Meteor. Soc. Japan, 91, 257272, doi:10.2151/jmsj.2013-302.

    • Search Google Scholar
    • Export Citation
  • Gopalan, K., , Wang N.-Y. , , Ferraro R. , , and Liu C. , 2010: Status of the TRMM 2A12 land precipitation algorithm. J. Atmos. Oceanic Technol., 27, 13431354, doi:10.1175/2010JTECHA1454.1.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., , Adler R. F. , , Morrissey M. M. , , Bolvin D. T. , , Cuttis S. , , Joyce R. , , McGavock B. , , and Susskind J. , 2001: Global precipitation at one degree resolution from multisatellite observations. J. Hydrometeor., 2, 3650, doi:10.1175/1525-7541(2001)002<0036:GPAODD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., and et al. , 2007: The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeor., 8, 3855, doi:10.1175/JHM560.1.

    • Search Google Scholar
    • Export Citation
  • Iguchi, T., , Kozu T. , , Meneghini R. , , Awaka J. , , and Okamoto K. , 2000: Rain-profiling algorithm for the TRMM precipitation radar. J. Appl. Meteor., 39, 20382052, doi:10.1175/1520-0450(2001)040<2038:RPAFTT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Iguchi, T., , Kozu T. , , Kwiatkowski J. , , Meneghini R. , , Awaka J. , , and Okamoto K. , 2009: Uncertainties in the rain profiling algorithm for the TRMM precipitation radar. J. Meteor. Soc. Japan,87, 1–30, doi:10.2151/jmsj.87A.1.

  • Kummerow, C., , Barnes W. , , Kozu T. , , Shiue J. , , and Simpson J. , 1998: The Tropical Rainfall Measuring Mission (TRMM) sensor package. J. Atmos. Oceanic Technol., 15, 809817, doi:10.1175/1520-0426(1998)015<0809:TTRMMT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kummerow, C., and et al. , 2001: The evolution of the Goddard profiling algorithm (GPROF) for rainfall estimation from passive microwave sensors. J. Appl. Meteor., 40, 18011820, doi:10.1175/1520-0450(2001)040<1801:TEOTGP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kummerow, C., , Ringerud S. , , Crook J. , , Randel D. , , and Berg W. , 2011: An observationally generated a priori database for microwave rainfall retrievals. J. Atmos. Oceanic Technol., 28, 113130, doi:10.1175/2010JTECHA1468.1.

    • Search Google Scholar
    • Export Citation
  • Lebsock, M. D., , and L’Ecuyer T. S. , 2011: The retrieval of warm rain from CloudSat. J. Geophys. Res., 116, D20209, doi:10.1029/2011JD016076.

    • Search Google Scholar
    • Export Citation
  • Liao, L., , and Meneghini R. , 2009: Changes in the TRMM version-5 and version-7 precipitation radar products due to orbit boost. J. Meteor. Soc. Japan, 87A, 93107, doi:10.2151/jmsj.87A.93.

    • Search Google Scholar
    • Export Citation
  • Liu, C., , and Zipser E. J. , 2009: “Warm rain” in the tropics: Seasonal and regional distribution based on 9 yr of TRMM data. J. Climate, 22, 767779, doi:10.1175/2008JCLI2641.1.

    • Search Google Scholar
    • Export Citation
  • Liu, C., , and Zipser E. J. , 2013: Regional variation of morphology of the organized convection in the tropics and subtropics. J. Geophys. Res. Atmos., 118, 453466, doi:10.1029/2012JD018409.

    • Search Google Scholar
    • Export Citation
  • Liu, C., , Zipser E. J. , , Cecil D. J. , , Nesbitt S. W. , , and Sherwood S. , 2008: A cloud and precipitation feature database from nine years of TRMM observations. J. Appl. Meteor. Climatol., 47, 27122728, doi:10.1175/2008JAMC1890.1.

    • Search Google Scholar
    • Export Citation
  • Malkus, J. S., 1954: Some results of a trade-cumulus cloud investigation. J. Meteor., 11, 220237, doi:10.1175/1520-0469(1954)011<0220:SROATC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Marshall, J. S., , and Palmer W. Mc K. , 1948: The distribution of raindrops with size. J. Meteor., 5, 165166, doi:10.1175/1520-0469(1948)005<0165:TDORWS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • McCollum, J. R., , Gruber A. , , and Ba M. B. , 2000: Discrepancy between gauges and satellite estimates of rainfall in equatorial Africa. J. Appl. Meteor., 39, 666679, doi:10.1175/1520-0450-39.5.666.

    • Search Google Scholar
    • Export Citation
  • Nesbitt, S. W., , Zipser E. J. , , and Kummerow C. D. , 2004: An examination of version 5 rainfall estimates from the TRMM microwave imager, precipitation radar, and rain gauges on global, regional, and storm scales. J. Appl. Meteor., 43, 10161036, doi:10.1175/1520-0450(2004)043<1016:AEOVRE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Petty, G. W., , and Li K. , 2013: Improved passive microwave retrievals of rain rate over land and ocean. Part I: Algorithm description. J. Atmos. Oceanic Technol.,30, 2493–2508, doi:10.1175/JTECH-D-12-00144.1.

  • Riehl, H., , and Malkus J. S. , 1958: On the heat balance in the equatorial trough zone. Geophysica, 6, 503538.

  • Seo, E., , Sohn B. , , and Liu G. , 2007: How TRMM precipitation radar and microwave imager retrieved rain rates differ. Geophys. Res. Lett., 34, L24803, doi:10.1029/2007GL032331.

    • Search Google Scholar
    • Export Citation
  • Shige, S., , Sasaki H. , , Okamoto K. , , and Iguchi T. , 2006: Validation of rainfall estimates from the TRMM precipitation radar and microwave imager using a radiative transfer model: 1. Comparison of the version-5 and -6 products. Geophys. Res. Lett., 33, L13803, doi:10.1029/2006GL026350.

    • Search Google Scholar
    • Export Citation
  • Shige, S., , Kida S. , , Ashiwake H. , , Kubota T. , , and Aonashi K. , 2013: Improvement of TMI rain retrievals in mountainous areas. J. Appl. Meteor. Climatol., 52, 242254, doi:10.1175/JAMC-D-12-074.1.

    • Search Google Scholar
    • Export Citation
  • Short, D. A., , and North G. , 1990: The beam filling error in the Nimbus 5 electronically scanning microwave radiometer observations of Global Atlantic Tropical Experiment rainfall. J. Geophys. Res., 95, 21872193, doi:10.1029/JD095iD03p02187.

    • Search Google Scholar
    • Export Citation
  • Short, D. A., , and Nakamura K. , 2010: Effect of TRMM orbit boost on radar reflectivity distributions. J. Atmos. Oceanic Technol., 27, 12471254, doi:10.1175/2010JTECHA1426.1.

    • Search Google Scholar
    • Export Citation
  • Spencer, R. W., , Goodman H. M. , , and Hood R. E. , 1989: Precipitation retrieval over land and ocean with SSM/I: Identification and characteristics of the scattering signal. J. Atmos. Oceanic Technol., 6, 254273, doi:10.1175/1520-0426(1989)006<0254:PROLAO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M., , Houze R. A. Jr., , and Yuter S. E. , 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor., 34, 19782007, doi:10.1175/1520-0450(1995)034<1978:CCOTDS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tustison, B. D., , Harris D. , , and Foufoula-Georgiou E. , 2001: Scale issues in verification of precipitation forecasts. J. Geophys. Res., 106, 11 77511 784, doi:10.1029/2001JD900066.

    • Search Google Scholar
    • Export Citation
  • Vivekanandan, J., , Turk J. , , and Bringi V. N. , 1991: Ice water path estimation and characterization using passive microwave radiometry. J. Appl. Meteor., 30, 14071421, doi:10.1175/1520-0450(1991)030<1407:IWPEAC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Wall, C., , Liu C. , , and Zipser E. , 2013: A climatology of tropical congestus using CloudSat. J. Geophys. Res. Atmos., 118, 64786492, doi:10.1002/jgrd.50455.

    • Search Google Scholar
    • Export Citation
  • Wang, N.-Y., , Liu C. , , Ferraro R. , , Wolff D. , , Zipser E. J. , , and Kummerow C. , 2009: The TRMM 2A12 land precipitation product—Status and future plans. J. Meteor. Soc. Japan, 87A, 237253, doi:10.2151/jmsj.87A.237.

    • Search Google Scholar
    • Export Citation
  • Wexler, R., , and Swingle D. M. , 1947: Radar storm detection. Bull. Amer. Meteor. Soc., 28, 159167.

  • Wilheit, T. T., 1986: Some comments on passive microwave measurement of rain. Bull. Amer. Meteor. Soc., 67, 12261232, doi:10.1175/1520-0477(1986)067<1226:SCOPMM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Wilheit, T. T., , and Kummerow C. D. , 2009: Use of the TRMM-PR for estimating the TMI beam filling correction. J. Meteor. Soc. Japan, 87A, 255263, doi:10.2151/jmsj.87A.255.

    • Search Google Scholar
    • Export Citation
  • Xu, W., , Zipser E. J. , , and Liu C. , 2009: Rainfall characteristics and convective properties of mei-yu precipitation systems over south China, Taiwan, and the South China Sea. Part I: TRMM observations. Mon. Wea. Rev., 137, 42614275, doi:10.1175/2009MWR2982.1.

    • Search Google Scholar
    • Export Citation
  • Yokoyama, C., , Zipser E. J. , , and Liu C. , 2014: TRMM-observed shallow versus deep convection in the eastern Pacific related to large-scale circulations in reanalysis datasets. J. Climate, 27, 55755592, doi:10.1175/JCLI-D-13-00315.1.

    • Search Google Scholar
    • Export Citation
  • Zhang, C., , McGauley M. , , and Bond N. A. , 2004: Shallow meridional circulation in the tropical eastern Pacific. J. Climate, 17, 133139, doi:10.1175/1520-0442(2004)017<0133:SMCITT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zhang, J., and et al. , 2011: National Mosaic and Multisensor QPE (NMQ) System: Description, results, and future plans. Bull. Amer. Meteor. Soc., 92, 13211338, doi:10.1175/2011BAMS-D-11-00047.1.

    • Search Google Scholar
    • Export Citation
  • Zipser, E. J., , Liu C. , , Cecil D. J. , , Nesbitt S. W. , , and Yorty D. P. , 2006: Where are the most intense thunderstorms on Earth? Bull. Amer. Meteor. Soc., 87, 10571071, doi:10.1175/BAMS-87-8-1057.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Global distributions of the unconditional mean precipitation from version 7 TRMM (a) PR (3A25) and (b) TMI (3A12) product observations during 1998–2012. Their differences are demonstrated with the (c) subtraction [(a) − (b)] and (d) fraction [(a)/(b)]. The contours of 1000 mm yr−1 mean precipitation are shown in (c) and (d). Four regions of interest with high precipitation rate and large difference between PR and TMI are shown with dashed boxes in (d).

  • View in gallery

    (a) Histogram of PR and TMI instantaneous rain rates at nadir pixels over land and ocean in 36°S–36°N. (b) Fractional rainfall contribution from different instantaneous rain rates at nadir pixels.

  • View in gallery

    (a) Global distributions of the number of nadir pixels with the precipitation indicated by PR but with no precipitation by TMI in 1° × 1° boxes. Note that there are more samples near 32° latitudes due to the sample biases of TRMM. (b) As in (a), but for nonzero precipitation detected by TMI and no precipitation by PR; (c) global distribution of fraction of nonzero PR precipitation nadir pixels with zero TMI precipitation in 1° × 1° boxes. The sample biases are removed with the fraction of PR precipitation nadir pixels with zero TMI. (d) As in (c), but for the fraction of nonzero precipitation TMI nadir pixels with zero PR precipitation. The contours of 1000 mm yr−1 mean precipitation are shown in (c) and (d).

  • View in gallery

    (a) Two-dimensional histogram of differences of PR and TMI rain rates and the PR echo-top height at stratiform nadir pixels (with both RPR and RTMI >1 mm h−1) over ocean. (b) As in (a), but for pixels over land. (c) Two-dimensional histogram of differences of PR and TMI rain rates and the PR echo-top height for convective nadir pixels over ocean. (d) As in (c), but for pixels over land.

  • View in gallery

    Examples of the RTPFs over (a) Oklahoma and (b) Yemen. Black contour and cross lines define the RTPF with an area of either PR or TMI rain within PR swath. Color fill shows the PR near-surface rain rate. Red dashed contour shows the area with TMI rain. Black and white background is TRMM Visible and Infrared Scanner (VIRS) infrared brightness temperature, respectively. Black dots are the location of lightning flashes detected by TRMM lightning imaging sensor. Case (a) over Oklahoma has a large raining area by TMI for which PR shows no rain and case (b) over Yemen has large raining area by PR for which TMI shows no rain.

  • View in gallery

    (a) Geographical distribution of fraction of small RTPFs (<200 km2) with area of TMI rain (ATMI) equal to zero in 1° × 1° boxes. (b) As in (a), but for small RTPFs with zero PR rain area (APR). (c) Fraction of large RTPFs (>2000 km2) that have ratio of areas of PR and TMI rain that are within 2 times of each other. (d) Fraction of large RTPFs with 4 times larger PR rain area. (e) Fraction of large RTPFs with 4 times larger TMI rain area. Contours are the number of RTPF samples in each 1° × 1° boxes.

  • View in gallery

    (a) Two-dimensional histogram of area and ratio of PR and TMI volumetric rainfall of RTPFs over ocean. (b) As in (a), but for RTPFs over land. Dashed lines are for the ratios of 0.5 and 2.

  • View in gallery

    (a) Geographical distribution of the fractions of small RTPFs (area <200 km2) with at least 2 times PR volumetric rainfall (VRPR) than TMI volumetric rainfall (VRTMI) in 1° × 1° boxes. (b) As in (a), but for small RTPFs with 2 times larger TMI volumetric rainfall. (c) Geographical distribution of the fractions of large RTPFs (area > 2000 km2) with at least 2 times PR volumetric rainfall (VRPR) than TMI volumetric rainfall (VRTMI) in 1° × 1° boxes. (d) As in (c), but for small RTPFs with 2 times larger TMI volumetric rainfall. Contours are number of RTPF samples with both PR and TMI volumetric rainfall >0 in each 1° × 1° boxes.

  • View in gallery

    (a) Two-dimensional histogram of maximum echo-top heights and differences between PR and TMI volumetric rainfall of RTPFs over ocean. (b) As in (a), but for over land.

  • View in gallery

    (a) Geographical distribution of the fractions of shallow RTPFs (maximum echo-top height <6 km) with at least 2 times PR volumetric rainfall (VRPR) than TMI volumetric rainfall (VRTMI) in 1° × 1° boxes. (b) As in (a), but for shallow RTPFs with 2 times larger TMI volumetric rainfall. (c) Geographical distribution of the fractions of deep RTPFs (maximum echo-top height >8 km) with at least 2 times the PR volumetric rainfall (VRPR) than TMI volumetric rainfall (VRTMI) in 1° × 1° boxes. (d) As in (c), but for deep RTPFs with 2 times larger TMI volumetric rainfall.

  • View in gallery

    (a) Two-dimensional cumulative histogram of minimum 85-GHz PCT and maximum PR near-surface rain rate of RTPFs over the Amazon (color fill) and Congo (contour). (b) Two-dimensional cumulative histogram of maximum 30-dBZ top height and maximum echo-top height of RTPFs over the Amazon (color fill) and Congo (contour). (c) Two-dimensional cumulative histogram of PR near-surface rain rate and 85-GHz PCT at nadir pixels over the Amazon (color fill) and Congo (contour). (d) Two-dimensional cumulative histogram of PR echo-top height and 30-dBZ top height at nadir pixels over the Amazon (color fill) and Congo (contour). Note that the maximum values of 30-dBZ height near the surface represent the RTPFs not having any radar echoes ≥30 dBZ in (b) and (d).

  • View in gallery

    As in Fig. 11, but for over the east Pacific (color fill) and north Indian Ocean (contour).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 89 89 21
PDF Downloads 64 64 4

Differences between the Surface Precipitation Estimates from the TRMM Precipitation Radar and Passive Microwave Radiometer Version 7 Products

View More View Less
  • 1 Department of Physical and Environmental Sciences, Texas A&M University Corpus Christi, Corpus Christi, Texas
  • | 2 Department of Atmospheric Sciences, University of Utah, Salt Lake City, Utah
© Get Permissions
Full access

Abstract

With 15 yr of the Tropical Rainfall Measuring Mission (TRMM) observations, the passive microwave radiometers [TRMM Microwave Imager (TMI)] and the precipitation radar (PR) report a close geographical distribution of annual precipitation between 36°S and 36°N. However, large discrepancies between PR and TMI precipitation retrievals are also found over several specific regions, such as central Africa, the Amazon, the tropical east Pacific, and north Indian Ocean. To understand these discrepancies, the PR near-surface and the TMI surface precipitation retrievals are compared at both pixel and precipitation system levels using collocated pixels and a precipitation feature database from 1998 to 2012. Over land, the TMI overestimates precipitation in deep and intense convective systems, but misses significant amounts of warm rainfall in shallow systems. Over the ocean, because of the partial beam filling of large footprints of the lower-frequency sensors, the TMI reports a larger precipitation area than the PR and underestimates the precipitation rate in the convective precipitation region. The TMI tends to overestimate precipitation compared to the PR in a large proportion of shallow systems over the tropical east Pacific and trade wind regions with large-scale descent. The PR tends to overestimate precipitation compared to the TMI in a large proportion of shallow systems over rainy oceans, such as the west Pacific and the Atlantic ITCZ. All these findings imply that there are still large uncertainties in the precipitation climatology over some regions. Further ground validation campaigns are still needed, especially over the ocean.

Corresponding author address: Dr. Chuntao Liu, Department of Physical and Environmental Sciences, Texas A&M University Corpus Christi, 6300 Ocean Dr., Corpus Christi, TX 78412-5892. E-mail: chuntao.liu@tamucc.edu

Abstract

With 15 yr of the Tropical Rainfall Measuring Mission (TRMM) observations, the passive microwave radiometers [TRMM Microwave Imager (TMI)] and the precipitation radar (PR) report a close geographical distribution of annual precipitation between 36°S and 36°N. However, large discrepancies between PR and TMI precipitation retrievals are also found over several specific regions, such as central Africa, the Amazon, the tropical east Pacific, and north Indian Ocean. To understand these discrepancies, the PR near-surface and the TMI surface precipitation retrievals are compared at both pixel and precipitation system levels using collocated pixels and a precipitation feature database from 1998 to 2012. Over land, the TMI overestimates precipitation in deep and intense convective systems, but misses significant amounts of warm rainfall in shallow systems. Over the ocean, because of the partial beam filling of large footprints of the lower-frequency sensors, the TMI reports a larger precipitation area than the PR and underestimates the precipitation rate in the convective precipitation region. The TMI tends to overestimate precipitation compared to the PR in a large proportion of shallow systems over the tropical east Pacific and trade wind regions with large-scale descent. The PR tends to overestimate precipitation compared to the TMI in a large proportion of shallow systems over rainy oceans, such as the west Pacific and the Atlantic ITCZ. All these findings imply that there are still large uncertainties in the precipitation climatology over some regions. Further ground validation campaigns are still needed, especially over the ocean.

Corresponding author address: Dr. Chuntao Liu, Department of Physical and Environmental Sciences, Texas A&M University Corpus Christi, 6300 Ocean Dr., Corpus Christi, TX 78412-5892. E-mail: chuntao.liu@tamucc.edu

1. Introduction

As a main component of the global water cycle, precipitation is not only a major source of freshwater over land, but latent heat release during precipitation processes is one main engine driving the global atmospheric circulation (Riehl and Malkus 1958). There have been many efforts to measure or estimate the surface precipitation globally, including surface rain gauges and remote sensing techniques. Surface rain gauges provide a relatively accurate measurement of local precipitation amount. However, these measurements are only available over regions easy to access. Precipitation measurements over high mountains, deserts, and forests are very difficult to obtain by rain gauges. Over ocean, only sparse observations are available from the gauges on buoys (e.g., Bowman 2005) due to the difficulties of access and the cost. Therefore, remote sensing techniques have been widely used to provide a large areal coverage of precipitation estimates.

The first attempt of remote sensing of precipitation is from the ground-based precipitation radars (Wexler and Swingle 1947). Radar reflectivity is sensitive to the large hydrometeors and near-surface radar echo return can be used to estimate the rainfall rate with some simplified empirical ZR relationships (Marshall and Palmer 1948; Atlas 1990). Surface precipitation can be estimated with a good spatial and temporal coverage with ground radar networks, such as National Mosaic and Multi-Sensor Quantitative Precipitation Estimation (QPE) system (Zhang et al. 2011). However, ground-based radars require regions easy to access and operate. Radar networks certainly cannot provide global coverage, especially over large areas of oceans, mountains, and deserts. Global surface precipitation estimation became possible only in the satellite era. Based on the cold infrared brightness temperature from global satellite images, with the assumption of relationships between the cold cloud tops and surface precipitation, the first global precipitation distribution was created using a simple global precipitation index (Arkin et al. 1994). Further, passive microwave radiance measurements from satellite were found to have better description of near-surface rainfall rate than the infrared, especially over ocean (Wilheit 1986).

In December 1997, the Tropical Rainfall Measuring Mission (TRMM; Kummerow et al. 1998) was launched with both precipitation radar (PR) and the passive TRMM Microwave Imager (TMI) onboard to provide a global precipitation estimate over the tropics and subtropics. With more than 16 yr of observations up to today, the PR and TMI provide a robust climatology of the global precipitation independently (Iguchi et al. 2000; Kummerow et al. 2001; Huffman et al. 2001, 2007). As shown in Fig. 1, the global precipitation estimates from the TRMM PR (Fig. 1a) and TMI (Fig. 1b) show similar geographical distributions. The mean zonal precipitation rates from PR and TMI estimates are close (Table 1). However, it is also clear that there are still large differences between the PR and TMI precipitation estimates over some regions (Figs. 1c,d). For example, over land, the PR derived ~20% more precipitation over central Africa than TMI, but ~20% less over the Amazon and the Maritime Continent. Over ocean, the PR derived 20% more precipitation than the TMI over the north Indian Ocean, but 20% less over the tropical east Pacific. These differences lead to the motivations for this study:

  • Why are there large differences in the precipitation estimates from radar and passive microwave observations over different regions?
  • Would these differences vary in precipitation systems with different properties? If so, how?
  • How can we improve precipitation retrievals based on what we learn from these differences?
Though these questions have been addressed in various ways in the past (e.g., Nesbitt et al. 2004; Shige et al. 2006; Seo et al. 2007; Wang et al. 2009), there is still lack of a thorough assessment of the precipitation rate differences between PR and TMI retrievals, especially for the new version 7 (V7) TRMM products (Iguchi et al. 2009; Awaka et al. 2009; Kummerow et al. 2011). In this study, first we compare the PR and TMI precipitation retrievals at the collocated PR nadir pixels. Then we compare the PR and TMI total rain volume within precipitation systems and examine the relationships between the PR versus TMI differences and the properties of precipitation systems. This paper is organized as follows. Section 2 describes the data and methodology used in this study. Section 3 compares the PR and TMI retrievals at the PR nadir pixels. Section 4 examines the PR versus TMI differences in precipitation features. The conclusions and discussion are presented in section 5.
Fig. 1.
Fig. 1.

Global distributions of the unconditional mean precipitation from version 7 TRMM (a) PR (3A25) and (b) TMI (3A12) product observations during 1998–2012. Their differences are demonstrated with the (c) subtraction [(a) − (b)] and (d) fraction [(a)/(b)]. The contours of 1000 mm yr−1 mean precipitation are shown in (c) and (d). Four regions of interest with high precipitation rate and large difference between PR and TMI are shown with dashed boxes in (d).

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

Table 1.

Mean precipitation rate derived from TRMM PR (3A25) and TMI (3A12) over tropics and subtropics during 1998–2012.

Table 1.

2. Data and methodology

Since the launch in December 1997, the first spaceborne precipitation radar, the TRMM PR, has been obtaining measurements of radar reflectivity with a global coverage between 36°S and 36°N. It has a horizontal resolution of about 4.2 km, vertical resolution of 250 m, and swath width of 220 km. To extend the duration of the mission, the orbit of the TRMM satellite was boosted from 350 to 402 km in August 2001. With this satellite boost, the footprint and sensitivity of the radar have some slight changes (Short and Nakamura 2010). In general, the consistency of the PR radar reflectivity measurements has been validated well with a research-quality ground-based radar (Liao and Meneghini 2009). The PR has been working normally since launch until a failure of a circuit board in June 2009. However, the backup device was used to continue the operation and the radar has been working fine since then. We are not aware of a report on the quality change of the radar reflectivity during this transition. However, to make sure that there is no big impact from the sensitivity changes of the radar on the results shown in this paper, we have repeated all analyses for two time periods, one from 1998 to 2012 and the other from 2002 to 2008. The results are very similar by using these two time periods. We choose to only present the results from 1998 to 2012 here.

The version 7 2A25 TRMM PR near-surface rainfall product is used in this study. The differences between version 6 (V6) and version 7 PR products are discussed in Iguchi et al. (2009), Awaka et al. (2009), and Funk and Schumacher (2013). In general, the near-surface rainfall rate is derived from PR reflectivity at the lowest range bin that is considered without contamination of the ground clutter. This varies from about 0.5 km over the ocean and 1–2 km over land. The details of the 2A25 TRMM PR rainfall rate retrieval algorithm are described by Iguchi et al. (2000, 2009). The products have the lowest precipitation rate of 0.1 mm h−1. However, a PR reliable signal is ~17 dBZ (before the satellite boost), which translates to about 0.4–0.5 mm h−1. Many weak rainfall rates between 0.1 and 0.4 mm h−1 can be missed or marginally detected. Based on the radar reflectivity values and horizontal (Steiner et al. 1995) and vertical gradients, convective and stratiform rainfall regions are separated with the 2A23 algorithm described by the Awaka et al. (1998, 2009). Considering the different microphysical properties of hydrometeors over the convective and stratiform precipitation regions, different groups of reflectivity to rain rate (ZR) relationships have been implemented in the 2A25 algorithm (Iguchi et al. 2000, 2009). Therefore, the comparisons of the retrievals are conducted over convective and stratiform regions.

TMI measures the microwave radiances at nine channels with horizontal and vertical polarization at 10, 19, 37, and 85 GHz with various footprint sizes from 63 × 37 to 5 × 7 km2 (Kummerow et al. 1998). The microwave radiances at low-frequency channels (10, 19, and 37 GHz) are sensitive to the emission from liquid hydrometeors at low levels. The microwave radiances at high frequencies (37 and 85 GHz) are depressed with scattering of ice hydrometeors that may be used to estimate the amount of ice in the vertical column (Vivekanandan et al. 1991). Because the ocean surface has a low microwave emissivity at lower frequencies, the emission signal from hydrometeors can be easily separated from the surface background. Therefore, it has been used to infer the surface rain rate over the ocean (Wilheit 1986). Kummerow et al. (2001) described the TRMM 2A12 algorithm to retrieve the surface precipitation rate over water surfaces by utilizing the radiances at multiple microwave frequencies with a Bayesian approach. Over land, however, the microwave emissivity varies significantly with surface temperature, soil moisture, and type (e.g., Ferraro et al. 2012). The emission signal from hydrometeors is difficult to separate from that from the ground. Therefore, simple empirical relationships have been derived to estimate the surface precipitation rate from the ice-scattering signals in microwave brightness temperatures at high-frequency channels (Ferraro 1997; Gopalan et al. 2010), with an assumption that the amount of ice aloft would eventually fall to the ground and that most liquid precipitation over land comes from melted ice hydrometeors. The TRMM V7 2A12 rain-rate retrieval algorithm from TMI consists of two separate algorithms over land (Gopalan et al. 2010) and over water surfaces (Kummerow et al. 2011). The improvements of the V7 versus V6 products are discussed in detail by Gopalan et al. (2010) and Kummerow et al. (2011). Over land, the empirical relationships between the ice-scattering signal and the surface precipitation are built upon collocated version 7 TMI brightness temperatures and PR near-surface rain rate (Gopalan et al. 2010). Over ocean, the TMI precipitation retrieval algorithm is a Bayesian approach built upon a database of precipitation rates at collocated PR and TMI pixels. The new concept of the probability of precipitation is introduced based on the fraction of the PR rainy pixels with similar TMI brightness temperatures in the database (Kummerow et al. 2011). The fraction of TMI pixels with PR rain (probability) decreases when the TMI precipitation rate is below 1 mm h−1 (figure not shown). These weak precipitation rates are only marginally detected by PR, especially below 0.4 mm h−1. Though sometimes PR does not detect rain with the similar TMI brightness temperatures, these weak precipitation rates are still probably legitimate and physically detectable by TMI with emission signal at low frequencies. In this study, we do not specifically utilize the probability of precipitation in the analysis. However, we have to keep in mind that the classification of weak precipitation rates over ocean is still an open question due to the different sensitivity of PR and TMI. In this study, the version 7 TMI 2A12 surface precipitation rates are collocated and compared with the PR 2A25 near-surface rain rates over land and ocean separately.

To demonstrate the differences between the TRMM PR and TMI retrievals, two different types of comparisons are conducted. The first is the pixel-by-pixel comparisons at nadir. The benefit of this type of comparison is the straightforward differences of the retrieval results at the same time and location from two independent measurements. First, we collocate the PR and TMI pixels by using the nearest neighbor method. The one difficulty of the collocation is the different scanning geometries by PR and TMI. PR scans nadir, but TMI scans conically. This could lead to mismatch between PR and TMI columns when there are significant amounts of ice aloft at high altitudes. Therefore, a parallax correction method has been introduced by shifting the TMI pixels forward/backward to match pixels when deep convection presents (Liu et al. 2008). However, this parallax correction would lead to the mismatch for pixels with shallow precipitation. In this study, a new parallax correction method is developed by shifting the PR pixel forward/backward only when the PR echo-top height exceeds 6 km and the path-integrated attenuation is greater than 0.4 dBZ. This method keeps good matches for pixels with shallow precipitation and improves the collocation for pixels with deep convection. After collocating the pixels, the properties of PR and TMI retrievals, including PR 2A25 near-surface rain rate (Iguchi et al. 2009), 2A12 surface precipitation rate (Kummerow et al. 2011), PR 2A23 echo-top height and rain type (Awaka et al. 2009), TMI polarization–corrected brightness temperatures (PCT; Spencer et al. 1989), and so on, are summarized for each pixel. To avoid the complication of the scanning geometry, only the collocated retrievals at the PR nadir pixels from 1998 to 2012 are analyzed in this study. Also we require at least 0.1 mm h−1 rain rate for each pixel to be eligible in the comparisons.

There is one caveat for intercomparing the rain rates at collocated PR and TMI pixels, the different uniform beam filling due to different footprint area of PR and TMI sensors. As described in Kummerow et al. (1998), the PR pixel area is about 2 times smaller than TMI 85-GHz pixels and about 120 times smaller than TMI 10-GHz pixels. Over land, TMI rain rates are retrieved based on the 85-GHz brightness temperatures (Kummerow et al. 2001; Gopalan et al. 2010). Over ocean, the Bayesian approach relies on most of the TMI channels with the footprint sizes ranging from 2 to 120 times of PR pixels. Rain rates are derived at much larger footprints than PR with a strong nonuniform beam-filling effect. There have been many studies addressing the issue of comparing precipitation observations with different spatial resolutions (e.g., Short and North 1990; Tustison et al. 2001; Wilheit and Kummerow 2009). In general, this leads to a more “smeared” precipitation system from TMI with relatively larger rain areas and weaker rain rates compared to the PR. This is especially true for shallow rain over the ocean where the TMI retrievals come mainly from the low-frequency, large area footprints. Therefore, a large TMI raining area is expected without PR rain over ocean. Also the high values of rain rates are averaged to lower values and this leads to a shift of the rain-rate histogram. Some of these effects were discussed by Tustison et al. (2001).

The second type of comparison is conducted at the precipitation system level. With a methodology similar to what is described in Nesbitt et al. (2004) and Liu et al. (2008), the precipitation radar and TMI precipitation features (RTPFs) are defined by grouping the contiguous, collocated pixels with precipitation indicated by either PR 2A25 near-surface rain or TMI surface precipitation greater than 0.1 mm h−1 (regardless of precipitation probability) product. The properties inside these precipitation features are calculated, including the total volumetric rain (sum of the instantaneous rain rate times rain area; this definition of volumetric rain does not include any information about the vertical dimension of the raining area), maximum echo-top heights, and so on. Then we compare the total rain volumes from PR 2A25 and TMI 2A12 products inside each RTPF. One benefit of this type of comparison is that the beam-filling and parallax correction is no longer a concern when we compare PR and TMI total rain volume at the precipitation system level. This approach may help us understand the relationships between the retrieval differences and the properties of precipitation systems. In this study, the RTPFs with at least four contiguous pixels are selected during 1998–2012. Because the different precipitation retrieval algorithms are used over land and ocean from TMI observations, we discuss the differences between PR and TMI precipitation estimates separately over land and ocean in each analysis below.

3. PR and TMI precipitation retrieval difference at nadir pixels

Over ocean, regardless of the probability of precipitation from TMI, there are nearly 72 million pixels found with at least 0.1 mm h−1 precipitation rate at the PR nadir pixels from either PR 2A25 near-surface rain or TMI 2A12 surface precipitation products from 15 yr of TRMM data. Over ocean, pixels tend to have more heavy rain from PR than TMI (Fig. 2a). There are many more pixels with weak rain rates by TMI. In all, TMI has the histogram of rain rate shifted to the weaker side compared to PR (Fig. 2a). Also TMI suggests a smaller rainfall contribution from heavy rainfall rates compared to PR (Fig. 2b). This is largely due to the beam-filling effect that extreme, high, rain values are averaged to lower values over a large area. Note that TMI has a larger contribution from the precipitation rate below 1.0 mm h−1 than PR. This is probably physical due to the sensitivity of TMI at weaker precipitation rates over the ocean; PR may miss weak rain <0.4 mm h−1 due to the minimum detectable signal of about 17 dBZ. For example, a large area of weak precipitation, such as drizzle from stratocumulus, is below the sensitivity of PR, but the total microwave radiance emission from the small hydrometeors could be well detected by TMI low-frequency channels with large footprint sizes. However, quantitative validation is still needed.

Fig. 2.
Fig. 2.

(a) Histogram of PR and TMI instantaneous rain rates at nadir pixels over land and ocean in 36°S–36°N. (b) Fractional rainfall contribution from different instantaneous rain rates at nadir pixels.

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

Over land, there are nearly 8 million pixels found with at least 0.1 mm h−1 precipitation rate at the PR nadir pixels from either PR 2A25 near-surface rain or TMI 2A12 surface precipitation products from 15 yr of TRMM data. A majority of the pixels have low rain rates. A larger proportion of pixels have TMI rain rates between 1 and 10 mm h−1 than PR rain rates over land (Fig. 2a). Thus, TMI indicates a larger rainfall contribution from the rain rate 1–10 km over land (Fig. 2b). However, there are more extreme rain rates >20 mm h−1 from the PR, and they contribute a considerable fraction of the total rainfall. Next, we categorize the differences between TMI and PR retrievals at pixel levels into rain area differences and bias of rain rates and discuss them separately.

a. Rain area differences

Rain area (detection) differences between PR and TMI retrievals are from two scenarios: TMI detects rain but PR does not and vice versa. To have a quantitative sense of area with both PR and TMI rain, and the fraction of rain volume from both the PR and TMI raining region, we summarized the PR nadir pixels with PR rain, TMI rain, and both over the tropics and subtropics separately. Then the fractions from the overlapped TMI and PR rain area to PR and TMI detected total rain area and fractions from the rain volume within the overlapped area to the total rain volume by PR and TMI are calculated. These have been examined with pixels with >0.1 and >1 mm h−1. The results are listed in Table 2.

Table 2.

Fractions of areas and precipitation from collocated PR and TMI nadir pixels with rain rates >0.1 and >1 mm h−1 over tropics and subtropics during 1998–2012.

Table 2.

In Table 2, only a small proportion (25%) of TMI rain area is overlapped by PR rain over tropical oceans. This is largely due to the beam-filling effect from large footprints of TMI low-frequency channels. This could also be from weak precipitation that is below the sensitivity of PR. The majority (92%) of PR raining pixels over tropical oceans are identified with precipitation by TMI. The PR suggests that pixels with >1 mm h−1 contribute a majority (>90%) of total precipitation over both land and ocean. However, TMI pixels >1 mm h−1 only contribute 73% over ocean. Over tropical oceans, TMI reports 27% of total precipitation from weak rain (<1 mm h−1), compared to 6% by PR. Also the pixel size mismatch between PR and TMI could be one major cause of the differences here. Since TMI retrieval over ocean relies on multiple channels with various footprint sizes, there is really no good match between the TMI and PR retrievals at the pixel levels. Therefore, the pixel level comparisons between the TMI and PR retrievals over ocean should be cautiously interpreted.

Over tropical land, about ⅔ of PR rain pixels (>0.1 mm h−1) are identified with TMI rain (>0.1 mm h−1) and vice versa (Table 2). Large proportions of the total precipitation (77% of PR, 79% of TMI) are from the pixels with both PR and TMI rain (>0.1 mm h−1). The fraction of precipitation from overlapping rain areas of PR and TMI is smaller over subtropical land (69% of PR, 59% of TMI), partially due to the difficulty for TMI to detect light wintertime precipitation and precipitation over high mountains.

Figure 3 shows the geographical distribution of population and fraction of PR-only and TMI-only raining pixels. Over ocean, more PR-only raining pixels are found in the subtropics than tropics. TMI-only raining pixels occur more frequently over regions with large annual precipitation (Fig. 3b). A large fraction of precipitation is from TMI-only raining pixels over the regions with large-scale descent, such as the southeast Pacific, Atlantic, and Indian Oceans off the coasts of the continents (Fig. 3d). It is known that stratocumulus and some shallow weak precipitation is below the sensitivity of the PR over these regions (Berg et al. 2010; Lebsock and L’Ecuyer 2011). It is interesting that PR-only raining pixels also contribute a notable fraction of total precipitation over these regions (Fig. 3c). Some shallow precipitation systems detected by PR are also missed by TMI, possibly due to the small size of the systems not resolved by large TMI footprints. Therefore, there is still a large uncertainty in the rain detection over this region.

Fig. 3.
Fig. 3.

(a) Global distributions of the number of nadir pixels with the precipitation indicated by PR but with no precipitation by TMI in 1° × 1° boxes. Note that there are more samples near 32° latitudes due to the sample biases of TRMM. (b) As in (a), but for nonzero precipitation detected by TMI and no precipitation by PR; (c) global distribution of fraction of nonzero PR precipitation nadir pixels with zero TMI precipitation in 1° × 1° boxes. The sample biases are removed with the fraction of PR precipitation nadir pixels with zero TMI. (d) As in (c), but for the fraction of nonzero precipitation TMI nadir pixels with zero PR precipitation. The contours of 1000 mm yr−1 mean precipitation are shown in (c) and (d).

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

Over land, PR-only raining pixels are mainly over the Amazon, the Maritime Continent, and some coastal regions where shallow warm rainfall occurs (Fig. 3a) (Liu and Zipser 2009). Because the TMI algorithm over land relies on the ice-scattering signal at 85 GHz, warm rainfall without ice-phase hydrometeors cannot be detected. In fact, more than 80% of warm rainfall with low echo-top heights (<4.5 km) from PR are missed by TMI over both tropics and subtropics. In the deeper systems, with PR echo-top heights >7 km, most of the PR rain pixels (80%–90%) are also detected by TMI over tropical land. About 30%–40% of PR rain pixels with echo top >5 km are missed by TMI over subtropical land and even more in the winter (figure not shown). Because of the low emissivity of microwave radiance at 85 GHz over desert and cold high mountains, the screening of the TMI land algorithm (Ferraro et al. 1998) removes a majority of the precipitation over these regions (Fig. 3c). Most of the pixels (>80%) with 85-GHz PCT colder than 250 K are detected with rain by PR. However, the screening is not perfect. There are still a large number of TMI-only pixels over cold mountains that are likely artifacts (Figs. 3b,d). Because the ice particles in very thick anvil clouds associated with deep convection may lead to a strong scattering signal at 85 GHz, many TMI-only pixels are from these scenarios, such as over central Africa and the Amazon (Fig. 3b).

b. Rain-rate differences

To investigate the rain-rate differences when both PR and TMI detect precipitation, the PR nadir pixels with at least 1 mm h−1 rain rate from both PR and TMI are selected because these pixels contribute the majority of the total rainfall (Table 2). But much weak precipitation in stratiform regions, and isolated warm rainfall over ocean, are excluded by this selection. For the rain rates greater than 1 mm h−1, there is relatively less uncertainty from the detection sensitivity, and the Bayesian probability of precipitation is close to 100%.

To understand the PR versus TMI rain-rate biases, the properties of the precipitation in the whole column are examined. Figure 4 demonstrates the variation of PR versus TMI rain-rate differences against radar echo-top height over the convective and stratiform region indicated by PR. Over ocean, though there is a large spread of PR versus TMI differences, the PR and TMI have generally consistent rain rates in the stratiform regions (Fig. 4a). Stratiform regions have larger uniform raining areas, so the beam-filling effect is small and PR and TMI are more consistent. However, over convective regions, there are many more pixels with high PR rain rates than TMI (Fig. 4c). This is because the beam-filling effect averages the high rain rates in convective regions. The fractions of pixels with consistent PR versus TMI rain rates (difference <1 mm h−1) are relatively higher in the subtropics than in the tropics (figure not shown). This is probably because subtropical systems have larger, more uniform raining areas where the influence of beam filling is smaller.

Fig. 4.
Fig. 4.

(a) Two-dimensional histogram of differences of PR and TMI rain rates and the PR echo-top height at stratiform nadir pixels (with both RPR and RTMI >1 mm h−1) over ocean. (b) As in (a), but for pixels over land. (c) Two-dimensional histogram of differences of PR and TMI rain rates and the PR echo-top height for convective nadir pixels over ocean. (d) As in (c), but for pixels over land.

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

Over land, TMI rain rates tend to be larger than PR rain rates when radar echo top is higher (e.g., >12 km), whether in convective or stratiform regions (Figs. 4b,d). This is due to the large ice water path leading to excessive ice-scattering signal that would overestimate the surface precipitation by TMI for deep convection. Many pixels with much higher TMI rain rates than PR are also found over desert regions, for example, the Sahel (figure not shown). This is likely due to the low emissivity at 85 GHz of those surfaces that have large brightness temperature depressions, leading to overestimation of rain rate by the TMI land algorithm. The V7 TMI land algorithm attempts to use different empirical relationships for convective and stratiform regions. Large footprint size and different convective and stratiform separation based only on the TMI brightness temperatures could create a mismatch between PR and TMI rain rates in convective and stratiform regions. That also explains some of the differences shown in Figs. 4b and 4d.

Though we have demonstrated the differences between PR and TMI rain rates at the PR nadir pixels and related them to the radar echo top and convective and stratiform regions and so on, it is still difficult to clearly interpret these differences due to the different footprint sizes and imperfect collocation between the PR and TMI pixels. Therefore, we use a different approach to avoid these effects by using precipitation features.

4. PR and TMI rain-rate differences in precipitation features

Instead of comparing the rain rates pixel by pixel, we examine total rain volumes from the PR and TMI in precipitation systems. The RTPFs are defined as areas with either PR or TMI contiguous raining pixels within the PR swath. As examples shown in Fig. 5, there are mesoscale convective systems (MCSs) for which TMI detects large areas of rainfall for which PR detects no rain reaching the surface (Fig. 5a), and those with PR raining areas missed entirely by the TMI (Fig. 5b). The supercell over Oklahoma had a large ice water content aloft in the anvil region with a strong ice-scattering signal at 85 GHz (Fig. 5a). However, these ice particles had not yet fallen to near the surface (at overpass time) so that the PR showed no detectable reflectivity at low levels. Therefore, the TMI land algorithm derived a large area of rainfall based on the ice-scattering signal in the anvil region, whereas the PR reported no rain. The second case (Fig. 8b) over Yemen had large areas of PR rain where TMI did not report rain due to the difficulty of rain screening over the desert region. These are just two extreme cases to demonstrate how we define RTPFs. The main benefit of using RTPFs is that when comparing the total rain volumes from PR and TMI in RTPFs, the beam filling and collocation are no longer a big concern, and at the same time we can relate the variation of the PR versus TMI retrieval biases with different precipitation system types.

Fig. 5.
Fig. 5.

Examples of the RTPFs over (a) Oklahoma and (b) Yemen. Black contour and cross lines define the RTPF with an area of either PR or TMI rain within PR swath. Color fill shows the PR near-surface rain rate. Red dashed contour shows the area with TMI rain. Black and white background is TRMM Visible and Infrared Scanner (VIRS) infrared brightness temperature, respectively. Black dots are the location of lightning flashes detected by TRMM lightning imaging sensor. Case (a) over Oklahoma has a large raining area by TMI for which PR shows no rain and case (b) over Yemen has large raining area by PR for which TMI shows no rain.

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

a. Rain area differences in precipitation features

With RTPFs, the first test is on the detection differences between PR and TMI. Figure 6 shows the geographical distribution of fraction of RTPFs with five different PR versus TMI rain area scenarios: small (<200 km2) RTPFs with rain from PR only and TMI only and large (>2000 km2) RTPFs with dominant rain area by PR, TMI, or both. Small RTPFs without TMI rain are mainly found over land (Fig. 6a). Large fractions of RTPFs over deserts do not have TMI rain due to the screening. It is worth noting that over most rainy land areas, TMI misses a majority of small precipitation systems (Fig. 6a), either due to the small size of the system, or the lack of ice scattering at 85 GHz in warm rainfall (e.g., the Amazon and the west coast of India).

Fig. 6.
Fig. 6.

(a) Geographical distribution of fraction of small RTPFs (<200 km2) with area of TMI rain (ATMI) equal to zero in 1° × 1° boxes. (b) As in (a), but for small RTPFs with zero PR rain area (APR). (c) Fraction of large RTPFs (>2000 km2) that have ratio of areas of PR and TMI rain that are within 2 times of each other. (d) Fraction of large RTPFs with 4 times larger PR rain area. (e) Fraction of large RTPFs with 4 times larger TMI rain area. Contours are the number of RTPF samples in each 1° × 1° boxes.

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

Small RTPFs without PR rainfall are mainly found over ocean, and many occur in the regions of large-scale descent such as the trade wind environment (Malkus 1954) (Fig. 6b). About half of the small RTPFs in the rainy regions of the ITCZ and SPCZ do not have PR rainfall. It is worth noting that a higher fraction of small RTPFs are found without PR rainfall over the tropical east Pacific. This might be related to the high liquid water path over the region that confuses the algorithm by misclassifying the cloud as precipitation (Berg et al. 2006). Nevertheless, there are still some difficulties to make PR and TMI consistent over this region.

For the precipitation systems with large sizes (RTPFs > 2000 km2), PR and TMI raining area are within a factor of 2 over land (Fig. 6c). However, TMI tends to underestimate the raining area in large MCSs over the Yangtze River region of China (Figs. 6c,d), where mei-yu fronts bring most of the rainfall during May–July (Xu et al. 2009) and a large proportion of total precipitation involves warm rain that is not detected by TMI. Because of the difficulty of detecting rainfall over the desert and cold mountain regions, TMI has false alarms (Fig. 6b) and misses the majority of rainfall over the Sahel, Saudi Arabia, Australia, Tibetan Plateau, and Rocky Mountains (Fig. 6d). Note that the high fraction of RTPFs without PR rainfall over northwest Australia is an artifact due to the agreement to turn off the PR there.

For large RTPFs, the TMI raining area is much larger than PR over ocean, especially over the large-scale descent regions in subtropics (Fig. 6e). This could be from both beaming filling and weak precipitation beyond the sensitivity of PR. It is worth noting that a higher fraction (80%–90%) of MCSs over the tropical east Pacific have TMI raining areas 4 times larger than the PR raining area, compared with those over tropical west Pacific (50%–70%) and Atlantic (60%–70%) (Fig. 6e). These differences may not be fully explained by the beam-filling effect.

b. Rain volume differences in precipitation features with different sizes

Though there are differences in the sensitivity and footprints of sensors, the ideal result for both TMI and PR would be to provide consistent total precipitation estimates for the same precipitation system in general, regardless of the beam-filling effect. To investigate the PR and TMI rain volume differences within the same precipitation systems, the total rain volume differences between PR (VRPR) and TMI (VRTMI) are examined in RTPFs with different sizes in two-dimensional histograms (Fig. 7). Then the geographical distribution of fractions of the small and large RTPFs with large PR or TMI rain volume differences are shown in Fig. 8.

Fig. 7.
Fig. 7.

(a) Two-dimensional histogram of area and ratio of PR and TMI volumetric rainfall of RTPFs over ocean. (b) As in (a), but for RTPFs over land. Dashed lines are for the ratios of 0.5 and 2.

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

Fig. 8.
Fig. 8.

(a) Geographical distribution of the fractions of small RTPFs (area <200 km2) with at least 2 times PR volumetric rainfall (VRPR) than TMI volumetric rainfall (VRTMI) in 1° × 1° boxes. (b) As in (a), but for small RTPFs with 2 times larger TMI volumetric rainfall. (c) Geographical distribution of the fractions of large RTPFs (area > 2000 km2) with at least 2 times PR volumetric rainfall (VRPR) than TMI volumetric rainfall (VRTMI) in 1° × 1° boxes. (d) As in (c), but for small RTPFs with 2 times larger TMI volumetric rainfall. Contours are number of RTPF samples with both PR and TMI volumetric rainfall >0 in each 1° × 1° boxes.

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

Over ocean, Most of the small RTPFs (size < 500 km2) have much higher PR rain volume than TMI when both PR and TMI indicate rain (Fig. 7a). The locations of small RTPFs with twice as much of rain volume from PR than TMI are spread over all oceans, especially over regions with many trade cumuli (Fig. 8a). One speculation is that because the TMI algorithm over the ocean is built upon the matched TMI and PR pixels after averaging the PR rain rate onto TMI footprints that this artificial beam filling has been built into the database. However, when matching the database to brightness temperatures from isolated showers, a lower rain rate is likely retrieved. Another possible reason is that we only used pixels with at least >0.1 mm h−1; a portion of rain from very weak rain retrievals from TMI are excluded by this method. In Fig. 7a, many oceanic precipitation features with size range of 1000–10 000 km2 have much larger TMI rain volumes than PR (Fig. 7a). They are more likely over the stratocumulus regions (Fig. 8d). TMI probably captures the large areas of weak rainfall over this region that are missed by PR due to the sensitivity. It is interesting to note that over the east Pacific, there is a large proportion (30%) of large RTPFs with much higher TMI rain volume than PR. It is known that the east Pacific ITCZ has a shallow meridional circulation (Zhang et al. 2004) and occasionally forms shallow convective lines (Liu and Zipser 2013). TMI tends to overestimate the rainfall over this region, possibly due to the high total column water vapor and low aerosol concentration over the region (Berg et al. 2006).

Over land, small RTPFs have much higher PR rain volume (Fig. 7b). Large proportions of small RTPFs with much higher PR rain volume are found at the northeast coast of South America, east coast of Africa, and Maritime Continent (Fig. 8a). Small RTPFs with much higher TMI rain volume occur more frequently over the Sahel, Argentina, and east coast of central China (Fig. 8b), in addition to the artifacts over cold mountains and desert regions. Though most of the large features have more consistent PR and TMI volumetric rainfall ratios (Fig. 7b), features with inconsistent PR and TMI rain volumes do occur over some specific regions. For example, there are more large RTPFs with twice as much PR rain volume than TMI over southeast China, western Ghats, and over desert regions (Fig. 8c), likely due to the missing warm rainfall and screening difficulty over deserts by TMI. Large RTPFs with large TMI rain volume occur more frequently over central Africa, which is consistent with the higher TMI precipitation estimates there (Fig. 1d).

c. Rain volume differences in precipitation features with different depths

As pointed out by earlier studies (Nesbitt et al. 2004; Seo et al. 2007), PR versus TMI rain-rate differences vary with the convective intensity of precipitation systems. A similar test is conducted with the ratio and subtraction of PR and TMI rain volume against the maximum PR echo-top height in Fig. 9. Then the geographical distribution of the fraction of shallow and deep RTPFs with large PR and TMI rain volume differences is shown in Fig. 10.

Fig. 9.
Fig. 9.

(a) Two-dimensional histogram of maximum echo-top heights and differences between PR and TMI volumetric rainfall of RTPFs over ocean. (b) As in (a), but for over land.

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

Fig. 10.
Fig. 10.

(a) Geographical distribution of the fractions of shallow RTPFs (maximum echo-top height <6 km) with at least 2 times PR volumetric rainfall (VRPR) than TMI volumetric rainfall (VRTMI) in 1° × 1° boxes. (b) As in (a), but for shallow RTPFs with 2 times larger TMI volumetric rainfall. (c) Geographical distribution of the fractions of deep RTPFs (maximum echo-top height >8 km) with at least 2 times the PR volumetric rainfall (VRPR) than TMI volumetric rainfall (VRTMI) in 1° × 1° boxes. (d) As in (c), but for deep RTPFs with 2 times larger TMI volumetric rainfall.

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

Over the ocean, there are two groups of shallow systems with different types of PR versus TMI rain volume differences (Fig. 9a). One group has a PR echo top in the range of 3–6 km with larger PR rain volume than TMI. These shallow RTPFs with more PR rain volume are over the ITCZ and SPCZ regions, especially over west Pacific and northern Indian Ocean (Fig. 10a). Many of these shallow systems are corresponding to isolated showers as shown in Fig. 8a. Another group of shallow systems has PR echo top below 3 km with more TMI rain volume than PR (Fig. 9a). The shallow oceanic RTPFs with more TMI rain volume are over regions populated with stratocumulus, in addition to the region south of the east Pacific ITCZ (Fig. 10b). Most deep RTPFs over ocean have consistent PR and TMI rain volumes (Fig. 9a). Over the east Pacific and large-scale descent regions, however, there are many deep RTPFs having much higher TMI rain volume (Fig. 10d).

Over land, the PR rain volume is much larger than that of the TMI in systems with echo top below 8 km (Fig. 9b). Shallow RTPFs with higher PR rain volumes occur more frequently near the coast, such as eastern South America, eastern Africa, and southeast China (Fig. 10a). These shallow systems have a large proportion of warm rainfall in a moist environment (Liu and Zipser 2009). There are many shallow RTPFs with higher TMI rain volume over high mountains (Fig. 10b). They are very likely due to the artifacts of TMI over cold surfaces. There is a large spread of PR versus TMI rain volume for deep systems over land (Fig. 9b). For deep systems with echo top above 8 km, RTPFs with larger PR rain volume are mostly over the desert regions (Fig. 10c), where screening removes most of rainfall by TMI. They also occur more frequently over southeast China and Bangladesh (Fig. 10c), where the large proportion of warm rainfall embedded in the deep systems could be missed by TMI over these regions. There are more deep systems with echo top reaching 12 km having larger TMI rain volume than PR over land (Fig. 9b). Many of these deep and intense systems are accompanied by large areas of thick anvil clouds that would be identified as rainfall by TMI. Though these ice particles may appear with PR echo at high altitudes, if PR detects no echo at near surface, PR still reports no rain. A good example of this is in Fig. 5a. Many of these RTPFs are mostly stratiform precipitation with large areas of thick anvil region with cold 85-GHz PCT that lead to a larger TMI rain volume.

d. PR versus TMI differences over specific regions

1) The Amazon versus central Africa

Though the rainfall differences over different land regions have been shown above, we now address the question of why there are opposite differences of PR versus TMI rain amounts over central Africa and the Amazon. McCollum et al. (2000) suggested that passive microwave sensor retrievals could overestimate the surface precipitation because of the high evaporation rate in precipitation systems with high cloud base or smaller hydrometeors with large aerosol loading in the relatively dry environment over central Africa. It is known that storms over Africa are more intense than those over the Amazon with a higher flash rate, 40-dBZ echo-top height, and colder 85 and 37 PCTs (Zipser et al. 2006). Comparing the Amazon with central Africa (Congo), RTPFs over Africa have much colder 85-GHz brightness temperatures for the similar PR near-surface rain rate (Figs. 11a,c). With the same echo-top height, African storms have higher 30-dBZ echo top when echo top is greater than 6 km (Figs. 11b,d). This means that large ice particles are lifted to higher altitudes in African precipitation systems, which would lead to stronger ice scattering at 85 GHz than in Amazon systems. African storms tend to have more large ice particles aloft and have relatively lower 85-GHz PCT (Figs. 11a,b). Thus, the TMI land algorithm overestimates the African surface rainfall according to these stronger brightness temperature depressions than the Amazon ones. Over the Amazon, there are large amounts of warm rainfall (Liu and Zipser 2009) missed by TMI, as well as large numbers of congestus (Wall et al. 2013) that may not have sufficient ice scattering over a large enough depth for TMI to infer heavy enough rain rates. Therefore, PR reports much higher rain volume than TMI there.

Fig. 11.
Fig. 11.

(a) Two-dimensional cumulative histogram of minimum 85-GHz PCT and maximum PR near-surface rain rate of RTPFs over the Amazon (color fill) and Congo (contour). (b) Two-dimensional cumulative histogram of maximum 30-dBZ top height and maximum echo-top height of RTPFs over the Amazon (color fill) and Congo (contour). (c) Two-dimensional cumulative histogram of PR near-surface rain rate and 85-GHz PCT at nadir pixels over the Amazon (color fill) and Congo (contour). (d) Two-dimensional cumulative histogram of PR echo-top height and 30-dBZ top height at nadir pixels over the Amazon (color fill) and Congo (contour). Note that the maximum values of 30-dBZ height near the surface represent the RTPFs not having any radar echoes ≥30 dBZ in (b) and (d).

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

2) Tropical east Pacific versus northern Indian Ocean

The differences between the properties of precipitation systems over the tropical east Pacific and northern Indian Ocean are shown in Fig. 12. Compared with those over the northern Indian Ocean, there are many more shallow RTPFs over the east Pacific with a minimum 85-GHz PCT warmer than 250 K (Fig. 12a). A lot of these shallow RTPFs over the east Pacific have small sizes, and PR detects no rain in them at all (Fig. 6b). Systems over the north Indian Ocean are relatively deeper with lower 85-GHz PCT (Figs. 12a,c). There is not much difference of the convective intensity in deep systems over these two regions (Figs. 12b,d). In addition to the missing shallow precipitation by PR over the east Pacific, it is likely that the moist environment over the east Pacific leads to higher rain-rate retrievals by TMI over the region than relatively dryer environment over the north Indian Ocean. However, these are pure speculations and further study is needed to fully understand the large difference between the PR and TMI rain rates over these two regions.

Fig. 12.
Fig. 12.

As in Fig. 11, but for over the east Pacific (color fill) and north Indian Ocean (contour).

Citation: Journal of Hydrometeorology 15, 6; 10.1175/JHM-D-14-0051.1

5. Summary and discussion

We have described the differences between the PR near-surface and TMI surface precipitation estimates from version 7 TRMM product at both the pixel and system levels. In general, there are still large differences between PR and TMI rain retrievals over various regions. Over the ocean, TMI tends to detect larger areas of rainfall with a shifted histogram of rain rates due to the partial beam filling of large footprint sizes of sensors. There are a large number of shallow and small oceanic precipitation systems detected by TMI, but not detected by PR over the regions with large-scale descent and over the tropical east Pacific. TMI tends to underestimate the rain volume in the shallow systems detected by PR over both land and ocean, but overestimate the rain volume in the deep intense systems over land. Most of these differences are consistent with conclusions of early study on the version 6 TRMM product (Nesbitt et al. 2004, their Table 2), such as beam filling, radiometrically cold surface, and convective intensity. However, the impact of the melting-layer emission on the TMI retrieval pointed out by Nesbitt et al. (2004) is not obvious in the results shown for the version 7 product. There is no obvious overestimation of TMI rain rates in the stratiform region (Fig. 4a). With the large database built from PR versus TMI matched rain profiles for version 7 2A12 algorithm, the inadequate consideration of melting-layer emission from model outputs is no longer an issue here. Note that we did not discuss the details of seasonal changes of the PR versus TMI retrieval differences. Both algorithms perform quite differently in subtropical winter systems (figure not shown).

There are a few lessons learned based on the results shown. Because of the “beam-filling” effect by different footprint sizes, the histograms of the rain rates from PR and TMI are shifted. Therefore, it is difficult to directly assess the different rain rates from PR and TMI at the pixel level. Though some detailed EOF analysis of rain-rate differences at pixel level could provide some hint in the biases (e.g., Seo et al. 2007), these approaches would mostly pick up the beam-filling effect (e.g., convective pixels have lower TMI rain rates). Comparing the PR versus TMI rain volumes at the system level provides a clearer assessment of the physical reasons behind the biases. Regardless of the beam-filling effect, there are a large number of small, shallow, and weak rainfall systems only detected by TMI over ocean. Are these precipitation systems real or false alarms? Further validation is needed. One potential approach could be using a collocated CloudSat dataset to validate these light precipitation systems.

This study found that there is a large discrepancy between PR and TMI retrievals over the east Pacific. A large amount of shallow rainfall over the region is detected by TMI but missed by PR. TMI also overestimates the rainfall in the deep systems over this region. The mechanisms of shallow circulation (Zhang et al. 2004; Yokoyama et al. 2014) and shallow convective lines (Liu and Zipser 2013) over this region still need further understanding. Therefore, further study and surface validation over this region is needed.

Over land, the main problem of the current TMI V7 2A12 product is the detection of warm rainfall and rainfall over desert and any radiometrically cold surfaces. To improve the algorithm over such surfaces, the information from low-frequency channels has to be incorporated after carefully taking into account highly variable surface emissivity over different surface types. Many efforts (e.g., Petty and Li 2013; Ferraro et al. 2012; Shige et al. 2013) are already pursuing this direction. It could be valuable to use precipitation features to validate these new algorithms focusing on warm rain, as well as precipitation systems over desert and cold mountain regions.

This study shows that precipitation features are useful in assessing the differences between PR and TMI precipitation retrievals. There are apparent relationships between the rainfall biases and the storm types over various regions. The main reason is that different types of precipitation systems have different microphysical properties, which are reflected by the different sensitivities to the hydrometeors at different altitudes by the TMI and PR algorithms. Currently, the retrieval algorithms still treat pixels independently. There is important information in the precipitation systems to improve the retrieval algorithms. However, how to utilize this information to improve the algorithms is still a challenge.

Acknowledgments

We appreciate Dr. Chris Kummerow’s contributions for the suggestions of the manuscript structure and the details of the 2A12 algorithms. This research was supported by NASA Precipitation Measurement Mission Grants NNX13AF73G and NNX11AG31G under the direction of Dr. Ramesh Kakar and NASA Grant NNX08AK28G under the direction of Dr. Erich Stocker. Thanks also go to Drs. Erich Stocker and John Kwiatkowski and the rest of the Precipitation Processing System (PPS) team at NASA Goddard Space Flight Center, Greenbelt, Maryland, for data processing assistance.

REFERENCES

  • Arkin, P. A., , Joyce R. J. , , and Janowiak J. E. , 1994: IR techniques: GOES precipitation index. Remote Sens. Rev., 11, 107124, doi:10.1080/02757259409532261.

    • Search Google Scholar
    • Export Citation
  • Atlas, D., 1990: Radar in Meteorology: Battan Memorial and 40th Anniversary Radar Meteorology Conference. Amer. Meteor. Soc., 806 pp.

  • Awaka, J., , Iguchi T. , , and Okamoto K. , 1998: Early results on rain type classification by the Tropical Rainfall Measuring Mission (TRMM) precipitation radar. Proc. 8th URSI Commission F Triennial Open Symp. on Wave Propagation and Remote Sensing, Aveiro, Portugal, URSI, 143146.

  • Awaka, J., , Iguchi T. , , and Okamoto K. , 2009: TRMM PR standard algorithm 2A23 and its performance on bright band detection. J. Meteor. Soc. Japan, 87A, 3152, doi:10.2151/jmsj.87A.31.

    • Search Google Scholar
    • Export Citation
  • Berg, W., , L’Ecuyer T. , , and Kummerow C. , 2006: Rainfall climate regimes: The relationship of regional TRMM rainfall biases to the environment. J. Appl. Meteor. Climatol., 45, 434454, doi:10.1175/JAM2331.1.

    • Search Google Scholar
    • Export Citation
  • Berg, W., , L’Ecuyer T. , , and Haynes J. M. , 2010: The distribution of rainfall over oceans from spaceborne radars. J. Appl. Meteor. Climatol., 49, 535543, doi:10.1175/2009JAMC2330.1.

    • Search Google Scholar
    • Export Citation
  • Bowman, K. P., 2005: Comparison of TRMM precipitation retrievals with rain gauge data from ocean buoys. J. Climate, 18, 178190, doi:10.1175/JCLI3259.1.

    • Search Google Scholar
    • Export Citation
  • Ferraro, R. R., 1997: Special Sensor Microwave Imager derived global rainfall estimates for climatological applications. J. Geophys. Res., 102, 16 71516 735, doi:10.1029/97JD01210.

    • Search Google Scholar
    • Export Citation
  • Ferraro, R. R., , Smith E. A. , , Berg W. , , and Huffman G. J. , 1998: A screening methodology for passive microwave precipitation retrieval algorithms. J. Atmos. Sci., 55, 15831600, doi:10.1175/1520-0469(1998)055<1583:ASMFPM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Ferraro, R. R., and et al. , 2012: An evaluation of microwave land surface emissivities over the continental United Stated to benefit GPM-era precipitation algorithms. IEEE Trans. Geosci. Remote Sens.,51, 378–398, doi:10.1109/TGRS.2012.2199121.

  • Funk, A., , and Schumacher C. , 2013: Analysis of rain classification over tropics by version 7 of TRMM 2A23 algorithm. J. Meteor. Soc. Japan, 91, 257272, doi:10.2151/jmsj.2013-302.

    • Search Google Scholar
    • Export Citation
  • Gopalan, K., , Wang N.-Y. , , Ferraro R. , , and Liu C. , 2010: Status of the TRMM 2A12 land precipitation algorithm. J. Atmos. Oceanic Technol., 27, 13431354, doi:10.1175/2010JTECHA1454.1.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., , Adler R. F. , , Morrissey M. M. , , Bolvin D. T. , , Cuttis S. , , Joyce R. , , McGavock B. , , and Susskind J. , 2001: Global precipitation at one degree resolution from multisatellite observations. J. Hydrometeor., 2, 3650, doi:10.1175/1525-7541(2001)002<0036:GPAODD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., and et al. , 2007: The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeor., 8, 3855, doi:10.1175/JHM560.1.

    • Search Google Scholar
    • Export Citation
  • Iguchi, T., , Kozu T. , , Meneghini R. , , Awaka J. , , and Okamoto K. , 2000: Rain-profiling algorithm for the TRMM precipitation radar. J. Appl. Meteor., 39, 20382052, doi:10.1175/1520-0450(2001)040<2038:RPAFTT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Iguchi, T., , Kozu T. , , Kwiatkowski J. , , Meneghini R. , , Awaka J. , , and Okamoto K. , 2009: Uncertainties in the rain profiling algorithm for the TRMM precipitation radar. J. Meteor. Soc. Japan,87, 1–30, doi:10.2151/jmsj.87A.1.

  • Kummerow, C., , Barnes W. , , Kozu T. , , Shiue J. , , and Simpson J. , 1998: The Tropical Rainfall Measuring Mission (TRMM) sensor package. J. Atmos. Oceanic Technol., 15, 809817, doi:10.1175/1520-0426(1998)015<0809:TTRMMT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kummerow, C., and et al. , 2001: The evolution of the Goddard profiling algorithm (GPROF) for rainfall estimation from passive microwave sensors. J. Appl. Meteor., 40, 18011820, doi:10.1175/1520-0450(2001)040<1801:TEOTGP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kummerow, C., , Ringerud S. , , Crook J. , , Randel D. , , and Berg W. , 2011: An observationally generated a priori database for microwave rainfall retrievals. J. Atmos. Oceanic Technol., 28, 113130, doi:10.1175/2010JTECHA1468.1.

    • Search Google Scholar
    • Export Citation
  • Lebsock, M. D., , and L’Ecuyer T. S. , 2011: The retrieval of warm rain from CloudSat. J. Geophys. Res., 116, D20209, doi:10.1029/2011JD016076.

    • Search Google Scholar
    • Export Citation
  • Liao, L., , and Meneghini R. , 2009: Changes in the TRMM version-5 and version-7 precipitation radar products due to orbit boost. J. Meteor. Soc. Japan, 87A, 93107, doi:10.2151/jmsj.87A.93.

    • Search Google Scholar
    • Export Citation
  • Liu, C., , and Zipser E. J. , 2009: “Warm rain” in the tropics: Seasonal and regional distribution based on 9 yr of TRMM data. J. Climate, 22, 767779, doi:10.1175/2008JCLI2641.1.

    • Search Google Scholar
    • Export Citation
  • Liu, C., , and Zipser E. J. , 2013: Regional variation of morphology of the organized convection in the tropics and subtropics. J. Geophys. Res. Atmos., 118, 453466, doi:10.1029/2012JD018409.

    • Search Google Scholar
    • Export Citation
  • Liu, C., , Zipser E. J. , , Cecil D. J. , , Nesbitt S. W. , , and Sherwood S. , 2008: A cloud and precipitation feature database from nine years of TRMM observations. J. Appl. Meteor. Climatol., 47, 27122728, doi:10.1175/2008JAMC1890.1.

    • Search Google Scholar
    • Export Citation
  • Malkus, J. S., 1954: Some results of a trade-cumulus cloud investigation. J. Meteor., 11, 220237, doi:10.1175/1520-0469(1954)011<0220:SROATC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Marshall, J. S., , and Palmer W. Mc K. , 1948: The distribution of raindrops with size. J. Meteor., 5, 165166, doi:10.1175/1520-0469(1948)005<0165:TDORWS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • McCollum, J. R., , Gruber A. , , and Ba M. B. , 2000: Discrepancy between gauges and satellite estimates of rainfall in equatorial Africa. J. Appl. Meteor., 39, 666679, doi:10.1175/1520-0450-39.5.666.

    • Search Google Scholar
    • Export Citation
  • Nesbitt, S. W., , Zipser E. J. , , and Kummerow C. D. , 2004: An examination of version 5 rainfall estimates from the TRMM microwave imager, precipitation radar, and rain gauges on global, regional, and storm scales. J. Appl. Meteor., 43, 10161036, doi:10.1175/1520-0450(2004)043<1016:AEOVRE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Petty, G. W., , and Li K. , 2013: Improved passive microwave retrievals of rain rate over land and ocean. Part I: Algorithm description. J. Atmos. Oceanic Technol.,30, 2493–2508, doi:10.1175/JTECH-D-12-00144.1.

  • Riehl, H., , and Malkus J. S. , 1958: On the heat balance in the equatorial trough zone. Geophysica, 6, 503538.

  • Seo, E., , Sohn B. , , and Liu G. , 2007: How TRMM precipitation radar and microwave imager retrieved rain rates differ. Geophys. Res. Lett., 34, L24803, doi:10.1029/2007GL032331.

    • Search Google Scholar
    • Export Citation
  • Shige, S., , Sasaki H. , , Okamoto K. , , and Iguchi T. , 2006: Validation of rainfall estimates from the TRMM precipitation radar and microwave imager using a radiative transfer model: 1. Comparison of the version-5 and -6 products. Geophys. Res. Lett., 33, L13803, doi:10.1029/2006GL026350.

    • Search Google Scholar
    • Export Citation
  • Shige, S., , Kida S. , , Ashiwake H. , , Kubota T. , , and Aonashi K. , 2013: Improvement of TMI rain retrievals in mountainous areas. J. Appl. Meteor. Climatol., 52, 242254, doi:10.1175/JAMC-D-12-074.1.

    • Search Google Scholar
    • Export Citation
  • Short, D. A., , and North G. , 1990: The beam filling error in the Nimbus 5 electronically scanning microwave radiometer observations of Global Atlantic Tropical Experiment rainfall. J. Geophys. Res., 95, 21872193, doi:10.1029/JD095iD03p02187.

    • Search Google Scholar
    • Export Citation
  • Short, D. A., , and Nakamura K. , 2010: Effect of TRMM orbit boost on radar reflectivity distributions. J. Atmos. Oceanic Technol., 27, 12471254, doi:10.1175/2010JTECHA1426.1.

    • Search Google Scholar
    • Export Citation
  • Spencer, R. W., , Goodman H. M. , , and Hood R. E. , 1989: Precipitation retrieval over land and ocean with SSM/I: Identification and characteristics of the scattering signal. J. Atmos. Oceanic Technol., 6, 254273, doi:10.1175/1520-0426(1989)006<0254:PROLAO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M., , Houze R. A. Jr., , and Yuter S. E. , 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor., 34, 19782007, doi:10.1175/1520-0450(1995)034<1978:CCOTDS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tustison, B. D., , Harris D. , , and Foufoula-Georgiou E. , 2001: Scale issues in verification of precipitation forecasts. J. Geophys. Res., 106, 11 77511 784, doi:10.1029/2001JD900066.

    • Search Google Scholar
    • Export Citation
  • Vivekanandan, J., , Turk J. , , and Bringi V. N. , 1991: Ice water path estimation and characterization using passive microwave radiometry. J. Appl. Meteor., 30, 14071421, doi:10.1175/1520-0450(1991)030<1407:IWPEAC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Wall, C., , Liu C. , , and Zipser E. , 2013: A climatology of tropical congestus using CloudSat. J. Geophys. Res. Atmos., 118, 64786492, doi:10.1002/jgrd.50455.

    • Search Google Scholar
    • Export Citation
  • Wang, N.-Y., , Liu C. , , Ferraro R. , , Wolff D. , , Zipser E. J. , , and Kummerow C. , 2009: The TRMM 2A12 land precipitation product—Status and future plans. J. Meteor. Soc. Japan, 87A, 237253, doi:10.2151/jmsj.87A.237.

    • Search Google Scholar
    • Export Citation
  • Wexler, R., , and Swingle D. M. , 1947: Radar storm detection. Bull. Amer. Meteor. Soc., 28, 159167.

  • Wilheit, T. T., 1986: Some comments on passive microwave measurement of rain. Bull. Amer. Meteor. Soc., 67, 12261232, doi:10.1175/1520-0477(1986)067<1226:SCOPMM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Wilheit, T. T., , and Kummerow C. D. , 2009: Use of the TRMM-PR for estimating the TMI beam filling correction. J. Meteor. Soc. Japan, 87A, 255263, doi:10.2151/jmsj.87A.255.

    • Search Google Scholar
    • Export Citation
  • Xu, W., , Zipser E. J. , , and Liu C. , 2009: Rainfall characteristics and convective properties of mei-yu precipitation systems over south China, Taiwan, and the South China Sea. Part I: TRMM observations. Mon. Wea. Rev., 137, 42614275, doi:10.1175/2009MWR2982.1.

    • Search Google Scholar
    • Export Citation
  • Yokoyama, C., , Zipser E. J. , , and Liu C. , 2014: TRMM-observed shallow versus deep convection in the eastern Pacific related to large-scale circulations in reanalysis datasets. J. Climate, 27, 55755592, doi:10.1175/JCLI-D-13-00315.1.

    • Search Google Scholar
    • Export Citation
  • Zhang, C., , McGauley M. , , and Bond N. A. , 2004: Shallow meridional circulation in the tropical eastern Pacific. J. Climate, 17, 133139, doi:10.1175/1520-0442(2004)017<0133:SMCITT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zhang, J., and et al. , 2011: National Mosaic and Multisensor QPE (NMQ) System: Description, results, and future plans. Bull. Amer. Meteor. Soc., 92, 13211338, doi:10.1175/2011BAMS-D-11-00047.1.

    • Search Google Scholar
    • Export Citation
  • Zipser, E. J., , Liu C. , , Cecil D. J. , , Nesbitt S. W. , , and Yorty D. P. , 2006: Where are the most intense thunderstorms on Earth? Bull. Amer. Meteor. Soc., 87, 10571071, doi:10.1175/BAMS-87-8-1057.

    • Search Google Scholar
    • Export Citation
Save