• Aonashi, K., and Coauthors, 2009: GSMaP passive, microwave precipitation retrieval algorithm: Algorithm description and validation. J. Meteor. Soc. Japan, 87A, 119136, doi:10.2151/jmsj.87A.119.

    • Search Google Scholar
    • Export Citation
  • Behrangi, A., , Hsu K.-L. , , Imam B. , , Sorooshian S. , , Huffman G. J. , , and Kuligowski R. J. , 2009: PERSIANN-MSA: A precipitation estimation method from satellite-based multispectral analysis. J. Hydrometeor., 10, 14141429, doi:10.1175/2009JHM1139.1.

    • Search Google Scholar
    • Export Citation
  • Garstang, M., , and Kummerow C. D. , 2000: The Joanne Simpson special issue on the Tropical Rainfall Measuring Mission (TRMM). J. Appl. Meteor., 39, 19611961, doi:10.1175/1520-0450(2001)040<1961:TJSSIO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Hou, A. Y., and Coauthors, 2014: The Global Precipitation Measurement Mission. Bull. Amer. Meteor. Soc., 95, 701722, doi:10.1175/BAMS-D-13-00164.1.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., 1997: Estimates of root-mean-square random error for finite samples of estimated precipitation. J. Appl. Meteor., 36, 11911201, doi:10.1175/1520-0450(1997)036<1191:EORMSR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., , and Bolvin D. T. , 2014: TRMM and other data precipitation data set documentation. TRMM Doc., 42 pp. [Available online at ftp://meso-a.gsfc.nasa.gov/pub/trmmdocs/3B42_3B43_doc.pdf.]

  • Huffman, G. J., and Coauthors, 2007: The TRMM Multi-satellite Precipitation Analysis: Quasi-global, multi-year, combined-sensor precipitation estimates at fine scale. J. Hydrometeor., 8, 3855, doi:10.1175/JHM560.1.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., , Adler R. F. , , Bolvin D. T. , , and Nelkin E. J. , 2010: The TRMM Multi-Satellite Precipitation Analysis (TMPA). Satellite Rainfall Applications for Surface Hydrology, F. Hossain and M. Gebremichael, Eds., Springer-Verlag, 3–22.

  • Huffman, G. J., , Bolvin D. T. , , Braithwaite D. , , Hsu K. , , Joyce R. , , Kidd C. , , Nelkin E. J. , , and Xie P. , 2015a: NASA Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG). Algorithm Theoretical Basis Doc., version 4.5, 26 pp. [Available online at http://pmm.nasa.gov/sites/default/files/document_files/IMERG_ATBD_V4.5.pdf.]

  • Huffman, G. J., , Bolvin D. T. , , and Nelkin E. J. , 2015b: Integrated Multi-satellitE Retrievals for GPM (IMERG) technical documentation. NASA Doc., 47 pp. [Available online at http://pmm.nasa.gov/sites/default/files/document_files/IMERG_doc.pdf.]

  • Huffman, G. J., , Bolvin D. T. , , and Nelkin E. J. , 2015c: Day 1 IMERG final run release notes. NASA Doc., 9 pp. [Available online at http://pmm.nasa.gov/sites/default/files/document_files/IMERG_FinalRun_Day1_release_notes.pdf.]

  • Janowiak, J. E., , Joyce R. J. , , and Yarosh Y. , 2001: A real-time global half-hourly pixel-resolution infrared dataset and its applications. Bull. Amer. Meteor. Soc., 82, 205217, doi:10.1175/1520-0477(2001)082<0205:ARTGHH>2.3.CO;2.

    • Search Google Scholar
    • Export Citation
  • Joyce, R. J., , Janowiak J. E. , , Arkin P. A. , , and Xie P. , 2004: CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hydrometeor., 5, 487503, doi:10.1175/1525-7541(2004)005<0487:CAMTPG>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kummerow, C., , Randel D. , , Kirstetter P. , , Wang N.-Y. , , Petkovic V. , , Kulie M. , , and Ferraro R. , 2014a: Global Precipitation Measurement (GPM) mission. Algorithm Theoretical Basis Doc., version 1.4, 46 pp. [Available online at http://rain.atmos.colostate.edu/ATBD/ATBD_GPM_Aug1_2014.pdf.]

  • Kummerow, C., , Randel D. , , Kirstetter P. , , Wang N.-Y. , , Petkovic V. , , Kulie M. , , and Ferraro R. , 2014b: Initial GPROF GPM results using a GPM derived database. Seventh IPWG Workshop on Precipitation Measurements, Tsukuba, Japan, International Precipitation Working Group. [Available online at http://www.isac.cnr.it/~ipwg/meetings/tsukuba-2014/pres/4-2_Kummerow.pdf.]

  • Liu, Z., 2015a: Comparison of precipitation estimates between version 7 3-hourly TRMM Multi-Satellite Precipitation Analysis (TMPA) near-real-time and research products. Atmos. Res., 153, 119133, doi:10.1016/j.atmosres.2014.07.032.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., 2015b: Comparison of versions 6 and 7 3-Hourly TRMM Multi-Satellite Precipitation Analysis (TMPA) research products. Atmos. Res., 163, 91101, doi:10.1016/j.atmosres.2014.12.015.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., , Ostrenga D. , , Teng W. , , and Kempler S. , 2012: Tropical Rainfall Measuring Mission (TRMM) precipitation data and services for research and applications. Bull. Amer. Meteor. Soc., 93, 13171325, doi:10.1175/BAMS-D-11-00152.1.

    • Search Google Scholar
    • Export Citation
  • Schneider, U., , Becker A. , , Meyer-Christoffer A. , , Ziese M. , , and Rudolf B. , 2011: Global Precipitation Analysis Products of the GPCC. GPCC Doc., DWD, 13 pp. [Available online at ftp://ftp.dwd.de/pub/data/gpcc/PDF/GPCC_intro_products_v2011.pdf.]

  • Sorooshian, S., , Hsu K.-L. , , Gao X. , , Gupta H. V. , , Imam B. , , and Braithwaite D. , 2000: Evaluation of PERSIANN system satellite–based estimates of tropical rainfall. Bull. Amer. Meteor. Soc., 81, 20352046, doi:10.1175/1520-0477(2000)081<2035:EOPSSE>2.3.CO;2.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (a) Mean precipitation of the IMERG Final Run monthly product, averaged in the boreal summer (JJA) of 2014. (b) Mean differences (IMERG minus 3B43). (c) Relative [with respect to the IMERG mean in (a)] difference (IMERG minus 3B43).

  • View in gallery

    As in Fig. 1, but for the boreal winter (DJF) of 2014/15.

  • View in gallery

    Comparison of precipitation estimates between IMERG and 3B43 for the region (46°–50°N, 90°–124°E) during the boreal summer of 2014 and the boreal winter of 2014/15. (a) A map showing the selected region, (b) probability of liquid precipitation phase for the summer, (c) probability of liquid precipitation phase for the winter, (d) scatterplot for the summer, and (e) scatterplot for the winter.

  • View in gallery

    As in Fig. 3, but for the region (32°–35°N, 80°–89°E) in the Tibetan Plateau.

  • View in gallery

    Zonal means of 3B43 and IMERG monthly products in the boreal summer (JJA) of 2014 and the boreal winter (DJF) of 2014/15: (a) land and ocean for JJA, (b) land only for JJA, (c) ocean only for JJA, (d) land and ocean for DJF, (e) land only for DJF, and (f) ocean only for DJF.

  • View in gallery

    Scatterplots between IMERG and 3B43 monthly products (left) over land and (right) over ocean for (a),(d) June; (b),(e) July; and (c),(f) August 2014. The red dashed line is a 1:1 line and the blue solid line represents linear fit (Y = a + bX). Relational statistics: b, slope; yint, a or Y intercept of the fit; count, number of grid points; bias XY, the mean difference; std diff, standard deviation of the difference; and xycorr, correlation coefficient.

  • View in gallery

    As in Fig. 6, but for (a),(d) December 2014; (b),(e) January 2015; and (c),(f) February 2015.

  • View in gallery

    Zonal means of 3B43 and IMERG monthly products over ocean in the (left) boreal summer (JJA) of 2014 and (right) boreal winter (DJF) of 2014/15, derived from (a),(c) HQ and (b),(d) IR.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 73 73 37
PDF Downloads 68 68 32

Comparison of Integrated Multisatellite Retrievals for GPM (IMERG) and TRMM Multisatellite Precipitation Analysis (TMPA) Monthly Precipitation Products: Initial Results

View More View Less
  • 1 Center for Spatial Information Science and Systems, George Mason University, Fairfax, Virginia
© Get Permissions
Full access

Abstract

Launched on 27 February 2014, the Global Precipitation Measurement (GPM) mission comprises an international constellation of satellites to provide the next generation of global observations of precipitation. Built upon the success of the widely used TRMM Multisatellite Precipitation Analysis (TMPA) products, the Integrated Multisatellite Retrievals for GPM (IMERG) products continue to make improvements in areas such as spatial and temporal resolutions and snowfall estimates, etc., which will be valuable for research and applications. During the transition from TMPA to IMERG, characterizing the differences between these two product suites is important in order for users to make adjustments in research and applications accordingly. In this study, the newly released IMERG Final Run monthly product is compared with the TMPA monthly product (3B43) in the boreal summer of 2014 and the boreal winter of 2014/15 on a global scale. The results show the IMERG monthly product can capture major heavy precipitation regions in the Northern and Southern Hemispheres reasonably well. Differences between IMERG and 3B43 vary with surface types and precipitation rates in both seasons. Over land, systematic differences are much smaller compared to those over ocean because of the similar gauge adjustment used in the two monthly products. Positive relative differences (IMERG > 3B43) are primarily found at low precipitation rates and negative differences (IMERG < 3B43) at high precipitation rates. Over ocean, negative systematic differences (IMERG < 3B43) prevail at all precipitation rates. Analysis of the passive microwave (PMW) and infrared (IR) monthly products from TMPA and IMERG shows the large systematic differences in the tropical oceans are closely associated with the differences in the PMW products.

Denotes Open Access content.

Corresponding author address: Zhong Liu, Center for Spatial Information Science and Systems, George Mason University, 4400 University Dr., Fairfax, VA 22030. E-mail: zliu@gmu.edu

Abstract

Launched on 27 February 2014, the Global Precipitation Measurement (GPM) mission comprises an international constellation of satellites to provide the next generation of global observations of precipitation. Built upon the success of the widely used TRMM Multisatellite Precipitation Analysis (TMPA) products, the Integrated Multisatellite Retrievals for GPM (IMERG) products continue to make improvements in areas such as spatial and temporal resolutions and snowfall estimates, etc., which will be valuable for research and applications. During the transition from TMPA to IMERG, characterizing the differences between these two product suites is important in order for users to make adjustments in research and applications accordingly. In this study, the newly released IMERG Final Run monthly product is compared with the TMPA monthly product (3B43) in the boreal summer of 2014 and the boreal winter of 2014/15 on a global scale. The results show the IMERG monthly product can capture major heavy precipitation regions in the Northern and Southern Hemispheres reasonably well. Differences between IMERG and 3B43 vary with surface types and precipitation rates in both seasons. Over land, systematic differences are much smaller compared to those over ocean because of the similar gauge adjustment used in the two monthly products. Positive relative differences (IMERG > 3B43) are primarily found at low precipitation rates and negative differences (IMERG < 3B43) at high precipitation rates. Over ocean, negative systematic differences (IMERG < 3B43) prevail at all precipitation rates. Analysis of the passive microwave (PMW) and infrared (IR) monthly products from TMPA and IMERG shows the large systematic differences in the tropical oceans are closely associated with the differences in the PMW products.

Denotes Open Access content.

Corresponding author address: Zhong Liu, Center for Spatial Information Science and Systems, George Mason University, 4400 University Dr., Fairfax, VA 22030. E-mail: zliu@gmu.edu

1. Introduction

Precipitation is an important variable in the global hydrological cycle. Over the years, satellite-based precipitation products (i.e., Huffman et al. 2007, 2010; Joyce et al. 2004; Sorooshian et al. 2000; Behrangi et al. 2009; Aonashi et al. 2009) have been widely used in research and applications around the world. An example is the TRMM (Garstang and Kummerow 2000) Multisatellite Precipitation Analysis (TMPA; Huffman et al. 2007, 2010) that consists of near-real-time and research products available from 1998 (2000 for the near–real time) onward. Launched on 27 February 2014, the Global Precipitation Measurement (GPM) mission (Hou et al. 2014) comprises an international network of satellites to provide the next generation of global observations of rain and snow. Built upon the success of the widely used TMPA products, the newly released Integrated Multisatellite Retrievals for GPM (IMERG) products (Huffman et al. 2015a,b,c) continue to make improvements in areas such as spatial and temporal resolutions, snowfall estimates, etc., which will be valuable for research and applications around the world. During the transition from TMPA to IMERG, characterizing systematic differences between the two product suites is important for researchers and application users to allow them to make adjustments in research and applications accordingly. For example, several studies in the past (i.e., Liu 2015a,b) have been carried out to compare TMPA products in different versions and between the TMPA research and the near-real-time products on a global scale, providing product users additional information not to be found in the product readme documents. Having ground observations or ground truth is always desirable in product comparison and validation; however, high-resolution precipitation measurements over land on a global scale are very difficult to obtain and even more so over ocean, if available. For the time being, validation or verification of satellite-based precipitation products still heavily relies on individuals who can obtain ground measurement data often on a local or national scale. For such reasons, ground validation will be left to those individual investigators. This paper will only focus on characterizing the difference between the two popular multisatellite monthly products.

Preliminary comparisons between the IMERG Final Run and the TMPA monthly products for June 2014 have been carried out by Huffman et al. (2015c). Their results (relevant to this study) show that a close similarity is found between the two monthly products over land since both use the same gauge product from the Global Precipitation Climatology Centre (GPCC; Schneider et al. 2011) and the same algorithm for bias correction (Huffman et al. 2015c). Nonetheless, more comprehensive comparisons are still needed to better understand their differences, such as in spatial and temporal domains, at different precipitation rates, over surface types, etc. In this study, the Day 1 IMERG Final Run monthly product (Huffman et al. 2015c) is compared with the TMPA monthly research product (3B43) in the boreal summer [June–August (JJA)] of 2014 and the boreal winter [December–February (DJF)] of 2014/15 on a global scale. Despite the limited data in the initial release of IMERG, results to be presented here show that systematic differences between the two products do exist and vary with surface types and precipitation rates. This paper is organized as follows: section 2 describes the data and methods, section 3 presents the results, and section 4 presents the summary and discussion.

2. Datasets

The TMPA algorithm (Huffman 1997; Huffman et al. 2007, 2010; Huffman and Bolvin 2014) consists of multiple independent precipitation estimates from various passive microwave (PMW) sources (Table 1), microwave-adjusted merged geo-infrared (IR) and monthly accumulated rain gauge analysis from GPCC (Schneider et al. 2011). The preprocessing of 3B43 is as follows (Huffman and Bolvin 2014): 1) all input PMW products in Table 1 are intercalibrated to TRMM Combined Instrument (TCI) precipitation estimates (TRMM product 3B31); 2) the IR estimates are computed using monthly matched microwave–IR histogram matching; and 3) then missing data in individual 3-hourly merged-microwave fields are filled with the IR estimates. When the preprocessing is complete, the 3-hourly multisatellite fields are summed for the month and combined with the monthly GPCC gauge analysis (Schneider et al. 2011) using inverse-error-variance weighting to form the best-estimate precipitation rate and root-mean-square (RMS) precipitation error estimates (Huffman and Bolvin 2014). All data fields of the TMPA research products are listed in Table 2.

Table 1.

List of PMW sources used in IMERG and TMPA (Huffman 1997; Huffman et al. 2015b; Huffman and Bolvin 2014) including Sounder for Atmospheric Profiling of Humidity in the Intertropics by Radiometry (SAPHIR), Advanced Technology Microwave Sounder (ATMS), Atmospheric Infrared Sounder (AIRS), Cross-Track Infrared Sounder (CRIS), and TRMM Combined Instrument (TCI) algorithms (2B31). TMI data ended on 8 Apr 2015.

Table 1.
Table 2.

Summary of the TMPA research (Huffman 1997; Huffman et al. 2007, 2010; Huffman and Bolvin 2014) and IMERG Final Run products (Huffman et al. 2015a,b,c). As of this writing, the coverage for IMERG is 60°N–60°S and will be extended to 90°N–90°S in future releases.

Table 2.

The GPM Day 1 IMERG algorithm (Huffman et al. 2015a,b,c) integrates several multisatellite retrievals from NASA TMPA (intersatellite calibration and gauge adjustment); NOAA Climate Prediction Center (CPC) morphing technique (CMORPH; Lagrangian time interpolation); PERSIANN at the University of California, Irvine (microwave calibrated IR using artificial neural networks); and the NASA Precipitation Processing System (input data assembly and processing). IMERG (Huffman et al. 2015a,b,c) contains precipitation estimates from PMW sensors (Table 1) on board various precipitation-relevant satellites in the GPM constellation and the zenith-angle-corrected, intercalibrated IR fields merged from several geostationary satellites (Janowiak et al. 2001). PMW precipitation estimates are computed using the 2014 version of the Goddard profiling algorithm (GPROF2014; Kummerow et al. 2014a). Major changes in GPROF2014 (Kummerow et al. 2014b) include the following:

  1. Pre-GPM launch database is used and data from the core GPM satellite are not yet used to create a priori databases.
  2. Over ocean, it is the same as TRMM.
  3. Over land, the algorithms have been fundamentally changed from TRMM, and surface radars [National Mosaic and Multi-Sensor QPE (NMQ)] over the United States were used to construct databases of observed surface rain and sensor brightness temperatures (Tb) for each radiometer; Special Sensor Microwave Imager/Sounder (SSMIS) database is used in the GPM Microwave Imager (GMI).
  4. For cold surface, AMSR2/Microwave Sounding Unit (MHS) was used with CloudSat rain and multiscale modeling framework (MMF) for physically consistent database and all sensors used AMSR2 and MHS channels.

By contrast, TMPA uses various early versions of GPROF (Huffman and Bolvin 2014). Table 1 lists the PMW sources for TMPA and IMERG. It is seen that fewer sensors are used in TMPA, and furthermore, none of the GPM sensors are used in TMPA (Huffman et al. 2015b; Huffman and Bolvin 2014), which suggests significant changes in PMW precipitation estimates from TMPA to IMERG. The merged IR data (Janowiak et al. 2001) are provided by the CPC. Finally, the PMW and IR precipitation estimates are used by the CMORPH–Kalman filter Lagrangian time interpolation to generate half-hourly estimates. Similarly, the monthly GPCC precipitation analysis (Schneider et al. 2011) is used for bias correction. IMERG contains monthly and half-hourly products (Table 2). Three products are in the half-hourly product suite (Huffman et al. 2015b,c): Early Run (latency ~6 h after observation time), Late Run (latency ~18 h), and Final Run (latency ~4 months). In this study, the Final Run products are used.

Table 2 is a summary of the four products from IMERG Final Run and TMPA research products used in this study. For the IMERG monthly product (Table 2), a new data field (probability of liquid precipitation) that helps to identify precipitation types (liquid, frozen, or mixed) has been added (Huffman et al. 2015b,c). In addition, the spatial resolution in IMERG has been increased to 0.1° and the spatial coverage in IMERG has been extended (60°N–60°S at present and to be expanded to 90°N–90°S in future releases) beyond the TMPA 50°N–50°S coverage (Huffman et al. 2015b,c). In this study, the spatial coverage is limited to 50°N–50°S to match that of TMPA. For the 3-hourly and half-hourly products, three more data fields (Table 2) have been added in the IMERG half-hourly product—probability of liquid precipitation, uncalibrated precipitation, and IR Kalman filter weight—to gain further insight on the multisatellite derived precipitation estimates.

Because of a difference in the spatial resolutions between IMERG and TMPA (Table 2), IMERG products need to be regridded to match the TMPA grid before comparisons can be made. In this study, a simple box-averaging method is used. To understand systematic differences, the half-hourly IMERG and 3-hourly TMPA products [high-quality precipitation from PMW sensors (HQ) and IR precipitation variables] in Table 2 are used to compute their monthly products. Both TMPA (version 7) and IMERG (version 03D) Final Run data in this study were downloaded from Mirador (http://mirador.gsfc.nasa.gov) at the NASA Goddard Earth Sciences Data and Information Services Center (GES DISC; Liu et al. 2012). There have been a few processing issues (Huffman and Bolvin 2014) with TMPA before, but all the TMPA data used in this study are current.

3. Results

a. Systematic difference

Figure 1a shows the mean IMERG monthly precipitation for the boreal summer (JJA) of 2014. It is seen that the IMERG monthly product can successfully capture several well-known regions of heavy precipitation, such as along the coasts of the Western Ghats in India, Myanmar (Burma), the western Philippines, and New Britain of Papua New Guinea, where the orographic effects take an important role in producing heavy precipitation on the windward side and the nearby ocean. Heavy precipitation zones in the eastern equatorial Pacific Ocean and Atlantic Ocean, both of which are associated with the intertropical convergence zone (ITCZ), are captured reasonably well (Fig. 1a).

Fig. 1.
Fig. 1.

(a) Mean precipitation of the IMERG Final Run monthly product, averaged in the boreal summer (JJA) of 2014. (b) Mean differences (IMERG minus 3B43). (c) Relative [with respect to the IMERG mean in (a)] difference (IMERG minus 3B43).

Citation: Journal of Hydrometeorology 17, 3; 10.1175/JHM-D-15-0068.1

Figure 1b shows mean systematic differences (IMERG minus 3B43) between the IMERG and 3B43 monthly products for the boreal summer (JJA) of 2014. Overall, small differences are found over land and much larger differences over ocean (Fig. 1b). The former is not a surprise because of the same GPCC gauge adjustment for bias correction applied to both products (Huffman et al. 2015c), as mentioned in the introduction. Over land, positive and negative differences are scattered. Over ocean, negative differences (IMERG < 3B43) prevail, and large negative differences are found in heavy precipitation regions such as the ITCZ, which may be associated with the significant changes in the PMW algorithms mentioned in section 2.

Figure 1c shows the relative (with respect to the IMERG mean in Fig. 1a) mean difference between the IMERG and 3B43 monthly products in the boreal summer of 2014. In agreement with Fig. 1b, small relative differences are found over land and large differences over ocean (Fig. 1c). A close examination shows that some large positive (IMERG > 3B43) and negative (IMERG < 3B43) differences over land scatter in light precipitation regions such as western China, the Sahara Desert, and the Arabian Peninsula. A similar situation is also found in the austral winter in light precipitation regions over land in the Southern Hemisphere (Fig. 1c), such as southern Africa, Australia, and Brazil. Over ocean, negative relative differences (IMERG < 3B43) prevail, except in the belt near 50°S where positive differences (IMERG > 3B43) dominate. In Fig. 1c, it is seen that, over ocean, small negative relative differences are found in heavy precipitation regions such as the ITCZ and large relative differences are predominantly found in light precipitation regions; however, some large positive differences are found in very light precipitation regions as well, such as the oceans off the coasts of California, northern Africa, Chile, and Peru.

Similar analysis (Fig. 2) was carried out for the boreal winter of 2014/15. Figure 2a shows the seasonal shift of precipitation regions and IMERG catches major precipitation regions such as the ITCZ and monsoon regions in the Southern Hemisphere. Similar to the boreal summer of 2014, small differences between IMERG and 3B43 are found over land (Fig. 2b) and large differences are found over ocean, especially in heavy precipitation regions in the tropics. The relative systematic difference analysis in Fig. 2c also shows similar results, with prevailing negative relative mean differences over ocean and small differences over land, except in some light precipitation regions such as in North Africa.

Fig. 2.
Fig. 2.

As in Fig. 1, but for the boreal winter (DJF) of 2014/15.

Citation: Journal of Hydrometeorology 17, 3; 10.1175/JHM-D-15-0068.1

To further illustrate the gauge adjustment issue in both boreal summer and winter, two areas were chosen—one in Mongolia and China (Fig. 3a) and the other in the Tibetan Plateau (Fig. 4a)—for their well-defined precipitation types (dominating liquid precipitation in summer and frozen in winter; Figs. 3b, 3c, 4b, 4c) and gauge availability. The former area contains more gauge measurements (on average, 50% gauge relative weight in the boreal summer and 59% in the boreal winter) than the latter (on average, 5% gauge relative weight in the boreal summer and 6% in the boreal winter). In the relatively more gauged area in Fig. 3, it is seen that both IMERG and 3B43 agree well with each other (Figs. 3d,e) in terms of high correlation coefficients (0.933 in the boreal summer and 0.8209 in the boreal winter). By contrast, in the less gauged area in the Tibetan Plateau (Fig. 4), the agreement between IMERG and 3B43 is not that good. Their correlation coefficients (Figs. 4d,e) are 0.502 in the boreal summer and 0.059 in the boreal winter. Nonetheless, the comparison statistics for the two areas demonstrate the importance of the gauge adjustment in describing their difference.

Fig. 3.
Fig. 3.

Comparison of precipitation estimates between IMERG and 3B43 for the region (46°–50°N, 90°–124°E) during the boreal summer of 2014 and the boreal winter of 2014/15. (a) A map showing the selected region, (b) probability of liquid precipitation phase for the summer, (c) probability of liquid precipitation phase for the winter, (d) scatterplot for the summer, and (e) scatterplot for the winter.

Citation: Journal of Hydrometeorology 17, 3; 10.1175/JHM-D-15-0068.1

Fig. 4.
Fig. 4.

As in Fig. 3, but for the region (32°–35°N, 80°–89°E) in the Tibetan Plateau.

Citation: Journal of Hydrometeorology 17, 3; 10.1175/JHM-D-15-0068.1

b. Zonal mean precipitation

Figure 5 shows the zonal mean precipitation analysis for the IMERG and 3B43 monthly products, averaged in the boreal summer of 2014 and the boreal winter of 2014/15 in several scenarios. The zonal means (Fig. 5a) in the boreal summer confirm the findings (Fig. 1b) shown earlier that precipitation estimates from the IMERG monthly product are, in general, lower than those of 3B43. The largest negative differences (IMERG < 3B43) are found in the ITCZ (Fig. 5a) and are followed by the heavy precipitation area over the ocean near New Britain of Papua New Guinea (Fig. 1a). However, positive differences (IMERG > 3B43) are found near 50°S, as found earlier in Fig. 1b as well, which perhaps links to the mixed precipitation near 50°S in the southern Indian Ocean and South Atlantic based on the IMERG accumulation-weighted probability of liquid precipitation phase (not shown). Over land (Fig. 5b), the differences in zonal means are small and almost identical, suggesting both products are similar because of the same gauge adjustment for bias correction applied to both products (Huffman et al. 2015c). By contrast, much larger differences are found over ocean (Fig. 5c), where the 3B43 zonal means are larger than those of IMERG, especially in the ITCZ, which resembles the pattern found in Fig. 5a. Over ocean, the positive differences mentioned earlier are found near 50°S. In the boreal winter of 2014/15, the situations (Figs. 5d–f) are quite similar to those in the boreal summer.

Fig. 5.
Fig. 5.

Zonal means of 3B43 and IMERG monthly products in the boreal summer (JJA) of 2014 and the boreal winter (DJF) of 2014/15: (a) land and ocean for JJA, (b) land only for JJA, (c) ocean only for JJA, (d) land and ocean for DJF, (e) land only for DJF, and (f) ocean only for DJF.

Citation: Journal of Hydrometeorology 17, 3; 10.1175/JHM-D-15-0068.1

c. Scatterplots and relative systematic differences at different precipitation rates

As seen earlier, both monthly products behave differently over land and ocean and large zonal differences between the two are found over ocean (Fig. 5); therefore, land and ocean need to be separated in the following analysis. Figure 6 contains scatterplots over land and ocean in JJA of 2014. Over land (Figs. 6a–c), precipitation estimates are less scattered near the 1:1 line, compared to those over ocean (Figs. 6d–f). As shown in Figs. 3 and 4, the GPCC gauge adjustment does make a difference in narrowing the difference between IMERG and 3B43, which may explain the finding here. In June and July, the blue (linear fit) lines are slightly above the red (1:1) line, suggesting slight negative (IMERG < 3B43) systematic differences over land. In August, both blue and red lines are almost completely overlapping each other, as indicated by the parameter b (1.0009). Over ocean, precipitation estimates are more scattered and the systematic differences are more significant (Figs. 6d–f). The negative systematic differences (IMERG < 3B43) remain consistent throughout the boreal summer. In the boreal winter of 2014/15 (Fig. 7), the similar situations are found, compared to those in the boreal summer in Fig. 6.

Fig. 6.
Fig. 6.

Scatterplots between IMERG and 3B43 monthly products (left) over land and (right) over ocean for (a),(d) June; (b),(e) July; and (c),(f) August 2014. The red dashed line is a 1:1 line and the blue solid line represents linear fit (Y = a + bX). Relational statistics: b, slope; yint, a or Y intercept of the fit; count, number of grid points; bias XY, the mean difference; std diff, standard deviation of the difference; and xycorr, correlation coefficient.

Citation: Journal of Hydrometeorology 17, 3; 10.1175/JHM-D-15-0068.1

Fig. 7.
Fig. 7.

As in Fig. 6, but for (a),(d) December 2014; (b),(e) January 2015; and (c),(f) February 2015.

Citation: Journal of Hydrometeorology 17, 3; 10.1175/JHM-D-15-0068.1

Table 3 lists relative (with respect to the IMERG mean) systematic differences between IMERG and 3B43 monthly products at different precipitation rates over land and ocean for the boreal summer of 2014 and the boreal winter of 2014/15. In the boreal summer, the largest positive relative systematic difference (IMERG > 3B43) over land is found at the lowest precipitation rate. However, it decreases rapidly and becomes negative (IMERG < 3B43) at 7.5 mm day−1 and above. Over ocean, negative relative systematic differences (IMERG < 3B43) prevail at all precipitation rates and large negative values are found at low precipitation rates. In Table 3, it is shown that precipitation rates over ocean are higher than those over land. In short, the findings here are consistent with those described earlier. Similar situations are found in the boreal winter (Table 3) as well: negative relative systematic differences dominate over ocean and smaller differences are found over land.

Table 3.

Relative (with respect to the IMERG mean) systematic differences (%) between IMERG and 3B43 monthly products over land and ocean as a function of precipitation rate (mm day−1) during the boreal summer (JJA) of 2014 and the boreal winter (DJF) of 2014/15.

Table 3.

d. Comparison of monthly means derived from passive microwave and infrared products

As mentioned in section 2, all monthly products from IMERG and TMPA contain precipitation estimates that are combined from microwave (HQ) and IR products through different algorithms. To better understand and explain the observed differences found earlier, it is necessary to compare their monthly products derived from HQ and IR separately. Since differences are small over land, as shown in the zonal means in Fig. 5, only those over ocean are examined. Figures 8a and 8c show the HQ zonal precipitation means averaged during the boreal summer of 2014 and the boreal winter of 2014/15, respectively. In the boreal summer, it is seen that 3B43 HQ zonal means are in general larger than those of IMERG, except near 50°S, and the largest differences are found in the ITCZ and the heavy precipitation area over the ocean near New Britain of Papua New Guinea (Fig. 8a). Near 50°S, it is seen that IMERG HQ estimates are larger than those of 3B43 (Fig. 8a), which may be associated with mixed precipitation types there. All these suggest that the large differences near the ITCZ are primarily due to the HQ product differences between 3B43 and IMERG. For IR (Fig. 8b), the differences are small, except over ocean in the Southern Hemisphere. The differences near 50°S are also quite small, compared to HQ. In the boreal winter of 2014/15, the situation is similar; namely, the largest differences are associated with the HQ products in the ITCZ (Fig. 8c). Unlike the boreal summer, some differences in IR are found in the latitudes north of 15°N (Fig. 8d).

Fig. 8.
Fig. 8.

Zonal means of 3B43 and IMERG monthly products over ocean in the (left) boreal summer (JJA) of 2014 and (right) boreal winter (DJF) of 2014/15, derived from (a),(c) HQ and (b),(d) IR.

Citation: Journal of Hydrometeorology 17, 3; 10.1175/JHM-D-15-0068.1

4. Summary and discussion

Two monthly products from TMPA and IMERG have been compared in the boreal summer of 2014 and the boreal winter of 2014/15 to characterize their systematic differences on a global scale. Initial results show that the IMERG monthly product can capture major heavy precipitation regions in both the Northern and Southern Hemispheres reasonably well. Systematic differences between IMERG and 3B43 vary with surface types and precipitation rates. Over land, systematic differences are smaller than over ocean because of the same GPCC gauge adjustment for bias correction used in IMERG and 3B43. Positive and negative mean differences scatter over land. By contrast, negative mean differences (IMERG < 3B43) prevail over ocean. Over land, positive relative mean differences (IMERG > 3B43) are primarily found at low precipitation rates and negative differences (IMERG < 3B43) at higher precipitation rates. Over ocean, negative relative mean differences prevail at all precipitation rates in both seasons. Analysis of monthly PMW and IR data shows that these differences, especially near the ITCZ, are closely associated with the differences between the TMPA and IMERG HQ products.

Although the changes in the PMW algorithms are listed in section 2, the observed large differences in PMW precipitation estimates over tropical oceans can be complicated by a number of factors, such as temporal resolution, grid resolution, the IMERG prelaunch database, new algorithms, different sensors used in IMERG and 3B43 (Table 1), different PMW calibrations, and the lack of similar parameters in 3B42 and 3B43 (Table 2). For example, the IMERG HQ product consists of PMW datasets from different sensors and satellites. As seen above, only the TMI dataset is identical to that in the IMERG HQ product and it is not clear for the rest of HQ products listed in Table 1. To further understand the PMW differences over ocean, analysis or comparison of orbital products is needed.

Over land, it is more complicated. In addition to the factors just mentioned, the GPCC gauge adjustment is another important factor in describing differences between IMERG and 3B43. Improvement in snowfall estimates is one of the main GPM mission features. However, as seen in the analysis above, differences over land in the boreal winter of 2014/15 are also small, raising questions such as how much improvement the IMERG algorithm can make in terms of snowfall estimates. Nonetheless, in-depth analysis or comparison of the half-hourly IMERG, orbital products, and ground observation is needed. The lack of uncalibrated parameter in 3B42 is another difficult factor to assess and remove the influence of GPCC gauge adjustment.

Likewise, the IR algorithms have been changed from TMPA to IMERG, although both use the same CPC merged IR product. For example, in TMPA, the monthly matched microwave–IR histogram matching algorithm is used in TMPA while the neural networks from the University of California, Irvine, are used in IMERG. The observed differences in Fig. 8 may be closely related to these changes.

For the time being, it is not clear whether the results found in this study will be similar in other boreal summer and winter seasons before 2014 because the data have not been released yet, suggesting more studies are needed. Similar studies can be extended to other seasons before 2014 as well when data are available.

Higher temporal resolution precipitation products such as half-hourly are frequently used in precipitation process studies and applications such as flooding, landslide, etc. It is necessary to continue and extend the comparison work to these products. Meanwhile, the additional data fields will very likely provide further insight on product differences, and efforts on analyzing them are also needed.

Acknowledgments

This project is supported by NASA Research Opportunities in Space and Earth Science-2010 (ROSES-2010), NNH10ZDA001N-ESDRERR, Appendix A.32: “Earth System Data Records Uncertainty Analysis.” Special thanks to the GES DISC. The TMPA and IMERG data were provided by the NASA Goddard Space Flight Center’s Mesoscale Atmospheric Processes Laboratory and Precipitation Processing System (PPS), which develop and compute the TMPA and IMERG as a contribution to TRMM and GPM, and archived at NASA GES DISC. I would like to extend thanks to the anonymous reviewers for their thorough review and constructive comments.

REFERENCES

  • Aonashi, K., and Coauthors, 2009: GSMaP passive, microwave precipitation retrieval algorithm: Algorithm description and validation. J. Meteor. Soc. Japan, 87A, 119136, doi:10.2151/jmsj.87A.119.

    • Search Google Scholar
    • Export Citation
  • Behrangi, A., , Hsu K.-L. , , Imam B. , , Sorooshian S. , , Huffman G. J. , , and Kuligowski R. J. , 2009: PERSIANN-MSA: A precipitation estimation method from satellite-based multispectral analysis. J. Hydrometeor., 10, 14141429, doi:10.1175/2009JHM1139.1.

    • Search Google Scholar
    • Export Citation
  • Garstang, M., , and Kummerow C. D. , 2000: The Joanne Simpson special issue on the Tropical Rainfall Measuring Mission (TRMM). J. Appl. Meteor., 39, 19611961, doi:10.1175/1520-0450(2001)040<1961:TJSSIO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Hou, A. Y., and Coauthors, 2014: The Global Precipitation Measurement Mission. Bull. Amer. Meteor. Soc., 95, 701722, doi:10.1175/BAMS-D-13-00164.1.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., 1997: Estimates of root-mean-square random error for finite samples of estimated precipitation. J. Appl. Meteor., 36, 11911201, doi:10.1175/1520-0450(1997)036<1191:EORMSR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., , and Bolvin D. T. , 2014: TRMM and other data precipitation data set documentation. TRMM Doc., 42 pp. [Available online at ftp://meso-a.gsfc.nasa.gov/pub/trmmdocs/3B42_3B43_doc.pdf.]

  • Huffman, G. J., and Coauthors, 2007: The TRMM Multi-satellite Precipitation Analysis: Quasi-global, multi-year, combined-sensor precipitation estimates at fine scale. J. Hydrometeor., 8, 3855, doi:10.1175/JHM560.1.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., , Adler R. F. , , Bolvin D. T. , , and Nelkin E. J. , 2010: The TRMM Multi-Satellite Precipitation Analysis (TMPA). Satellite Rainfall Applications for Surface Hydrology, F. Hossain and M. Gebremichael, Eds., Springer-Verlag, 3–22.

  • Huffman, G. J., , Bolvin D. T. , , Braithwaite D. , , Hsu K. , , Joyce R. , , Kidd C. , , Nelkin E. J. , , and Xie P. , 2015a: NASA Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG). Algorithm Theoretical Basis Doc., version 4.5, 26 pp. [Available online at http://pmm.nasa.gov/sites/default/files/document_files/IMERG_ATBD_V4.5.pdf.]

  • Huffman, G. J., , Bolvin D. T. , , and Nelkin E. J. , 2015b: Integrated Multi-satellitE Retrievals for GPM (IMERG) technical documentation. NASA Doc., 47 pp. [Available online at http://pmm.nasa.gov/sites/default/files/document_files/IMERG_doc.pdf.]

  • Huffman, G. J., , Bolvin D. T. , , and Nelkin E. J. , 2015c: Day 1 IMERG final run release notes. NASA Doc., 9 pp. [Available online at http://pmm.nasa.gov/sites/default/files/document_files/IMERG_FinalRun_Day1_release_notes.pdf.]

  • Janowiak, J. E., , Joyce R. J. , , and Yarosh Y. , 2001: A real-time global half-hourly pixel-resolution infrared dataset and its applications. Bull. Amer. Meteor. Soc., 82, 205217, doi:10.1175/1520-0477(2001)082<0205:ARTGHH>2.3.CO;2.

    • Search Google Scholar
    • Export Citation
  • Joyce, R. J., , Janowiak J. E. , , Arkin P. A. , , and Xie P. , 2004: CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hydrometeor., 5, 487503, doi:10.1175/1525-7541(2004)005<0487:CAMTPG>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kummerow, C., , Randel D. , , Kirstetter P. , , Wang N.-Y. , , Petkovic V. , , Kulie M. , , and Ferraro R. , 2014a: Global Precipitation Measurement (GPM) mission. Algorithm Theoretical Basis Doc., version 1.4, 46 pp. [Available online at http://rain.atmos.colostate.edu/ATBD/ATBD_GPM_Aug1_2014.pdf.]

  • Kummerow, C., , Randel D. , , Kirstetter P. , , Wang N.-Y. , , Petkovic V. , , Kulie M. , , and Ferraro R. , 2014b: Initial GPROF GPM results using a GPM derived database. Seventh IPWG Workshop on Precipitation Measurements, Tsukuba, Japan, International Precipitation Working Group. [Available online at http://www.isac.cnr.it/~ipwg/meetings/tsukuba-2014/pres/4-2_Kummerow.pdf.]

  • Liu, Z., 2015a: Comparison of precipitation estimates between version 7 3-hourly TRMM Multi-Satellite Precipitation Analysis (TMPA) near-real-time and research products. Atmos. Res., 153, 119133, doi:10.1016/j.atmosres.2014.07.032.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., 2015b: Comparison of versions 6 and 7 3-Hourly TRMM Multi-Satellite Precipitation Analysis (TMPA) research products. Atmos. Res., 163, 91101, doi:10.1016/j.atmosres.2014.12.015.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., , Ostrenga D. , , Teng W. , , and Kempler S. , 2012: Tropical Rainfall Measuring Mission (TRMM) precipitation data and services for research and applications. Bull. Amer. Meteor. Soc., 93, 13171325, doi:10.1175/BAMS-D-11-00152.1.

    • Search Google Scholar
    • Export Citation
  • Schneider, U., , Becker A. , , Meyer-Christoffer A. , , Ziese M. , , and Rudolf B. , 2011: Global Precipitation Analysis Products of the GPCC. GPCC Doc., DWD, 13 pp. [Available online at ftp://ftp.dwd.de/pub/data/gpcc/PDF/GPCC_intro_products_v2011.pdf.]

  • Sorooshian, S., , Hsu K.-L. , , Gao X. , , Gupta H. V. , , Imam B. , , and Braithwaite D. , 2000: Evaluation of PERSIANN system satellite–based estimates of tropical rainfall. Bull. Amer. Meteor. Soc., 81, 20352046, doi:10.1175/1520-0477(2000)081<2035:EOPSSE>2.3.CO;2.

    • Search Google Scholar
    • Export Citation
Save