Intercomparison of the Version-6 and Version-7 TMPA precipitation products over high and low latitudes basins with independent gauge networks: Is the newer version better in both real-time and post-real-time analysis for water resources and hydrologic extremes?

Highlights

• We intercompared Version-6 and -7 TMPA estimates by independent gauge networks.

• The Version-7 TMPA generally represents a substantial improvement over Version-6.

• The incorporation of SSMIS in Version-7 provides more high-quality microwave data.

• The Version-7 real-time estimates still have much room to improve at high latitude.

• The new Version7 TMPA estimates still underestimate the rainfall of typhoon events.

Summary

The TRMM Multi-satellite Precipitation Analysis (TMPA) system underwent an important upgrade in early 2013, at which the newest Version-7 TMPA products were formally released. In this study, the TMPA successive versions, original Version-6 and current Version-7, were evaluated and intercompared by using independent gauge observation networks for a 7-yr (2003–09) period over two representative basins in China at different latitude bands. The TMPA products studied are the Version-6 and Version-7 real-time 3B42RT estimates (RTV6 and RTV7) and post-real-time 3B42 estimates (V6 and V7). Assessments indicate that RTV7 represents a substantial improvement over RTV6 with respect to the systematic bias in the low-latitude Mishui basin, reaching similar accuracy levels as with the gauge-adjusted research products. But, such improvement was not found in the high-latitude Laohahe basin, suggesting that the current Version 7 TMPA real-time estimates still have much room for improvement at high latitudes. On the other hand, the post-real-time research product V7, which is expected to provide better precipitation information for water resources management in ungauged regions, generally outperforms V6 over both gauged basins and has the best performance among the four standard TMPA estimates. The seasonal analyses show that the new Version-7 algorithm notably reduces the bias between TMPA and observations during winter months for the low-latitude Mishui basin, but fails to effectively alleviate the serious overestimation for winter precipitation occurring in the high-latitude Laohahe basin. The study also reveals that all the TMPA products significantly underestimate high rain rates over the Mishui basin, especially for strong typhoon events during summer. Thus, caution should be exercised when applying the current Version-7 TMPA products for simulation and prediction of hydrologic extremes associated with heavy rainfall, such as floods or landslides.

Introduction

The growing availability of high-resolution satellite precipitation products provides hydrologists with tremendous opportunities to improve hydrological process simulation and flood prediction capacity in medium to large river basins, especially in the developing countries and remote regions where in situ measurement networks are sparse (Anagnostou, 2004, Hong et al., 2006, Hossain and Lettenmaier, 2006, Ebert et al., 2007, Gourley et al., 2010, Yong et al., 2010, Bitew and Gebremichael, 2011, Wang et al., 2012, Xue et al., 2013, Yong et al., 2013a; among many others). As a prelude to the forthcoming Global Precipitation Measurement (GPM) mission (http://pmm.nasa.gov/GPM), the current operational TRMM Multi-satellite Precipitation Analysis (TMPA), which combines low-Earth orbiting passive microwave (PMW, with better retrievals) data and geostationary infrared (IR, with excellent sampling) data from almost all available precipitation-related satellite platforms, is intended to produce the best estimates of quasi-global precipitation (50° N–S). In the present implementation, the standard TMPA products are available at 0.25° × 0.25°, 3-hourly resolution in both real-time (i.e., TMPA-RT) and post-real-time (i.e., research quality product) to accommodate different needs for a wide range of researchers and users (Huffman et al., 2007).

During the past 10 years, the TMPA algorithm produced at the National Aeronautics and Space Administration (NASA) Goddard Space Flight Center (GSFC) has undergone three major upgrades due to the new sensors used and algorithmic upgrades. First, Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) on Aqua and three Advanced Microwave Sounding Unit-B (AMSU-B) sensors onboard the National Oceanic and Atmospheric Administration (NOAA)-series satellites (NOAA-15, -16, -17) were introduced into the PMW mosaics in the TMPA algorithm on 3 February 2005. The use of AMSR-E and AMSU-B nearly doubles the typical combined microwave coverage from the 50° N–S latitude band from ∼45% to nearly 80% (Huffman et al., 2007). This has resulted in improvements in the performance of TMPA real-time estimates by reducing the errors associated with the inadequate samples by combining PMW information from three prior NOAA Polar-orbiting Operational Environmental Satellite (POES) Satellites (Su et al., 2011, Yong et al., 2012). Second, Version-6 of the TMPA real-time products was released on 17 February 2009. Compared to the original Version 5, one primary feature of Version-6 is that a climatological calibration algorithm (CCA) was applied in TMPA’s real-time estimates so as to reduce the systematic bias. For this new calibration procedure, the developers determined a local histogram matching of 3 hourly 2A12 (TRMM Microwave Imager, or TMI) to 2B31 (TRMM Combined Instrument, or TCI) and ratio adjustment of TCI to 3B43V6 (another monthly TRMM product), computed from 10 years of coincident data to establish the climatology for each calendar month. Then, the TMI-TCI and TCI-3B43 calibrations are successively applied to the Version 5 TMPA-RT estimates to create the calibrated precipitation field in the Version-6 TMPA-RT datasets. As a result, the systematic errors in the TMPA-RT estimates were generally alleviated over land from the global perspective (Huffman et al., 2010). A detailed description regarding the above two revisions of TMPA can be found in Yong et al., 2012, Yong et al., 2013b.

The last crucial upgrade for TMPA system occurred on 25 June 2012, on which the Version-7 of the TMPA real-time products was released. Relative to original Version-6, the new Version-7 TMPA system has more-substantial upgrades: (1) The Special Sensor Microwave Imager Sounder (SSMIS) on Defense Meteorological Satellite Program (DMSP) F-16 satellites were incorporated into TMPA-RT, which is anticipated to further improve the data accuracy of microwave precipitation estimates. (2) The new Global Precipitation Climatology Centre (GPCC V2.2) full analysis, substituting the old GPCC monitoring product before April 2005 and the Climate Assessment and Monitoring System (CAMS) analysis thereafter used in Version-6, is employed as the gauge analysis of CCA in the TMPA-RT and the month-to-month gauge adjustments in the research product 3B42V7. The GPCP V2.2 includes a slight upgrade to the gauge analysis input datasets, particularly over China. In November 2012, unfortunately, the AMSU data were omitted in the first release of TMPA precipitation estimates including the TMPA-RT data series (3B40RT, 3B41RT, and 3B42RT) and the official Version-7 research products (3B42V7 and 3B43V7). Afterward, the TMPA developers quickly corrected this processing issue and retrospectively produced these products again. On 28 January 2013, the latest Version-7 TMPA-RT (2000-now) and research products (2000–2010) were formally released so as to provide the users a new backlog for validation and application activities. Meanwhile, the retrieval developers called upon all the TMPA users to immediately switch to the newest Version-7 datasets as soon as practical. More details about this update can be seen in Huffman and Bolvin (2013).

The dynamical updates of TMPA depicted above can help the users to better understand and analyze the changes of data accuracy and hydrologic potential of the TMPA products across the different developing stages. In previous studies, many efforts have been made to investigate the error and uncertainty characteristics of the Version-6 TMPA products (Scheel et al., 2010, Shen et al., 2010, Yong et al., 2010, Gourley et al., 2011, Samaniego et al., 2011, Stampoulis and Anagnostou, 2012, Chen et al., 2013; among others). Furthermore, several recent studies have reported on the improvement of the Version-6 TMPA over earlier versions (e.g., Su et al., 2011, Yong et al., 2012). As the era toward a Global Precipitation Measurement mission (GPM) approaches, focus on new measurements will become inevitable. Moreover, the latest Version-7 will likely be the last one for TMPA data. Thus, the objective of this study is designed to quantify the improvements of the latest Version-7 TMPA estimates relative to its predecessor (Version-6) over two typical basins in China at different latitude bands. This ground validation exercise will be relevant and useful in the GPM era, as the lessons learned in regards to TMPA’s error characteristics will certainly carry over to GPM algorithms. In this study, we will focus on: (1) What are the spatiotemporal error characteristics of four standard TMPA products including the real-time Version-6 3B42RT (hereafter refer to as “RTV6”) and post-real-time 3B42V6 (hereafter “V6”), and the corresponding Version-7 3B42RT (hereafter “RTV7”) and 3B42V7 (hereafter “V7”), and how much do they differ? (2) What changes happened to the TMPA estimates in the rainfall intensity distributions after the latest system upgrade? (3) What implications do results from this study have on future hydrologic utility (e.g., water resources planning and management, hydrologic extremes monitoring and warning) of the Version-7 TMPA datasets in ungauged basins? Therefore, it seems timely for this paper to employ the independent ground observations over two representative basins at different latitudes to directly evaluate and intercompare the TMPA successive version products.

In the next section, we describe the study basins and the datasets used. A presentation of the results and analysis follows in Section 3. Summarizing remarks and conclusions finalize the paper in Section 4.