Comparison of TMPA-3B42 Versions 6 and 7 Precipitation Products with Gauge-Based Data over India for the Southwest Monsoon Period

Satya Prakash National Centre for Medium Range Weather Forecasting, Earth System Science Organization, Ministry of Earth Sciences, Noida, India

Search for other papers by Satya Prakash in
Current site
Google Scholar
PubMed
Close
,
Ashis K. Mitra National Centre for Medium Range Weather Forecasting, Earth System Science Organization, Ministry of Earth Sciences, Noida, India

Search for other papers by Ashis K. Mitra in
Current site
Google Scholar
PubMed
Close
,
Imranali M. Momin National Centre for Medium Range Weather Forecasting, Earth System Science Organization, Ministry of Earth Sciences, Noida, India

Search for other papers by Imranali M. Momin in
Current site
Google Scholar
PubMed
Close
,
D. S. Pai India Meteorological Department, Pune, India

Search for other papers by D. S. Pai in
Current site
Google Scholar
PubMed
Close
,
E. N. Rajagopal National Centre for Medium Range Weather Forecasting, Earth System Science Organization, Ministry of Earth Sciences, Noida, India

Search for other papers by E. N. Rajagopal in
Current site
Google Scholar
PubMed
Close
, and
S. Basu National Centre for Medium Range Weather Forecasting, Earth System Science Organization, Ministry of Earth Sciences, Noida, India

Search for other papers by S. Basu in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

The upgraded version 7 (V7) of the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA) products is available to the user community. In this paper, two successive versions of the TMPA-3B42 research monitoring product, version 6 (V6) and V7, at the daily scale are evaluated over India during the southwest monsoon with gauge-based data for a 13-yr (1998–2010) period. Over typical monsoon rainfall zones, biases are improved by 5%–10% in V7 over the regions of higher rainfall like the west coast, northeastern, and central India. A similar reduced bias is seen in V7 over the rain-shadow region located in southeastern India. In terms of correlation, anomaly correlation, and RMSE, a marginal improvement is seen in V7. Additionally, in all-India summer monsoon rainfall amounts, mean, interannual values, and standard deviation show an overall improvement in V7. Different skill metrics over typical subregions within India show an improvement of the monsoon rainfall representation in V7. Rainfall frequency in different categories also indicates an overall improvement in V7 across all scales and subregions. Over central India regions associated with the monsoon transients, the sign of the bias has changed toward a positive bias. Even if the bias in the frequency of the occurrence of light rain has improved in V7, the values still show a large difference compared to observations. Though both V6 and V7 are able to represent the anomalous dry/wet regions during contrasting monsoon years, V7 shows some improvement in amplitude of those anomalies over V6. In general, V7 has considerably improved over V6 and will continue to be in demand from various sectors of observed rainfall data users.

Corresponding author address: Dr. Ashis K. Mitra, Scientist-F, NCMRWF, ESSO/MoES, A-50, Sector-62, Noida UP-201 309, India. E-mail: ashis.mitra@nic.in

Abstract

The upgraded version 7 (V7) of the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA) products is available to the user community. In this paper, two successive versions of the TMPA-3B42 research monitoring product, version 6 (V6) and V7, at the daily scale are evaluated over India during the southwest monsoon with gauge-based data for a 13-yr (1998–2010) period. Over typical monsoon rainfall zones, biases are improved by 5%–10% in V7 over the regions of higher rainfall like the west coast, northeastern, and central India. A similar reduced bias is seen in V7 over the rain-shadow region located in southeastern India. In terms of correlation, anomaly correlation, and RMSE, a marginal improvement is seen in V7. Additionally, in all-India summer monsoon rainfall amounts, mean, interannual values, and standard deviation show an overall improvement in V7. Different skill metrics over typical subregions within India show an improvement of the monsoon rainfall representation in V7. Rainfall frequency in different categories also indicates an overall improvement in V7 across all scales and subregions. Over central India regions associated with the monsoon transients, the sign of the bias has changed toward a positive bias. Even if the bias in the frequency of the occurrence of light rain has improved in V7, the values still show a large difference compared to observations. Though both V6 and V7 are able to represent the anomalous dry/wet regions during contrasting monsoon years, V7 shows some improvement in amplitude of those anomalies over V6. In general, V7 has considerably improved over V6 and will continue to be in demand from various sectors of observed rainfall data users.

Corresponding author address: Dr. Ashis K. Mitra, Scientist-F, NCMRWF, ESSO/MoES, A-50, Sector-62, Noida UP-201 309, India. E-mail: ashis.mitra@nic.in

1. Introduction

The Indian monsoon is a major component of the South Asian monsoon system, which plays an important role in the earth’s climate system and its water cycle. The southwest monsoon dominates the socioeconomic livelihood over India, including agriculture, water resources, and energy sectors. Through observations, numerical modeling, and use of supercomputers, the scientific community has advanced its understanding and prediction in short, medium, and extended ranges. Weather and climate models are being continuously improved for better monsoon understanding and prediction at different time scales ranging from days to seasons. These models are also crucial when used as climate models to project the future changes in the monsoon at regional and larger scales related to climate change and their uncertainties. For this, there is a need of accurate observed daily rainfall data covering both land and oceanic regions around India. This observed rainfall data could be used to validate or verify weather and climate models and hence provide useful feedback for model development.

Because of recent advancements in sensor techniques, data analysis algorithms, and computing techniques, satellite observations have been showing rapid progress. These satellite data are now being used with confidence for real-time applications by end users (Mitra et al. 2003, 2009). With the availability of rainfall data from the Tropical Rainfall Measuring Mission (TRMM), researchers and model developers for the Indian monsoon got a major boost. Two key microwave (MW) precipitation sensors, the TRMM Microwave Imager (TMI) and the Precipitation Radar (PR), on board TRMM made it the foremost satellite platform to analyze the tropical precipitation and its variability. TRMM is designed to monitor and study the tropical precipitation characteristics and associated latent heating (Kummerow et al. 1998) and has been found very useful for monsoon studies.

The TRMM Multisatellite Precipitation Analysis (TMPA) research monitoring product, one of the most widely used high-resolution merged precipitation products for various hydrometeorological applications (Mitra et al. 2009, 2013; Liu et al. 2012; Wang et al. 2014), takes advantages of the rich constellation of MW and infrared (IR) satellite-borne sensors along with available rain gauge data over land (Huffman et al. 2007, 2010). For the widest usage and applications of this precipitation product, a number of studies have been made in the last few years to validate this product at regional and seasonal scales (Sapiano and Arkin 2009; Villarini 2010; Scheel et al. 2011; Karaseva et al. 2012; Prakash et al. 2012, 2013; Prakash and Gairola 2014). However, there are relatively few studies available for the Indian land region (Rahman et al. 2009; Nair et al. 2009; Uma et al. 2013; Prakash et al. 2014b), which consists of varied topography and rainfall distributions.

Rahman et al. (2009) evaluated the TMPA-3B42 version 6 (V6) rainfall estimates with gridded rain gauge–based datasets from the India Meteorological Department (IMD) at daily scales and noticed that the TMPA product depicts the pattern and variability of rainfall reasonably well at synoptic scales. Nair et al. (2009) independently evaluated this multisatellite rainfall product close to the Western Ghats mountain range in India using high-density rain gauge measurements for the southwest monsoon period of 1998–2004. They noticed that TMPA picks up the west-to-east rainfall gradient along the west coast well, but the rainfall maxima over the Western Ghats are not captured adequately. More recently, Uma et al. (2013) validated daily TMPA-3B42 V6 rainfall products with gridded IMD rain gauge–based data for the period of 1998–2007 and showed that TMPA picks up the major prominent large-scale features of the southwest or summer monsoon rainfall, like flood and drought and active and break spells, reasonably well. However, TMPA shows significant biases that vary across regions and seasons over India. Furthermore, the TMPA dataset performs better than the other contemporary high-resolution multisatellite precipitation products like Climate Prediction Center (CPC) morphing technique (CMORPH), the Naval Research Laboratory (NRL)-Blend technique, and Precipitation Estimation from Remotely Sensed Imagery using Artificial Neural Networks (PERSIANN) over the Indian monsoon region (Prakash et al. 2014b). Any single satellite-based rainfall monitoring product would have sampling errors (Bell et al. 1990). However, the final TMPA product, because of its multisatellite combination and calibration, does not have severe sampling issues.

The TMPA research version precipitation monitoring product version 7 (V7) was released in December 2012 after retrospective processing of the various inputs with the new V7 algorithm. The differences between these two successive versions of TMPA products are mostly over the land and coastal regions, which are mainly attributable to the changes in rain gauge analysis information built into the new V7 algorithm (Huffman and Bolvin 2013). The main causes are as follows.

  1. TMPA V6 used the Global Precipitation Climatology Centre (GPCC), version 2, monitoring analysis from January 1998 to April 2005 and the National Oceanic and Atmospheric Administration (NOAA)/CPC Climate Anomaly Monitoring System (CAMS) analysis from May 2005 to June 2010. However, TMPA V7 uses the latest GPCC, version 4, full analysis from 1998 to 2010 and GPCC, version 6, monitoring analysis thereafter.

  2. TMPA V6 included deficient Advanced Microwave Sounding Unit (AMSU) precipitation estimates from January 2000 to May 2007, particularly from January 2003 to May 2007; the current improved AMSU algorithm is now applied throughout.

  3. The TRMM Combined Instrument (TCI) product showed a modest increase from V6 to V7.

  4. The calibrated MW data are about 3%–5% higher than the TCI calibrator.

More recently, Wang et al. (2014) showed that the new version (V7) of precipitation estimates from TMI and PR are considerably improved compared to V6 estimates. However, the west coast of India and the foothills of the Himalayas are among the regions having still larger uncertainty. The upgraded version of TRMM PR–derived near-surface rainfall (3A25V7) also showed an overall improvement over the South Asian monsoon region as compared to V6 (Prakash et al. 2014a). It is therefore important to evaluate this new version of the precipitation product using a dense network of gauge observations over the Indian monsoon region for a wide range of applications.

Recently, some researchers have quantified the similarities and differences between these two successive versions of the TMPA product (Prakash et al. 2013; Chen et al. 2013a,b; Yong et al. 2014; Zulkafli et al. 2014). Prakash et al. (2013) compared V6 and V7 rainfall products at a monthly scale over the tropical oceans with rain gauge data from the available buoys for 1998–2010 and showed that underestimation of high rainfall by V6 is considerably improved by 5%–8% in V7. The comparison of both versions of the TMPA-3B42 product over China (Chen et al. 2013a; Yong et al. 2014), the continental United States (Chen et al. 2013b), and the Andean–Amazon river basins (Zulkafli et al. 2014) showed that V7 performs better, in general, than V6, with some exceptions. Furthermore, Prakash and Gairola (2014) validated the TMPA-3B42 V7 product over the tropical Indian Ocean from buoy gauge data at daily scales for the period of 2004–11. They noticed that the new version of the TMPA product slightly overestimates rainfall at most of the buoy locations even though it underestimates heavy rain of more than 100 mm day−1 and light rainfall of less than 0.5 mm day−1. In this study, we compare the TMPA-3B42 V6 and V7 products with gridded rain gauge–based data from IMD over India for the southwest monsoon season (June–September), as India receives about 60%–90% of its annual rainfall during this season, which has crucial socioeconomic impacts. The comparison is done at a daily time scale for a 13-yr period (1998–2010).

2. Data used

a. TMPA-3B42 precipitation data

The TMPA dataset is a finescale (0.25° latitude–longitude and 3-hourly resolution) merged multisatellite precipitation product available over a global belt ranging from 50°S to 50°N (Huffman et al. 2007, 2010). The TMPA-3B42 research estimates are produced in four stages: 1) the MW precipitation estimates are calibrated against TCI, which includes TRMM PR; 2) the IR precipitation estimates are created using the calibrated MW precipitation; 3) the MW and IR estimates are combined; and 4) rescaling to monthly gauge data is applied. For this rescaling, the gauge database from GPCC is used. For India, it is seen that, on average, around 260 observations are used during the monsoon season. The TMPA rainfall product is available in post–real time and near–real time, based on calibration by the TCI and TMI precipitation products, respectively. The post-real-time product is also known as the research monitoring product and is available at three distinct time scales, namely, 3-hourly, daily, and monthly. The TMPA V6 precipitation product has gone through major changes, as discussed in the previous section, and after retrospective processing of the various inputs with the new V7 algorithm, the TMPA V7 product was formally released in December 2012. The V6 product was available only until June 2011. Hence, in this study, we used data until the 2010 monsoon season. The TMPA-3B42 research monitoring product V6 data and revised V7 data at the 3-hourly scale used in the present study were obtained from the TRMM website (http://disc2.nascom.nasa.gov/tovas/).

b. Gauge-based rainfall data

For the evaluation of these two successive versions of the TMPA-3B42 precipitation product over India, high-resolution rain gauge–derived daily upgraded gridded rainfall data developed by IMD (Rajeevan and Bhate 2009) are used. These gridded rainfall data are developed using rainfall observations from about 3500 gauge stations (on a typical day) over India (Fig. 1), followed by proper quality-control checks such as data consistency at a specific gauge location, proper care of missing and unexpected anomalous rainfall values, etc. For our study period, it is seen that, on average, about 2300 gauge stations were used in this gridded dataset. These daily station data are interpolated into a regular grid of 0.5° latitude–longitude using a standard Shepard’s optimal interpolation method, which is widely used and is independent of the first-guess field. This gridded rainfall dataset is intercompared with other available standard gauge-based gridded rainfall datasets, namely, the Asian Precipitation–Highly-Resolved Observational Data Integration Towards Evaluation of the Water Resources (APHRODITE), which showed that the IMD dataset has a realistic representation of rainfall observations and is widely used for various meteorological and climatological applications. This IMD rain gauge–based dataset uses many more observations (around 2300 gauges) than are used in TMPA (around 260 gauges). Since the IMD dataset has many more gauges, it is quite appropriate to use this as a reference product for validation and intercomparison for the Indian monsoon period. In spite of the greater number of gauges in the IMD data, there are still some regions that have fewer gauges spread across a greater area (Fig. 1), where some uncertainty is expected in this type of validation study. Since most of the precipitation occurs as rainfall over India, we interchangeably use precipitation and rainfall in this paper.

Fig. 1.
Fig. 1.

Location of IMD rain gauges used for the preparation of the gridded dataset (Rajeevan and Bhate 2009). Boxes A, B, and C are the subregions having dense networks of rain gauges used for detailed analyses.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

3. Methodology

As the IMD rain gauge data computes daily rainfall ending at 0300 UTC (Rajeevan and Bhate 2009), daily rainfall from the TMPA-3B42 is computed by accumulating the available eight 3-hourly estimates ending at 0300 UTC for one-to-one comparison. The reference IMD gauge-based gridded dataset is available at 0.5° resolution. For this intercomparison study, both versions of 3B42 were upscaled to the same resolution of 0.5°. The two versions of the TMPA-3B42 product are compared with IMD gridded rain gauge–based data using visual verification and continuous verification statistics. Visual verification qualitatively compares the maps of satellite-based estimates with ground-truth observations whereas the continuous verification statistics quantitatively measure the accuracy of rainfall. Different statistical parameters like correlation coefficient r, anomaly correlation coefficient (ACC), bias, and root-mean-square error (RMSE) are calculated using Eqs. (1)(4) in order to evaluate the satellite-derived rainfall with observations (Wilks 2006):
e1
e2
e3
and
e4
where xi and yi are the observed (rain gauge) and estimated (TMPA-3B42) rainfall values, x and y are their respective means, and n is the number of matches.
The significance of the regional differences between the TMPA-3B42 and IMD rain gauge data are analyzed using a paired Student’s t test defined as (Shin et al. 2011)
e5
where is the standard deviation of . For a two-tailed Student’s t test using the 1586 pairs (122 days yr−1) of the daily mean rain rates, values larger than 2.58 are significant at the 99% level and values larger than 1.98 are significant at the 95% level.

Indian monsoons exhibit considerable rainfall variability associated with typical monsoon synoptic weather systems and varied topography in the region. Moreover, the spatial distributions of rain gauges are not homogeneous over India. Hence, the evaluation of these two successive versions of TMPA-3B42 rainfall products has been done at the regional scale, where more gauge information is available in the reference IMD data. For this purpose, three boxes—box A (13°–17°N, 74°–78°E), box B (10°–12°N, 77°–80°E), and box C (24°–27°N, 84°–87°E)—have been selected (Fig. 1) associated with typical rainfall regimes and quality-verifying data. Box A is a unique region with varied orography where the western side gets high rainfall because of the forced ascent of the monsoon low-level jet and the eastern side (leeward) gets low or negligible rainfall. Box B is a rain-shadow region where the error associated with satellite-derived rainfall is assumed to be very high because of a large gradient of rainfall. Box C is a plain land region without major orography where monsoon low-pressure systems move and provide considerable rainfall in the monsoon season. Thus, these three regions of India are very good test beds for the evaluation.

4. Results and discussion

a. Spatial comparison of TMPA-3B42 V6 and V7 rainfall data

The mean seasonal rainfall averaged for the southwest monsoon period of 1998–2010 from the IMD rain gauge–based data and the TMPA-3B42 V7 and V6 products are shown in Fig. 2. Prominent rainfall features like high rainfall along the west coast and northeastern and central India and low rainfall over the southeast peninsula (rain-shadow region) and northwestern India are qualitatively well captured by all of these rainfall products. However, the magnitude of rainfall slightly differs among them. In V7, the areal spread of low rainfall over the rain-shadow region (in peninsular India) is slightly larger than V6 and closer to IMD rain gauge–based data. Moreover, V7 shows better agreement with observed data than V6 in terms of spread of higher rainfall areas over northeastern and central India. The regions of higher rainfall along the west coast and eastern India also agree better in V7.

Fig. 2.
Fig. 2.

Spatial distributions of mean rainfall (mm day−1) over India during the southwest monsoon season for the period 1998–2010 from daily IMD rain gauge–based observations and TMPA-3B42 V6 and V7 rainfall products.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

The bias between V7 and V6 and between both versions and the gauge-based data are presented in Fig. 3. TMPA-3B42 V7 shows an increase in rainfall along the west coast of India by about 2–4 mm day−1 as compared to V6. The increase in rainfall is as high as more than 4 mm day−1 over northeastern India in V7, whereas the increase ranges from 1 to 3 mm day−1 over central and eastern India. It may be noted that the enhanced rainfall amounts are over those regions where higher rainfall is usually associated with the monsoon synoptic conditions. The eastern regions (Odisha and Jharkhand states), where most of the rainfall is associated with the monsoon lows and depressions, show enhanced rainfall amounts in V7, which is a positive development in the new V7 dataset. Moreover, the rainfall over the leeward side of the west coast shows a decrease ranging from 1 to 3 mm day−1 with respect to the V6 product, resulting in reduced bias in V7. These new V7 data show reduced rainfall over regions where the rainfall is usually lower during the monsoon. The spatial distributions of bias (Fig. 3) and associated Student’s t score (Fig. 4) in V7 and V6 products with respect to IMD rain gauge–based data show that the overestimation of rainfall over the rain-shadow region by V6 is significantly improved in V7. Although the underestimation of the monsoon rainfall along the west coast of India by V6 is notably improved in V7, TMPA-3B42 V7 still shows less rainfall over this region as compared to the IMD rain gauge–based data. However, V7 shows a significant overestimation of rainfall by 1–2 mm day−1 over central India along 20°N. Moreover, V7 shows an overestimation of the monsoon rainfall over northeastern India that is less in V6.

Fig. 3.
Fig. 3.

Spatial distributions of bias (mm day−1) over India during the southwest monsoon season for the period 1998–2010 from daily IMD rain gauge–based observations and TMPA-3B42 V6 and V7 rainfall products.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

Fig. 4.
Fig. 4.

Spatial distributions of Student’s t score showing the bias in daily TMPA-3B42 V6 and V7 rainfall products with respect to IMD rain gauge–based observations at 99% (|t| > 2.58) and 95% (|t| > 1.98) significance levels over India during the southwest monsoon season for the period 1998–2010.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

Figures 5, 6, and 7 show the spatial distributions of correlation coefficient, ACC, and RMSE of V7 and V6 with respect to observed IMD rainfall data. Both V7 and V6 show similar patterns of correlation coefficient as well as RMSE over India. Higher correlation is observed over central India and along the west coast, whereas lower correlation exists over the rain-shadow region and northern hilly regions (Jammu and Kashmir states) of India. The ACC in Fig. 6 shows the correlations of the daily rainfall anomalies computed with respect to their own mean. The patterns of ACC from both products look similar. However, there is marginal improvement in V7 seen in many regions. This also confirms that both datasets are able to capture the daily monsoon transients, where the rainfall amounts are usually higher (west coast, central, and eastern India). The magnitude of RMSE is larger over the high rainfall regions and smaller over the low rainfall areas, except in southern peninsular India. RMSE is marginally decreased in V7 over the rain-shadow region compared to V6.

Fig. 5.
Fig. 5.

Spatial distributions of the correlation coefficient of daily TMPA-3B42 V6 and V7 rainfall products with respect to IMD rain gauge–based observations over India during the southwest monsoon season for the period 1998–2010.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

Fig. 6.
Fig. 6.

Spatial distributions of ACC of daily TMPA-3B42 V6 and V7 rainfall products with respect to IMD rain gauge–based observations over India during the southwest monsoon season for the period 1998–2010.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

Fig. 7.
Fig. 7.

Spatial distributions of RMSE (mm day−1) of daily TMPA-3B42 V6 and V7 rainfall products with respect to IMD rain gauge–based observations over India during the southwest monsoon season for the period 1998–2010.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

b. Comparison of TMPA-3B42 V6 and V7 rainfall data at all-India scale

In this section, the comparison of daily TMPA-3B42 V6 and V7 rainfall is done at the all-India scale with IMD rain gauge–based data for the southwest monsoon period of 1998–2010. Figure 8 shows the density scatterplots of daily all-India summer monsoon rainfall (AISMR) from V7 and V6 products with respect to IMD rain gauge–based data, which clearly shows that the underestimation of rainfall by V6 is noticeably improved by about 9% in V7. The RMSE shows marginal improvement by about 2% in V7 than V6, whereas the correlation coefficient is maintained at the same value. The higher slope in V7 compared to V6 shows better linear fit of data with IMD rain gauge–based observations. This simple analysis again shows that V7 rainfall is slightly improved over India as a whole when compared to the previous V6 product. The AISMR rainfall for each year during the study period is now compared. Figure 9 illustrates the interannual variations in mean, standard deviation, bias, and slope of AISMR from IMD rain gauge–based, V6, and V7 products. The mean and bias show that V6 underestimates AISMR during each year, whereas it is notably improved in V7. The bias against the IMD rain gauge–based observations is marginally positive during some individual years in V7, which was always negative in V6. The interannual variations of standard deviation and slope also suggest that V7 is closer to IMD rain gauge–based data than V6 at the all-India scale.

Fig. 8.
Fig. 8.

Density scatterplots of TMPA-3B42 (a) V7 and (b) V6 daily AISMR with respect to IMD rain gauge–based observations for the period 1998–2010. Blue dashed line shows the 1:1 line; the equation of linear-fit line, number of matches, correlation coefficient, bias, and RMSE are also given.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

Fig. 9.
Fig. 9.

Interannual variations of (a) mean, (b) std dev, (c) bias, and (d) slope of daily AISMR for the period 1998–2010 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets. Bias in TMPA-3B42 V7 and V6 daily AISMR is computed against IMD rain gauge–based observations.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

Finally, the capability of both versions of the TMPA-3B42 rainfall product is examined at different rainfall rate ranges. The frequency distributions of daily AISMR from IMD rain gauge, V6, and V7 are presented in Fig. 10, which shows that V7 underestimates light rainfall (0.5–2 mm day−1) as compared to gauge-based data. The V6 rainfall data show an overestimation of AISMR ranging from 2 to 6 mm day−1, which is considerably reduced in the V7 product. Moreover, V6 underestimates AISMR by more than 8 mm day−1, which is also improved in the V7 product. This clearly shows that, at all rainfall ranges except very light rain (0.5–2 mm day−1), V7 is superior to V6 in terms of frequency distribution at different categories.

Fig. 10.
Fig. 10.

Frequency distribution of daily AISMR for the period 1998–2010 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

c. Comparison of TMPA-3B42 V6 and V7 rainfall data at regional scale

Figure 11 illustrates the density scatterplots of V7 and V6 products with IMD rain gauge–based data for the three regions of interest within India. Over box A, the negative bias of −16.09% in V6 is reduced to −11.74% in V7, and RMSE is also reduced by about 4% in V7, which depicts that the rainfall estimates are noticeably improved in V7 versus V6 over this region. Similar to box A, considerable improvement in bias and RMSE can be seen in V7 versus V6 over box B. However, RMSE is on the order of 200% in both versions, which clearly indicates the problem associated with satellite-based rainfall estimation methods over this rain-shadow region. Over box C, a bias of −9.10% is observed for V6, whereas the bias is about 3% in V7, indicating an improvement in V7. Overall, an improvement in terms of bias is seen for all three regions. The corresponding improvements in correlation and RMSE are marginal in most regions. From these results, it is seen that the V7 bias gets better, but the results are somewhat equivocal for correlation and RMSE, suggesting that the main improvement in V7 is from the gauge analysis. However, the individual precipitation events are driven by the satellite data.

Fig. 11.
Fig. 11.

Density scatterplots of TMPA-3B42 (left) V7 and (right) V6 daily monsoon rainfall with respect to IMD rain gauge–based data averaged over (a),(b) box A; (c),(d) box B; and (e),(f) box C for the period 1998–2010. Blue dashed lines show 1:1 lines; and correlation coefficient, bias, and RMSE are also given.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

The frequency distributions of area-averaged daily rainfall for the three boxes are shown in Fig. 12 in terms of percentage number matching of points lying in a range. In a box, from each dataset, the number count is computed for a particular frequency range and then converted to a percentage in relation to the total number count in all categories. This is repeated for all three datasets. This procedure is similar to the computations shown in Fig. 10, except for three different regions. Both V6 and V7 show biases in different ranges of rainfall. However, in V7 there is a clear improvement across the scales over V6 for all three regions under consideration. For light rainfall (0.5–2 mm day−1), a larger bias is seen in TMPA V6 and V7 rainfall estimates. However, V7 data have improved slightly in this light rainfall category. The interannual variations of mean seasonal rainfall, its standard deviation, and the bias with respect to IMD rain gauge–based data for the three boxes are shown in Fig. 13. Interestingly, V6 systematically underestimates rainfall over boxes A and C, whereas it overestimates rainfall over box B, except during 2004. The mean and standard deviation also show that the rainfall estimates are notably improved over boxes A and B in V7 versus V6. For box C, a region of passing monsoon transient weather systems, V7 now shows a slightly positive bias (enhanced rainfall) as compared to the earlier negative bias in V6.

Fig. 12.
Fig. 12.

Frequency distribution of daily monsoon rainfall averaged over (a) box A, (b) box B, and (c) box C from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets for the period 1998–2010.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

Fig. 13.
Fig. 13.

Interannual variations of (left) mean, (middle) std dev, and (right) bias of monsoon rainfall averaged over (a)–(c) box A, (d)–(f) box B, and (g)–(i) box C for the period 1998–2010 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets. Bias in TMPA-3B42 V7 and V6 daily rainfall is computed against IMD rain gauge–based observations.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

d. Comparison of TMPA-3B42 V6 and V7 rainfall data during contrasting monsoon years

Monsoon seasonal rainfall shows considerable interannual rainfall variability at all-India and regional scales. There is also a need to verify the model-predicted seasonal rainfall anomalies for the Indian monsoon in operations and research. In this section, TMPA-3B42 V6 and V7 rainfall data are compared with IMD rain gauge–based observations for two pairs of contrasting monsoon years. The mean daily rainfall anomalies for the southwest monsoon seasons of 2002 and 2003 from IMD rain gauge, V7, and V6 data are presented in Fig. 14. These seasonal anomalies are computed with respect to the corresponding mean values of each dataset computed for the period of study. The seasonal rainfall for India as a whole was 81% of the long period average (deficit monsoon) in 2002, whereas it was 102% of the long period average (normal monsoon) in 2003, as reported by the IMD. The spatial patterns of rainfall anomalies are in good agreement with each other qualitatively for both the deficient and normal monsoon seasons of 2002 and 2003, respectively. It is seen that both V6 and V7 are able to represent the large-scale anomalous dry and wet regions during the contrasting monsoon years. The major difference between gauge-based data and satellite-based rainfall products is observed over northeastern India, the west coast, and rain-shadow regions during both years. Moreover, the positive anomaly of rainfall over India is underestimated by the TMPA products quantitatively as compared to gauge-based data in 2003, although the underestimation is rather less in V7 than V6 product. It is noted that the amplitude of anomalous dry/wet regions has improved marginally in V7. Figure 15 shows the time series of daily AISMR for both years, which shows a similar kind of rainfall variability by all three products. Anomalous lower values in July 2002 contributed to the below-normal values for the seasonal monsoon as a whole. The AISMR values are 5.78, 5.82, and 5.31 mm by IMD rain gauge, V7, and V6, respectively, during 2002, whereas they show 7.29, 7.32, and 6.70 mm during 2003. The AISMR values during these two contrasting monsoon years also indicate that V7 is closer to IMD observed values as compared to V6.

Fig. 14.
Fig. 14.

Spatial distributions of mean daily rainfall anomaly (mm day−1) over India for the southwest monsoon seasons of 2002 and 2003 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

Fig. 15.
Fig. 15.

Time series of daily AISMR for 2002 and 2003 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets. Mean AISMR values for the southwest monsoon period are also given in parentheses.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

The second pair of contrasting monsoon years are chosen as 2007 (normal monsoon) and 2009 (deficient monsoon) for the comparison. The seasonal rainfall over India was 106% of the long period average in 2007, whereas it was 78% of its long period average in 2009, as per the IMD report. The monsoon year 2009 was the second drought year in this decade after 2002 and the third-highest deficient rainfall year since 1901. The mean daily rainfall anomalies for the southwest monsoon seasons of 2007 and 2009 from IMD rain gauge, V7, and V6 data are illustrated in Fig. 16. During 2007, the positive rainfall anomaly over southern India and the negative anomaly over central and northern India are very well depicted qualitatively by the three products. However, the TMPA products show larger areal spread of positive anomaly over southern India than gauge-based data in both V6 and V7. It is seen that the amplitude of wet/dry biases has improved in V7. The positive anomalies in the west coast and eastern India have improved in V7. Even for 2009, a dry monsoon season, we see a marginal improvement in V7. The time series of daily AISMR from the three rainfall products for 2007 and 2009 are shown in Fig. 17, which gives similar results as discussed for Fig. 15. Below-normal AISMR during June and the first half of August in 2009 is picked up very well by the three rainfall products. The AISMR values of 8.33 mm in 2007 and 6.20 mm in 2009 by V7 are seen to be closer to IMD observed values during two contrasting years, confirming an improvement in V7 over its predecessor V6.

Fig. 16.
Fig. 16.

Spatial distributions of mean daily rainfall anomaly (mm day−1) over India for the southwest monsoon seasons of 2007 and 2009 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

Fig. 17.
Fig. 17.

Time series of daily AISMR for 2007 and 2009 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets. Mean AISMR values for the southwest monsoon period are also given in parentheses.

Citation: Journal of Hydrometeorology 16, 1; 10.1175/JHM-D-14-0024.1

5. Conclusions

Weather and climate in India are strongly affected by monsoons, and the ecosystem and human society in the country are under strong influence of the monsoon season. To understand the monsoon, accurate rainfall data are essential. Rainfall data retrieved from the TRMM satellite observations have been accumulated for a longer period, and as it is still in orbit, it provides us with continuous data. TRMM data are being used in real-time applications and numerical model verification and development. In this paper, we verified the quality of an upgraded research monitoring version of the TMPA-3B42 data (V7) against its predecessor (V6) with IMD gridded rain gauge–based data over India from 1998 to 2010 of the southwest monsoon season at the daily scale.

Over the typical monsoon rainfall regimes, biases were significantly improved in V7 versus V6 over regions of higher rainfall amounts such as over the west coast and northeastern and central India. A similar reduced bias was seen in V7 over the rain-shadow region of southeastern peninsular India. These improvements in biases were noted at the 95% significance level. A marginal improvement was also seen in V7 versus V6 in terms of different skill metrics like correlation, ACC, and RMSE. Moreover, the mean, interannual values, and standard deviation also showed an overall improvement in V7 data at the all-India scale. Over the central India regions associated with the monsoon transients, the sign of bias has changed toward a positive bias. Additionally, V7 showed an improvement of the monsoon rainfall representation for the typical subregions within India. The regional analyses of these two versions of TMPA-3B42 rainfall products showed that V7 is noticeably better than V6 along the west coast of India and the rain-shadow region of southeastern peninsular India. The RMSE was reduced by about 38% and the overestimation of monsoon rainfall was also improved by about 25% in V7 versus V6 over the rain-shadow region of southeastern peninsular India. Rainfall frequency at different categories also showed that V7 indicates an improvement across all scales and subregions over V6, in general. Even though bias in the frequency of occurrence of light rain had improved in V7, the values still showed a large difference compared to observations. Though both V6 and V7 were able to represent the anomalous dry/wet regions of contrasting monsoon years, V7 showed some improvement in amplitude of those anomalies over V6. Overall, results indicated that V7 is considerably improved over V6 over India and can be used for the monsoon studies and numerical model output verification. This study, along with other studies for different regions of the globe (Prakash et al. 2013; Chen et al. 2013a,b; Yong et al. 2014; Wang et al. 2014; Zulkafli et al. 2014), suggests that, overall, the V7 product is superior to V6. Both V7 and V6 are TRMM products calibrated with gauge values, and now V7 real-time data are also available for real-time applications. These improved V7 products can be used for merging local gauges to produce the final merged products over India (Mitra et al. 2009, 2013).

TMPA-3B42 V6 data were a huge success, and this study reinforces that V7 data are further improved and will continue to be in demand from various sectors. A new satellite project called the Global Precipitation Measurement (GPM) has begun, and its core satellite was launched in February 2014 to obtain global precipitation data with advanced instrument capabilities beyond TRMM. Satellite-derived precipitation data from the GPM still may have estimation errors depending on seasons and/or regions. Thus, this type of validation study will continue to document the reliability of datasets and to keep the user community informed about the quality. Data from upcoming and recent satellites like GPM, Indian National Satellite 3D (INSAT-3D), and Megha-Tropiques and other new types of in situ data from automatic weather stations, automatic rain gauges, and radars will be highly useful for producing the monsoon rainfall data products. In a future study, we plan to undertake the evaluation of V7 data for other seasons of India and particularly for the northeast monsoon of southern India coinciding with winter.

Acknowledgments

The TMPA data from the Goddard Earth Sciences Data and Information Services Center (GES DISC) and the rain gauge–based data from IMD are gratefully acknowledged. Thanks are owed to Dr. G. J. Huffman for helpful discussions and to three anonymous reviewers for invaluable suggestions. This work was undertaken at NCMRWF to meet the goals of the MoES-NERC SAPRISE project on the Changing Water Cycle.

REFERENCES

  • Bell, T. L., Abdullah A. , Martin R. L. , and North G. R. , 1990: Sampling errors for satellite-derived tropical rainfall: Monte Carlo study using a space–time stochastic model. J. Geophys. Res., 95, 21952205, doi:10.1029/JD095iD03p02195.

    • Search Google Scholar
    • Export Citation
  • Chen, S., and Coauthors, 2013a: Similarity and difference of the two successive V6 and V7 TRMM multisatellite precipitation analysis performance over China. J. Geophys. Res. Atmos., 118, 13 060–13 074, doi:10.1002/2013JD019964.

    • Search Google Scholar
    • Export Citation
  • Chen, S., and Coauthors, 2013b: Evaluation of the successive V6 and V7 TRMM multisatellite precipitation analysis over the continental United States. Water Resour. Res., 49, 81748186, doi:10.1002/2012WR012795.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., and Bolvin D. T. , 2013: TRMM and other data precipitation data set documentation. NASA Global Change Master Directory Doc., 40 pp. [Available online at ftp://precip.gsfc.nasa.gov/pub/trmmdocs/3B42_3B43_doc.pdf.]

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

  • Karaseva, M. O., Prakash S. , and Gairola R. M. , 2012: Validation of high-resolution TRMM-3B43 precipitation product using rain gauge measurements over Kyrgyzstan. Theor. Appl. Climatol., 108, 147157, doi:10.1007/s00704-011-0509-6.

    • Search Google Scholar
    • Export Citation
  • 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
  • Liu, Z., Ostrenga D. , Teng W. , and Kempler S. , 2012: 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
  • Mitra, A. K., Dasgupta M. , Singh S. V. , and Krishnamurti T. N. , 2003: Daily rainfall for Indian monsoon region from merged satellite and rain gauge values: Large-scale analysis from real-time data. J. Hydrometeor., 4, 769781, doi:10.1175/1525-7541(2003)004<0769:DRFTIM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Mitra, A. K., Bohra A. K. , Rajeevan M. N. , and Krishnamurti T. N. , 2009: Daily Indian precipitation analyses formed from a merge of rain-gauge with TRMM TMPA satellite derived rainfall estimates. J. Meteor. Soc. Japan, 87A, 265279, doi:10.2151/jmsj.87A.265.

    • Search Google Scholar
    • Export Citation
  • Mitra, A. K., Momin I. M. , Rajagopal E. N. , Basu S. , Rajeevan M. N. , and Krishnamurti T. N. , 2013: Gridded daily Indian monsoon rainfall for 14 seasons: Merged TRMM and IMD gauge analyzed values. J. Earth Syst. Sci., 122, 11731182, doi:10.1007/s12040-013-0338-3.

    • Search Google Scholar
    • Export Citation
  • Nair, S., Srinivasan G. , and Nemani R. , 2009: Evaluation of multi-satellite TRMM derived rainfall estimates over a western state of India. J. Meteor. Soc. Japan, 87, 927939, doi:10.2151/jmsj.87.927.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., and Gairola R. M. , 2014: Validation of TRMM-3B42 precipitation product over the tropical Indian Ocean using rain gauge data from the RAMA buoy array. Theor. Appl. Climatol., 115, 451460, doi:10.1007/s00704-013-0903-3.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., Mahesh C. , Gairola R. M. , and Pal P. K. , 2012: Comparison of high-resolution TRMM-based precipitation products during tropical cyclones in the north Indian Ocean. Nat. Hazards, 61, 689701, doi:10.1007/s11069-011-0055-7.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., Mahesh C. , and Gairola R. M. , 2013: Comparison of TRMM Multi-satellite Precipitation Analysis (TMPA)-3B43 version 6 and 7 products with rain gauge data from ocean buoys. Remote Sensing Letters, 4, 677685, doi:10.1080/2150704X.2013.783248.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., Mitra A. K. , Momin I. M. , Rajagopal E. N. , and Basu S. , 2014a: Agreement between monthly land rainfall estimates from TRMM-PR and gauge-based observations over South Asia. Remote Sens. Lett., 5, 558567, doi:10.1080/2150704X.2014.934401.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., Sathiyamoorthy V. , Mahesh C. , and Gairola R. M. , 2014b: An evaluation of high-resolution multisatellite rainfall products over the Indian monsoon region. Int. J. Remote Sens., 35, 30183035, doi:10.1080/01431161.2014.894661.

    • Search Google Scholar
    • Export Citation
  • Rahman, S. H., Sengupta D. , and Ravichandran M. , 2009: Variability of Indian summer monsoon rainfall in daily data from gauge and satellite. J. Geophys. Res., 114, D17113, doi:10.1029/2008JD011694.

    • Search Google Scholar
    • Export Citation
  • Rajeevan, M., and Bhate J. , 2009: A high resolution daily gridded rainfall dataset (1971–2005) for mesoscale meteorological studies. Curr. Sci., 96, 558562.

    • Search Google Scholar
    • Export Citation
  • Sapiano, M. R. P., and Arkin P. A. , 2009: An intercomparison and validation of high-resolution satellite precipitation estimates with 3-hourly gauge data. J. Hydrometeor.,10, 149–166, doi:10.1175/2008JHM1052.1.

  • Scheel, M. L. M., Rohrer M. , Huggel Ch. , Villar D. S. , Silvestre E. , and Huffman G. J. , 2011: Evaluation of TRMM multi-satellite Precipitation Analysis (TMPA) performance in the central Andes region and its dependency on spatial and temporal resolution. Hydrol. Earth Syst. Sci., 15, 26492663, doi:10.5194/hess-15-2649-2011.

    • Search Google Scholar
    • Export Citation
  • Shin, D. B., Kim J. H. , and Park H. J. , 2011: Agreement between monthly precipitation estimates from TRMM satellite, NCEP reanalysis, and merged gauge–satellite analysis. J. Geophys. Res., 116, D16105, doi:10.1029/2010JD015483.

    • Search Google Scholar
    • Export Citation
  • Uma, R., Lakshmi Kumar T. V. , Narayanan M. S. , Rajeevan M. , Bhate J. , and Kumar K. N. , 2013: Large scale features and assessment of spatial scale correspondence between TMPA and IMD rainfall datasets over Indian land mass. J. Earth Syst. Sci., 122, 573588, doi:10.1007/s12040-013-0312-0.

    • Search Google Scholar
    • Export Citation
  • Villarini, G., 2010: Evaluation of the research-version TMPA rainfall estimate at its finest spatial and temporal scales over the Rome metropolitan area. J. Appl. Meteor. Climatol., 49, 25912602, doi:10.1175/2010JAMC2462.1.

    • Search Google Scholar
    • Export Citation
  • Wang, J. J., Adler R. F. , Huffman G. J. , and Bolvin D. , 2014: An updated TRMM composite climatology of tropical rainfall and its validation. J. Climate, 27, 273284, doi:10.1175/JCLI-D-13-00331.1.

    • Search Google Scholar
    • Export Citation
  • Wilks, D. S., 2006: Statistical Methods in the Atmospheric Sciences.2nd ed. Academic, 648 pp.

  • Yong, B., and Coauthors, 2014: 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? J. Hydrol., 508, 7787, doi:10.1016/j.jhydrol.2013.10.050.

    • Search Google Scholar
    • Export Citation
  • Zulkafli, Z., Buytaert W. , Onof C. , Manz B. , Tarnavsky E. , Lavado W. , and Guyot J. L. , 2014: A comparative performance analysis of TRMM 3B42 (TMPA) versions 6 and 7 for hydrological applications over Andean–Amazon river basins. J. Hydrometeor., 15, 581592, doi:10.1175/JHM-D-13-094.1.

    • Search Google Scholar
    • Export Citation
Save
  • Bell, T. L., Abdullah A. , Martin R. L. , and North G. R. , 1990: Sampling errors for satellite-derived tropical rainfall: Monte Carlo study using a space–time stochastic model. J. Geophys. Res., 95, 21952205, doi:10.1029/JD095iD03p02195.

    • Search Google Scholar
    • Export Citation
  • Chen, S., and Coauthors, 2013a: Similarity and difference of the two successive V6 and V7 TRMM multisatellite precipitation analysis performance over China. J. Geophys. Res. Atmos., 118, 13 060–13 074, doi:10.1002/2013JD019964.

    • Search Google Scholar
    • Export Citation
  • Chen, S., and Coauthors, 2013b: Evaluation of the successive V6 and V7 TRMM multisatellite precipitation analysis over the continental United States. Water Resour. Res., 49, 81748186, doi:10.1002/2012WR012795.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., and Bolvin D. T. , 2013: TRMM and other data precipitation data set documentation. NASA Global Change Master Directory Doc., 40 pp. [Available online at ftp://precip.gsfc.nasa.gov/pub/trmmdocs/3B42_3B43_doc.pdf.]

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

  • Karaseva, M. O., Prakash S. , and Gairola R. M. , 2012: Validation of high-resolution TRMM-3B43 precipitation product using rain gauge measurements over Kyrgyzstan. Theor. Appl. Climatol., 108, 147157, doi:10.1007/s00704-011-0509-6.

    • Search Google Scholar
    • Export Citation
  • 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
  • Liu, Z., Ostrenga D. , Teng W. , and Kempler S. , 2012: 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
  • Mitra, A. K., Dasgupta M. , Singh S. V. , and Krishnamurti T. N. , 2003: Daily rainfall for Indian monsoon region from merged satellite and rain gauge values: Large-scale analysis from real-time data. J. Hydrometeor., 4, 769781, doi:10.1175/1525-7541(2003)004<0769:DRFTIM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Mitra, A. K., Bohra A. K. , Rajeevan M. N. , and Krishnamurti T. N. , 2009: Daily Indian precipitation analyses formed from a merge of rain-gauge with TRMM TMPA satellite derived rainfall estimates. J. Meteor. Soc. Japan, 87A, 265279, doi:10.2151/jmsj.87A.265.

    • Search Google Scholar
    • Export Citation
  • Mitra, A. K., Momin I. M. , Rajagopal E. N. , Basu S. , Rajeevan M. N. , and Krishnamurti T. N. , 2013: Gridded daily Indian monsoon rainfall for 14 seasons: Merged TRMM and IMD gauge analyzed values. J. Earth Syst. Sci., 122, 11731182, doi:10.1007/s12040-013-0338-3.

    • Search Google Scholar
    • Export Citation
  • Nair, S., Srinivasan G. , and Nemani R. , 2009: Evaluation of multi-satellite TRMM derived rainfall estimates over a western state of India. J. Meteor. Soc. Japan, 87, 927939, doi:10.2151/jmsj.87.927.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., and Gairola R. M. , 2014: Validation of TRMM-3B42 precipitation product over the tropical Indian Ocean using rain gauge data from the RAMA buoy array. Theor. Appl. Climatol., 115, 451460, doi:10.1007/s00704-013-0903-3.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., Mahesh C. , Gairola R. M. , and Pal P. K. , 2012: Comparison of high-resolution TRMM-based precipitation products during tropical cyclones in the north Indian Ocean. Nat. Hazards, 61, 689701, doi:10.1007/s11069-011-0055-7.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., Mahesh C. , and Gairola R. M. , 2013: Comparison of TRMM Multi-satellite Precipitation Analysis (TMPA)-3B43 version 6 and 7 products with rain gauge data from ocean buoys. Remote Sensing Letters, 4, 677685, doi:10.1080/2150704X.2013.783248.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., Mitra A. K. , Momin I. M. , Rajagopal E. N. , and Basu S. , 2014a: Agreement between monthly land rainfall estimates from TRMM-PR and gauge-based observations over South Asia. Remote Sens. Lett., 5, 558567, doi:10.1080/2150704X.2014.934401.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., Sathiyamoorthy V. , Mahesh C. , and Gairola R. M. , 2014b: An evaluation of high-resolution multisatellite rainfall products over the Indian monsoon region. Int. J. Remote Sens., 35, 30183035, doi:10.1080/01431161.2014.894661.

    • Search Google Scholar
    • Export Citation
  • Rahman, S. H., Sengupta D. , and Ravichandran M. , 2009: Variability of Indian summer monsoon rainfall in daily data from gauge and satellite. J. Geophys. Res., 114, D17113, doi:10.1029/2008JD011694.

    • Search Google Scholar
    • Export Citation
  • Rajeevan, M., and Bhate J. , 2009: A high resolution daily gridded rainfall dataset (1971–2005) for mesoscale meteorological studies. Curr. Sci., 96, 558562.

    • Search Google Scholar
    • Export Citation
  • Sapiano, M. R. P., and Arkin P. A. , 2009: An intercomparison and validation of high-resolution satellite precipitation estimates with 3-hourly gauge data. J. Hydrometeor.,10, 149–166, doi:10.1175/2008JHM1052.1.

  • Scheel, M. L. M., Rohrer M. , Huggel Ch. , Villar D. S. , Silvestre E. , and Huffman G. J. , 2011: Evaluation of TRMM multi-satellite Precipitation Analysis (TMPA) performance in the central Andes region and its dependency on spatial and temporal resolution. Hydrol. Earth Syst. Sci., 15, 26492663, doi:10.5194/hess-15-2649-2011.

    • Search Google Scholar
    • Export Citation
  • Shin, D. B., Kim J. H. , and Park H. J. , 2011: Agreement between monthly precipitation estimates from TRMM satellite, NCEP reanalysis, and merged gauge–satellite analysis. J. Geophys. Res., 116, D16105, doi:10.1029/2010JD015483.

    • Search Google Scholar
    • Export Citation
  • Uma, R., Lakshmi Kumar T. V. , Narayanan M. S. , Rajeevan M. , Bhate J. , and Kumar K. N. , 2013: Large scale features and assessment of spatial scale correspondence between TMPA and IMD rainfall datasets over Indian land mass. J. Earth Syst. Sci., 122, 573588, doi:10.1007/s12040-013-0312-0.

    • Search Google Scholar
    • Export Citation
  • Villarini, G., 2010: Evaluation of the research-version TMPA rainfall estimate at its finest spatial and temporal scales over the Rome metropolitan area. J. Appl. Meteor. Climatol., 49, 25912602, doi:10.1175/2010JAMC2462.1.

    • Search Google Scholar
    • Export Citation
  • Wang, J. J., Adler R. F. , Huffman G. J. , and Bolvin D. , 2014: An updated TRMM composite climatology of tropical rainfall and its validation. J. Climate, 27, 273284, doi:10.1175/JCLI-D-13-00331.1.

    • Search Google Scholar
    • Export Citation
  • Wilks, D. S., 2006: Statistical Methods in the Atmospheric Sciences.2nd ed. Academic, 648 pp.

  • Yong, B., and Coauthors, 2014: 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? J. Hydrol., 508, 7787, doi:10.1016/j.jhydrol.2013.10.050.

    • Search Google Scholar
    • Export Citation
  • Zulkafli, Z., Buytaert W. , Onof C. , Manz B. , Tarnavsky E. , Lavado W. , and Guyot J. L. , 2014: A comparative performance analysis of TRMM 3B42 (TMPA) versions 6 and 7 for hydrological applications over Andean–Amazon river basins. J. Hydrometeor., 15, 581592, doi:10.1175/JHM-D-13-094.1.

    • Search Google Scholar
    • Export Citation
  • Fig. 1.

    Location of IMD rain gauges used for the preparation of the gridded dataset (Rajeevan and Bhate 2009). Boxes A, B, and C are the subregions having dense networks of rain gauges used for detailed analyses.

  • Fig. 2.

    Spatial distributions of mean rainfall (mm day−1) over India during the southwest monsoon season for the period 1998–2010 from daily IMD rain gauge–based observations and TMPA-3B42 V6 and V7 rainfall products.

  • Fig. 3.

    Spatial distributions of bias (mm day−1) over India during the southwest monsoon season for the period 1998–2010 from daily IMD rain gauge–based observations and TMPA-3B42 V6 and V7 rainfall products.

  • Fig. 4.

    Spatial distributions of Student’s t score showing the bias in daily TMPA-3B42 V6 and V7 rainfall products with respect to IMD rain gauge–based observations at 99% (|t| > 2.58) and 95% (|t| > 1.98) significance levels over India during the southwest monsoon season for the period 1998–2010.

  • Fig. 5.

    Spatial distributions of the correlation coefficient of daily TMPA-3B42 V6 and V7 rainfall products with respect to IMD rain gauge–based observations over India during the southwest monsoon season for the period 1998–2010.

  • Fig. 6.

    Spatial distributions of ACC of daily TMPA-3B42 V6 and V7 rainfall products with respect to IMD rain gauge–based observations over India during the southwest monsoon season for the period 1998–2010.

  • Fig. 7.

    Spatial distributions of RMSE (mm day−1) of daily TMPA-3B42 V6 and V7 rainfall products with respect to IMD rain gauge–based observations over India during the southwest monsoon season for the period 1998–2010.

  • Fig. 8.

    Density scatterplots of TMPA-3B42 (a) V7 and (b) V6 daily AISMR with respect to IMD rain gauge–based observations for the period 1998–2010. Blue dashed line shows the 1:1 line; the equation of linear-fit line, number of matches, correlation coefficient, bias, and RMSE are also given.

  • Fig. 9.

    Interannual variations of (a) mean, (b) std dev, (c) bias, and (d) slope of daily AISMR for the period 1998–2010 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets. Bias in TMPA-3B42 V7 and V6 daily AISMR is computed against IMD rain gauge–based observations.

  • Fig. 10.

    Frequency distribution of daily AISMR for the period 1998–2010 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets.

  • Fig. 11.

    Density scatterplots of TMPA-3B42 (left) V7 and (right) V6 daily monsoon rainfall with respect to IMD rain gauge–based data averaged over (a),(b) box A; (c),(d) box B; and (e),(f) box C for the period 1998–2010. Blue dashed lines show 1:1 lines; and correlation coefficient, bias, and RMSE are also given.

  • Fig. 12.

    Frequency distribution of daily monsoon rainfall averaged over (a) box A, (b) box B, and (c) box C from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets for the period 1998–2010.

  • Fig. 13.

    Interannual variations of (left) mean, (middle) std dev, and (right) bias of monsoon rainfall averaged over (a)–(c) box A, (d)–(f) box B, and (g)–(i) box C for the period 1998–2010 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets. Bias in TMPA-3B42 V7 and V6 daily rainfall is computed against IMD rain gauge–based observations.

  • Fig. 14.

    Spatial distributions of mean daily rainfall anomaly (mm day−1) over India for the southwest monsoon seasons of 2002 and 2003 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets.

  • Fig. 15.

    Time series of daily AISMR for 2002 and 2003 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets. Mean AISMR values for the southwest monsoon period are also given in parentheses.

  • Fig. 16.

    Spatial distributions of mean daily rainfall anomaly (mm day−1) over India for the southwest monsoon seasons of 2007 and 2009 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets.

  • Fig. 17.

    Time series of daily AISMR for 2007 and 2009 from IMD rain gauge–based observations and TMPA-3B42 V7 and V6 datasets. Mean AISMR values for the southwest monsoon period are also given in parentheses.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 531 125 8
PDF Downloads 243 74 6