Performance Evaluation of GPM IMERG Precipitation Products over the Tropical Oceans Using Buoys

Rajani Kumar Pradhan aFaculty of Environmental Sciences, Czech University of Life Sciences Prague, Prague, Czech Republic

Search for other papers by Rajani Kumar Pradhan in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0002-4091-4267
and
Yannis Markonis aFaculty of Environmental Sciences, Czech University of Life Sciences Prague, Prague, Czech Republic

Search for other papers by Yannis Markonis in
Current site
Google Scholar
PubMed
Close
Free access

Abstract

The majority of global precipitation falls in tropical oceans. Nonetheless, due to the lack of in situ precipitation measurements, the number of studies over the tropical oceans remains limited. Similarly, the performance of IMERG products over the tropical oceans is unknown. In this context, this study quantitatively evaluates the 20 years (2001–20) of IMERG V06 Early, Late, and Final products against the in situ buoys’ estimates using the pixel–point approach at a daily scale across the tropical oceans. Results show that IMERG represents well the mean spatial pattern and spatial variation of precipitation, though significant differences exist in the magnitude of precipitation amount. Overall, IMERG notably overestimates precipitation across the tropical ocean, with maxima over the western Pacific and Indian Oceans, while it performs better over the eastern Pacific and Atlantic Oceans. Moreover, irrespective of the region, IMERG sufficiently detects precipitation events (i.e., >0.1 mm day−1) for high-precipitation regions, though it significantly overestimates the magnitude. Despite IMERG’s detection issues of precipitation events over the regions with lower precipitation, it is in good agreement with the buoys in total precipitation estimation. The positive hit bias and false alarm bias are the major contributions to the overall total positive bias. Furthermore, the detection capability of IMERG tends to decline with increasing precipitation rates. In terms of IMERG runs, the IMERG Final product performs slightly better than the Early and Late runs. More detailed studies over the tropical oceans are required to better characterize the biases and their sources.

© 2023 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Rajani Kumar Pradhan, pradhan@fzp.czu.cz

Abstract

The majority of global precipitation falls in tropical oceans. Nonetheless, due to the lack of in situ precipitation measurements, the number of studies over the tropical oceans remains limited. Similarly, the performance of IMERG products over the tropical oceans is unknown. In this context, this study quantitatively evaluates the 20 years (2001–20) of IMERG V06 Early, Late, and Final products against the in situ buoys’ estimates using the pixel–point approach at a daily scale across the tropical oceans. Results show that IMERG represents well the mean spatial pattern and spatial variation of precipitation, though significant differences exist in the magnitude of precipitation amount. Overall, IMERG notably overestimates precipitation across the tropical ocean, with maxima over the western Pacific and Indian Oceans, while it performs better over the eastern Pacific and Atlantic Oceans. Moreover, irrespective of the region, IMERG sufficiently detects precipitation events (i.e., >0.1 mm day−1) for high-precipitation regions, though it significantly overestimates the magnitude. Despite IMERG’s detection issues of precipitation events over the regions with lower precipitation, it is in good agreement with the buoys in total precipitation estimation. The positive hit bias and false alarm bias are the major contributions to the overall total positive bias. Furthermore, the detection capability of IMERG tends to decline with increasing precipitation rates. In terms of IMERG runs, the IMERG Final product performs slightly better than the Early and Late runs. More detailed studies over the tropical oceans are required to better characterize the biases and their sources.

© 2023 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Rajani Kumar Pradhan, pradhan@fzp.czu.cz

1. Introduction

Covering 71% of Earth’s surface and receiving about 78% of global total precipitation, the ocean plays an important role in Earth’s climate system and hydrological cycle (Trenberth et al. 2007; Vargas Godoy et al. 2021). The tropical oceans, which receive a major proportion of total oceanic precipitation, have significant effects on the global radiation budget. Therefore, understanding the amount, rate, and distribution of precipitation on the tropical oceans not only assists in the accurate estimation of the global water cycle and energy fluxes but also enhances our understating of the processes of the global water cycle over the ocean. Thus, the precise quantification of tropical oceanic precipitation is one of utmost research importance.

One of the reasons for poor understanding of and few studies on the tropical ocean is a lack of an observational network over oceans. At the present time, to bridge this important information gap, satellite estimations have been providing an important and promising source of precipitation information for data-scarce regions like mountains and over oceans. Since the launch of Tropical Rainfall Measuring Mission (TRMM) core instrument in 1998, the TRMM Multisatellite Precipitation Analysis (TMPA) has been one of the most widely used satellite precipitation products over the tropics (Huffman et al. 2007). Following its success, the Global Precipitation Measurement (GPM) mission launched its core satellite in early 2014 (Hou et al. 2013; Liu 2016). Thereafter, several studies have evaluated the performance of GPM products, either by validating them using reference gauge observations or by comparing them with TRMM products across a range of climates (Pradhan et al. 2022).

Despite the scarcity of reference datasets, there have been efforts to evaluate the satellite precipitation over the ocean by using radar observations, island gauges, or buoy observations. Many studies have performed quantitative analysis of satellite precipitation over the tropical oceans using the gauge datasets from buoys. Bowman (2005) compared the TRMM precipitation using 25 buoys in the Pacific Ocean and reported the validation challenges in point-area averaging between the satellite and gauges. Sahany et al. (2010) assessed the TMPA with buoy observations for the estimation of the diurnal cycle over the Indian Ocean and found, overall, good agreement between these two products. In addition, Sapiano and Arkin (2009) evaluated the TMPA over the Pacific Ocean using buoys’ observations and revealed an underestimation by TMPA. Prakash et al. (2011) compared the TRMM Microwave Imager (TMI) with the available buoy observations in the Indian, Pacific, and Atlantic Oceans. They found reasonable agreement in precipitation rate between the TMI and buoy observations over the Atlantic Ocean, followed by the Pacific Ocean and the Indian Ocean.

In recent times, there have been limited attempts to assess the performance of IMERG products over the ocean. For instance, Prakash et al. (2018) first evaluated the IMERG Final (IMERG-F) V04 products over the north Indian Ocean against the buoys’ observations from March 2014 to December 2015. They noticed substantial positive bias and false alarms in IMERG estimates despite IMERG’s better improvement compared to the TRMM. Kucera and Klepp (2018) evaluated the IMERG V03 products over the ocean and revealed an underestimation of IMERG compared to the Ocean Rainfall and Ice-Phase Precipitation Measurement Network (OceanRAIN). Similarly, IMERG V05 has been evaluated using OceanRAIN and the Level-3 Dual-Frequency Precipitation Radar (3DPRD) as references and revealed an overall underestimation in IMERG precipitation compared to OceanRAIN, despite an accurate detection of precipitation events (Khan and Maggioni 2019). In addition, IMERG comparison with TMPA and buoy observations over the tropical oceans revealed that IMERG performs best when representing the mean precipitation at the Pacific and Indian Oceans, except for the high-precipitation regions of the Atlantic (Wu and Wang 2019). Evaluation results of IMERG-F V06 monthly estimates with the Pacific Rainfall Database (PACRAIN) atoll daily observations over the ocean also showed an overall overestimation by IMERG (Bolvin et al. 2021). The study also revealed that IMERG tends to underestimate light precipitation and overestimate intense precipitation rates.

It should also be noted that most of the aforementioned studies are based on the TMPA dataset. Thus, so far, very little is known regarding the GPM IMERG products’ performance. In particular, the most recent IMERG version, V06, has not yet been evaluated at a daily scale. In this context, the main objective of the present study is to comprehensively investigate the performance of the IMERG V06B precipitation estimates using observation buoys across the tropical oceans. This will provide important information for the user community, the GPM ground validation group, and algorithm developers, and thus improve its applicability over the remote ocean regions.

The paper is organized as follows. After this introduction, the second section briefly describes the datasets methodological approach used in this study. The third section presents the results, including spatial and temporal evaluation of IMERG. The fourth section discusses the significant findings, along with some limitations of the study, and provides recommendations for future studies. Finally, the fifth section summarizes and draws conclusions about the findings.

2. Data and methods

a. IMERG

IMERG is a gridded precipitation product and algorithm that merges a variety of satellite precipitation estimations from the GPM constellation (Huffman et al. 2015). The algorithm uses the GPM core satellite to intercalibrate and merge individual passive microwave products (Hou et al. 2013; Liu 2016). IMERG implements the PERSIANN-CCS (Hong et al. 2004) and GPROF (Kummerow et al. 2015) algorithms to retrieve precipitation estimates from individual sensors. The product then merges the selected infrared (IR) and passive microwave (PMW) sensors using the Kalman filter-based CMORPH techniques (Joyce and Xie 2011) and produces gridded precipitation products with 0.1° × 0.1° spatial and 0.5-hourly temporal resolution (Huffman et al. 2020). Although IMERG V06 is extended beyond 60°N–60°S (i.e., until 90°N–90°S), it is partially outside of that latitude band (i.e., only limited to non-ice-covered surfaces). Based on the application and availability to the user community, IMERG has Early (IMERG-E), Late (IMERG-L), and Final runs with a latency of 4 h, 12 h, and 3.5 months, respectively (Huffman et al. 2020). Unlike the Early and Late run products, the Final run product is bias corrected with the Global Precipitation Climatology Centre (GPCC) gauge products at monthly scale. In this study, the IMERG V06 Early, Late, and Final run daily products are evaluated over the period of 2001–20 (Table 1).

Table 1.

Summary of the datasets used in this study.

Table 1.

b. Buoys

The in situ data from the Global Tropical Moored Buoy Array (GTMBA) program were used to evaluate the IMERG products over the tropical oceans. This program includes the Tropical Atmosphere Ocean/Triangle Trans-Ocean Buoy Network (TAO/TRITON) in the Pacific (McPhaden et al. 1998), the Prediction and Research Moored Array in the Tropical Atlantic (PIRATA) (Bourlès et al. 2008), and the Research Moored Array for African–Asian–Australian Monsoon Analysis and Prediction (RAMA) in the Indian Ocean (McPhaden et al. 2009). The buoys are equipped with a Y.M. Young R3125 rain gauge and measure precipitation around 3 m above the sea surface. The capacitance type of gauges measures precipitation at 1-min intervals. Later, the daily precipitation rates (in mm h−1) are determined based on the differences in the 10-min accumulation (Serra et al. 2001). The relevant datasets of each of the buoy networks (i.e., TAO, PIRATA, and RAMA) can be accessed from the GTMBA website (https://www.pmel.noaa.gov/gtmba/) or the OceanSITES website (http://www.oceansites.org/data/index.html) (Table 1). An important feature of the buoy observations is that, unlike overland rain gauges, they exclude island orographic and other thermodynamic effects (Bowman et al. 2005; Prakash et al. 2013; McPhaden et al. 2009). The buoy observations provide an independent precipitation source, as IMERG did not use them for the gauge correction over the ocean. According to Serra et al. (2001), errors in buoy estimations due to evaporation, sloshing, sea spray, and so forth are negligible. However, errors from the wind undercatch would be substantial based on the wind speed. Moreover, no additional wind correction is applied to the buoy observations.

For the Indian Ocean, 25 RAMA buoys were selected, with data from 2004 to 2020 used. Among the buoys, seven are situated in the Northern Hemisphere and the remaining are in the Southern Hemisphere. For the Atlantic Ocean, 18 buoys were selected (8 in the northern Atlantic), and the data range from 2001 to 2020. For the Pacific Ocean, 40 buoys were chosen, 22 buoys in the western Pacific and 18 in the eastern Pacific, and the data ranged from 2001 to 2020. The detailed spatial distribution of the buoys over the tropical oceans is depicted in Fig. 1.

Fig. 1.
Fig. 1.

Spatial distribution of mean daily precipitation of buoys and IMERG-F for the period 2001–20 over the tropical oceans.

Citation: Journal of Hydrometeorology 24, 10; 10.1175/JHM-D-22-0216.1

c. Methodology

This study is focused on the tropical oceans, covering an extent of 25°N–S in both the hemispheres (Fig. 1). The selection of the study area is driven by two main reasons: (i) the significance of the region in global water and energy balance, and (ii) the availability of the in situ observational datasets (i.e., the buoys). To compare the grided satellite estimates with point-based buoy measurements, the point–pixel based approach was employed (Xie et al. 2022; Chen et al. 2018). Following the extraction of IMERG pixels (0.1° × 0.1°) overlying the buoy stations, a pairwise time series was prepared for the evaluation. Although buoy datasets are available from 1998 onward, the datasets from 2001 to 2020 were selected for the analysis to match the IMERG period. Due to the lack of continuous time series of the buoy datasets, only the days present both in the buoys’ data and the IMERG were considered for the analysis. Consequently, the sample size for each ocean was different, ranging from 5809 days over the Indian Ocean to 7519 days over the Pacific Ocean. In addition, for the basin-scale comparison, the mean areal precipitation of all the IMERG pixels and buoys falling inside the particular basin is taken into account. To account for spatial heterogeneity in precipitation rate and to better understand the regional influence on IMERG, the Pacific Ocean was divided into eastern Pacific (longitude < 0°) and western Pacific (longitude > 0°). The buoy precipitation rate was converted from intensity rates (mm h−1) to daily precipitation rates (mm day−1) to match the IMERG format, and the evaluation methodology was carried out on a daily time scale.

The standard continuous and categorical metrics were used to quantitatively evaluate the IMERG products against the buoys’ observations. The main continuous metrics include bias [Eq. (1)], mean absolute error [MAE; Eq. (2)], and root-mean-square error [RMSE; Eq. (3)]. The Pearson correlation coefficient [COR; Eq. (4)] was used to estimate the cross-correlation between the two datasets:
Bias=(SiGi)N,
MAE=|SiGi|N,
RMSE=(SiGi)2N, and
COR=[(SiS¯)(GiG¯)]2[(SiS¯)2(GiG¯)2],
where Si and Gi represent the IMERG and buoy-measured precipitation, and S¯ and G¯ refer to their mean precipitation, respectively.
Unlike the continuous skill scores, which were used for precipitation volume estimation, the detection skill scores represent the IMERG’s ability to detect precipitation events. A 2 × 2 contingency table was constructed with four variables: hit (H), miss (M), false alarm (F), and correct negative (CN) for the IMERG–buoy pairs (Table 2). Based on these scores, the three most commonly used categorical metrics, namely the probability of detection (POD), false alarm ratio (FAR), and critical success index (CSI), were calculated [Eqs. (5)(7)]:
POD=HH+M,
FAR=FH+F, and
CSI=HH+M+F.
POD represents the total number of correctly detected precipitation events in the total number of events. FAR is the ratio of the total number of times IMERG detects a false event to the total events. The CSI reflects a more balanced score between the POD and FAR. In this study, a rain/no-rain threshold of 0.1 mm day−1 was used to compute the metrics (values < 0.1 are very low and may be considered noise). This threshold was selected to account for light precipitation events, which make up a significant portion of the total precipitation events. For instance, IMERG-F has 12%, 35%, 32%, and 10% of precipitation events between 0.1 and 1 mm day−1 for the Indian, Atlantic, eastern Pacific, and western Pacific Oceans, respectively. Although the detection threshold is subjective (Behrangi et al. 2012), previous studies have also used a threshold of 0.1 mm day−1 for estimating the detection scores (Tian and Peters-Lidard 2007; Wu et al. 2018). In addition to the main threshold, categorical metrics were also examined with varying thresholds ranging from 0.1 to 1.0 mm day−1. The results demonstrate that while the choice of threshold has a slight effect on the categorical metrics, the overall pattern remains consistent (see Fig. S1 in the online supplemental material).
Table 2.

Contingency table of satellite and buoys precipitation data.

Table 2.
To further assess the precipitation volume error, an error decomposition method was applied (Wang et al. 2022; Tian et al. 2009). This method divides the total bias (e) into three independent constituents: hit bias (h), missed bias (m), and false alarm bias (f). The hit bias, which is generated from the hit events, can be either positive (an overestimate of the volume from the detected events) or negative (an underestimate of the volume from the detected events). On the other hand, the false alarm bias, which comes from the IMERG’s falsely identified events when there is no precipitation detection by buoys, obviously can only lead to positive bias. Similarly, the missed bias can only have negative bias since it arises from the precipitation events that are not detected by the IMERG. It is also noted that the total bias may be smaller than the individual biases since they may cancel each other out, leading to a smaller total error (Wang et al. 2022). Therefore, by decomposing the total error into individual components, we can gain important insight into the main contributors to the total error. This relationship between the individual components and the total error can be represented as follows:
e=hm+f.

3. Results

a. Mean daily precipitation maps

The basic visual inspection of satellite and buoy precipitation is one of the simplest verification methods between the two datasets. The mean daily precipitation for IMERG-F and buoys from 2001 to 2020 is shown in Fig. 1. Even though a perfect match between IMERG-F’s gridded areal precipitation estimates and buoy point measurements is not accomplished, IMERG-F accurately represents the spatial pattern of buoy precipitation across tropical oceans. This is in line with other studies (Pradhan et al. 2022) that found IMERG-F provided a better representation of the spatiotemporal pattern of precipitation than the other satellite estimates across a variety of climatic and regional conditions. IMERG-F estimates are in good agreement with the buoys in the high-precipitation regions, such as the intertropical convergence zone (ITCZ), over the Pacific and the Indian Oceans. Additionally, it also represents well the low-precipitation areas of the eastern Pacific and Atlantic Oceans. However, IMERG-F substantially overestimates precipitation, with the bias varying with location. The mean daily precipitation for buoys ranges from 0 to 11 mm day−1, whereas the corresponding range for IMERG-F is nearly twice as large (0–25 mm day−1). Given that we are comparing the entire tropical oceanic precipitation with very few point measurements, such differences are expected. Moreover, since the buoy point measurements are sparser in spatial distribution compared to IMERG’s complete coverage, it is very likely that buoys could have missed some precipitation from high-precipitation regions due to their absence over those areas.

b. Point–pixel evaluation

The statistical performance of various IMERG products in comparison to buoy precipitation data has been evaluated using several metrics such as bias, RMSE, COR, POD, FAR, and CSI. The daily-scale estimates for each buoy–IMERG pair have been computed separately for different oceanic regions (Fig. 2 and Table 3). All the IMERG runs tend to overestimate the buoy precipitation, (except IMERG-F over the eastern Pacific), though the magnitude varies across oceanic basin. In terms of bias, IMERG-F shows the best performance over the eastern Pacific, with a mean bias of −0.07 mm day−1 (−3%), which is much lower than the mean bias observed at the Indian (1.70 mm day−1; 38%), Atlantic (0.33 mm day−1; 16%), and western Pacific (2.37 mm day−1; 36%) Oceans. The high precipitation of the Indian Ocean (mean 4.43 mm day−1) and western Pacific Ocean (mean 6.67 mm day−1) could be a reason for the overestimation. On the other hand, while the Indian and western Pacific Oceans have the highest RMSE values (Fig. 2b), the Atlantic (395%) and eastern Pacific (351%) Oceans have the highest RMSE values in terms of relative errors (Table 3). Despite IMERG’s lower bias over the eastern Pacific, the higher variability in terms of RMSE indicates the possibility that the positive and negative biases may offset each other, resulting in a low total bias. In addition, a similar pattern is also observed for COR, with the highest values over the western Pacific (0.61) and Indian (0.56) Oceans, followed by the eastern Pacific (0.55) and Atlantic (0.54) Oceans (Fig. 2c). In terms of IMERG runs, IMERG-F products perform slightly better than the IMERG-E and IMERG-L products. This is because, while none of the IMERG runs has any gauge correction over the ocean, there is a slight difference among the runs. The early run only has forward propagation, whereas the late and final runs have both backward and forward propagation. Moreover, the only difference between the late and final runs over the ocean is the period of calibration.

Fig. 2.
Fig. 2.

Evaluation metrics of daily IMERG precipitation over the tropical oceans for the period 2001–20. (a) Bias (mm day−1), (b) RMSE (mm day−1), (c) COR, (d) POD, (e) FAR, and (f) CSI.

Citation: Journal of Hydrometeorology 24, 10; 10.1175/JHM-D-22-0216.1

Table 3.

Volumetric scores (mm day−1) for all the IMERG runs in comparison with buoys at daily scale.

Table 3.

Interestingly, the analysis of detection of precipitation events reveals that the categorical scores are quite opposite to the volumetric scores. All the IMERG runs show good detection capability over the western Pacific (0.86) and Indian (0.88) Oceans, where the volumetric scores are worse. In fact, both the Indian and western Pacific Oceans have similar POD values, with a median greater than 0.85 (Fig. 2d). However, the distribution of values for the Indian Ocean is slightly narrower compared to the western Pacific, indicating IMERG’s high detection capability compared with all the buoys in the Indian Ocean, whereas there is some spatial heterogeneity for the western Pacific (values vary from 0.70 to 0.92). Similar to POD, IMERG has the best FAR over the western Pacific, with the lowest median FAR (0.21), followed by the Indian (0.41), eastern Pacific (0.48), and Atlantic (0.54) Oceans (Fig. 2e). This could possibly suggest that the IMERG’s high detection capability over the Atlantic (i.e., POD > 0.6) comes at the expense of a high false alarm ratio. Furthermore, IMERG has high variability of POD and FAR among the buoys for the Atlantic and eastern Pacific Oceans compared to the Indian and the western Pacific Oceans. Similar results were also shown for CSI, with the highest values over the western Pacific (0.69), followed by the Indian (0.54), eastern Pacific (0.40), and Atlantic (0.39) Oceans (Fig. 2f).

c. Spatial distribution of categorical scores

To better understand the regional performance of IMERG detection skills, we further investigate the spatial distribution of categorical metrics, including the POD, FAR, and CSI, in different regions (Fig. 3). Since the performance difference among the three IMERG data products is negligible, we only present the detection scores of IMERG-F products. In terms of POD spatial distribution, IMERG-F exhibited relatively low variability over the western Pacific (0.7–0.9) and Indian (0.8–0.9) Oceans. However, IMERG-F has demonstrated high variability in its detection capability over the eastern Pacific and Atlantic Oceans, with POD values varying between 0.3 and 0.9. For the eastern Pacific, in the low-precipitation zone along and south of the equator between 95° and 120°W, the POD scores are considerably lower (<0.5). In contrast, buoys located along the high-precipitation band north of the equator have better performance, with POD > 0.7. Similar to the eastern Pacific, POD scores over the Atlantic Ocean follow the high- and low-precipitation regions, with the highest detection scores in the western Atlantic along the equatorial ITCZ band. Moderate detection scores (0.5–0.7) are found throughout the rest of the low-precipitation regions. However, IMERG-F’s lowest values (<0.5) are reported in the southernmost and northernmost low-precipitation regions along the 10° and 22°W meridians, respectively.

Fig. 3.
Fig. 3.

Spatial distribution of precipitation detection metrics across the tropical oceans at daily scale.

Citation: Journal of Hydrometeorology 24, 10; 10.1175/JHM-D-22-0216.1

The majority of IMERG-F grid cells exhibit comparatively better FAR performance over the Indian and western Pacific regions, with most values varying between 0.3 and 0.5. Among them, an outlier is the northernmost Arabian Sea, a low-precipitation region with a FAR of around 0.8, that stands out from the rest of the Indian Ocean (FAR < 0.3). Over the eastern Pacific, most of the IMERG-F grid cells have very high FARs, with values reaching 0.71. In particular, in the low-precipitation region along and south of the equator, FARs range between 0.5 and 0.7. In contrast, to the north of the equator, along the high-precipitation ITCZ band, IMERG-F demonstrates good performance, with FAR values < 0.5. Over the Atlantic Ocean, FAR shows high variability, varying from 0.29 to 0.89. Although no particular spatial pattern exists for the distribution of FAR, similar to the POD, the low FAR values are observed in the high-precipitation regions at the western Atlantic, whereas the worst performance (>0.8) is reported over low-precipitation regions. The analysis of the regional patterns further confirms that IMERG-F has better detection ability and fewer false alarms in the high-precipitation regions than in the low-precipitation regions.

The CSI results are quite similar to those of the POD. IMERG-F exhibits notable variability in its performance over the eastern Pacific and the Atlantic Ocean, with scores varying between 0.1 and 0.7. In fact, in the low-precipitation zone along and south of the equator in the eastern Pacific, no IMERG-F pixel has a CSI greater than 0.5, which indicates a generally lower level of detection skill. This also indicates that the low POD and excessive FAR in this region reduce the CSI. In contrast, over the northern high-precipitation regions between 5° and 10°N latitude, IMERG-F has a moderate CSI (0.5–0.7). Similar to the POD, the high-precipitation regions in the western Atlantic show good detection (>0.5), whereas the worst (<0.3) is observed in low-precipitation regions. On average, the northern Atlantic has comparatively better CSI values than the southern Atlantic. This pattern could be attributed to the greater extent of the low-precipitation region over the southern Atlantic relative to the North Atlantic.

d. Spatial distribution of volumetric scores

The spatial distribution of bias varies from −1.10 to 4.68 mm day−1, indicating an overall and more pronounced overestimation of precipitation than underestimation throughout the tropical oceans (Fig. 4). The best performance of IMERG-F is observed over the Atlantic and eastern Pacific Oceans, with bias varying between −1.10 and 1.01 mm day−1. Despite the similar bias range between the Atlantic and eastern Pacific Oceans, most of the buoys (17 of 18) over the Atlantic show positive bias, indicating an overall overestimation by IMERG-F. However, in the case of the eastern Pacific, bias tends to be the other way around, that is, most of the buoys (11 of 18) show negative bias, revealing an overall underestimation by IMERG-F. Except for one buoy (bias = 1.01) that lies in the northern Atlantic (high-precipitation ITCZ bands) between 2° and 8°N latitude, IMERG-F shows consistently low bias with values < 0.8 mm day−1. In contrast, bias values in the Indian Ocean display more variability, ranging up to 4 mm day−1, with comparatively less bias over the Arabian Sea than the Bay of Bengal, in line with previous findings using Ocean Moored Buoys in the Northern Indian Ocean (OMNI) buoys (Prakash et al. 2018). This difference may be explained by the difference in precipitation distribution across the basins, with the Bay of Bengal generally receiving more precipitation than the Arabian Sea. On the other hand, IMERG-F exhibits the highest overestimation over the western Pacific, with most of the grid cells having a bias between 2 and 4 mm day−1. The buoys, in particular those located between the 130° and 160°E longitude, have a bias of >2 mm day−1.

Fig. 4.
Fig. 4.

Spatial distribution of volumetric metrics across the tropical oceans at a daily scale.

Citation: Journal of Hydrometeorology 24, 10; 10.1175/JHM-D-22-0216.1

Unlike the bias, the spatial distribution of RMSE values across the tropical oceans exhibits considerable variability, especially over the Atlantic and eastern Pacific Oceans (Fig. 4). The IMERG-F dataset shows comparatively better RMSE performance over the Atlantic and eastern Pacific Oceans than in the Indian and western Pacific Oceans. The worst performance is yielded over the western Pacific, with most of the IMERG-F grid cells reporting RMSE values of >20 mm day−1. Likewise, IMERG-F also exhibits a moderately high RMSE over the Indian Ocean, with most locations having RMSE values of >10 mm day−1. On the other hand, in the eastern Pacific, IMERG-F exhibits lower RMSE (<10 mm day−1) over the low-precipitation region along and south of the equator. In contrast, in the high-precipitation region north of the equator, RMSE values range from 5 to 25 mm day−1. Unlike the other oceanic regions, IMERG-F presents the highest variability over the Atlantic, with RMSE varying from 0.63 to 26.34 mm day−1. In fact, most of the IMERG-F grid cells in the southern Atlantic (low-precipitation region) have comparatively lower RMSE (<10 mm day−1) than their corresponding northern counterparts. Especially along the high-precipitation region between 0° and 12°N latitude, IMERG-F has the highest RMSE, with values greater than 15 mm day−1, whereas the low-precipitation regions between the 15° and 20°N/S latitude in both hemispheres have the lowest RMSE (<5 mm day−1). Nonetheless, when it comes to relative RMSE values (i.e., standardized by their means), in particular, over the Atlantic and eastern Pacific, the regions with low-precipitation exhibit the highest values compared to the high-precipitation regions (Fig. S2).

On the other hand, the spatial distribution of correlation between the IMERG-F and buoy estimates is quite diverse. Furthermore, unlike bias and RMSE, it does not exhibit any particular spatial pattern (Fig. 4). The correlation values range from 0.2 to 9.78. Not surprisingly, the eastern Pacific exhibits the greatest variability, with values ranging from <0.3 to >0.7. On the other hand, the Atlantic Ocean has values between 0.3 and 0.7. IMERG-F pixels along the 30°–40°W longitude regions, in particular, have a low correlation, with values < 0.5. For the Indian Ocean, the majority of the IMERG-F pixels (11) have a correlation of >0.6, with most (8) located south of the equator. Correlation values <0.5 are predominantly observed in the IMERG-F pixels in the regions between 1°S and 10°N. In contrast to the other oceanic basins, IMERG-F pixels across the western Pacific Ocean exhibit strong correlation with the buoy-recorded precipitation, with most having values between 0.6 and 0.7. These results are consistent with previous findings on buoy and CMORPH correlation values, which have been reported to range between 0.5 and 0.7 (Xie et al. 2017).

One of the main reasons for the high variability of IMERG-F’s performance over the Atlantic and eastern Pacific Oceans compared to the Indian and western Pacific Oceans could be attributed to the spatial distribution of precipitation patterns in these regions. The western Pacific and Indian Oceans exhibit a more extensive high-precipitation zone, resulting in buoys being located within homogeneous precipitation regions (Fig. 1). In contrast, the high-precipitation zone of the eastern Pacific and Atlantic Oceans is narrower and limited to the ITCZ region, leaving out more regions on either side of the equator with very low precipitation. A substantially greater number of buoys are located in the low-precipitation regions than in the high-precipitation ITCZ regions, leading to higher variability in IMERG-F’s performance in these regions.

e. Basin-scale evaluation

In this section, the spatial mean precipitation of IMERG-F grid cells for each oceanic region and their corresponding buoys are compared for each oceanic basin (Fig. 5). IMERG-F has better agreement with buoy precipitation over the Atlantic and eastern Pacific Oceans than over the Indian and western Pacific Oceans. For the Indian Ocean, IMERG-F tends to substantially overestimate buoy precipitation, with high-density peaks above the 1:1 line. The notable overestimation of the higher-precipitation magnitudes could have caused the comparatively high RMSE and MAE. For the Atlantic and eastern Pacific, IMERG-F and buoy precipitation are in good agreement and are comparatively less scattered than in the Indian and western Pacific Oceans. Especially for higher magnitudes, IMERG-F shows comparatively less overestimation than in the Indian and western Pacific Oceans. On the other hand, the higher-density precipitation peaks along the 1:1 line imply a better correlation for the western Pacific. Similar to the Indian Ocean, the high RMSE and MAE of IMERG-F over the western Pacific Ocean could be attributed to the relatively high mean precipitation intensities.

Fig. 5.
Fig. 5.

Scatter density plot between IMERG-F and buoys’ daily precipitation (mm day−1) over the tropical oceans (the red line denotes the quantile–quantile matchups).

Citation: Journal of Hydrometeorology 24, 10; 10.1175/JHM-D-22-0216.1

Furthermore, the evaluation metrics significantly improved for all the oceanic basins compared to the IMERG-F–buoys individual point–pixel errors. For instance, the RMSE values for the Indian, Atlantic, eastern Pacific, and western Pacific are reduced from 16.1, 8.3, 9.08, and 18.52 from the point–pixel scale to 8.83, 3.65, 3.06, and 7.2 for the basin scale, respectively. Considering the smoothing effect of spatial aggregation, it is generally expected that the agreement between the regional mean precipitation is better than the individual IMERG-F’s point–pixel agreement (Behrangi and Wen 2017). This is also in line with other findings (Tan and Santo 2018; Wang et al. 2017), which also confirmed the positive effect of scale on satellite errors. Nonetheless, the pattern of IMERG-F performance remains similar to the previous results, that is, IMERG-F’s best performance is over the Atlantic and eastern Pacific Oceans, followed by the Indian and western Pacific Oceans.

f. Frequency

To better understand the IMERG-F performance in terms of representing the frequency of different precipitation rates, the total precipitation is divided into various intensities (Fig. 6). Overall, IMERG-F underestimates the occurrence frequency of no-rain days, compared to the buoys, throughout the tropical oceans, although the extent of underestimation varies among the regions. The underestimation is greatest for the Atlantic Ocean (10%), followed by the eastern Pacific (5.2%), Indian (5%), and western Pacific (3%) Oceans. Similar results were also previously reported for IMERG-F by Prakash et al. (2018) over the Indian Ocean, though the reference data were different (i.e., they were from OMNI buoys). Even though the rain/no-rain threshold of 0.1 mm day−1 could probably influence the results, the primary reason could be attributed to the point–pixel comparison (Behrangi et al. 2012). For instance, as IMERG-F pixels represent around 11 km × 11 km over the equator, it is highly probable that the light rainfall could occur anywhere in the grid and probably be missed by the buoys if it does not occur precisely over the buoy location. Consequently, buoys can record a higher number of nonrainy days than can the IMERG-F.

Fig. 6.
Fig. 6.

IMERG-F capability in reproducing the occurrence frequency of different precipitation intensities across the tropical oceans.

Citation: Journal of Hydrometeorology 24, 10; 10.1175/JHM-D-22-0216.1

Furthermore, except in the Indian Ocean, IMERG-F tends to overestimate the light precipitation (0.1–2 mm day−1) throughout the tropical oceans. Especially, in the Atlantic and eastern Pacific, the overestimation of light precipitation is more pronounced than in the western Pacific Ocean. This is due to the frequency distribution over the Atlantic and eastern Pacific Oceans, which is right-skewed, indicating the dominance of light precipitation. On the other hand, IMERG-F’s overestimation of the heavy precipitation events (>10 mm day−1) is more pronounced over the Indian and western Pacific Oceans than the Atlantic and eastern Pacific Oceans. IMERG-F’s overestimation of heavy precipitation was also reported over the Indian (Prakash et al. 2018) and Pacific (Bolvin et al. 2021) Oceans. Again, this is due to the higher fraction of heavy precipitation over the Indian and western Pacific Oceans. Moreover, this may be the one probable reason for IMERG-F’s overall overestimation of precipitation over the Indian and western Pacific Oceans. Finally, IMERG-F tends to slightly underestimate moderate precipitation (3–5 mm day−1) in all basins, except for the Atlantic, where it agrees with the buoys.

g. Error decomposition

To further understand the IMERG estimation errors, the total bias is divided into hit bias, missed bias, and false alarm bias (Fig. 7). Unless a specific IMERG run is mentioned, the results reported here refer to all three IMERG runs. Overall, the positive hit bias and false alarm bias are the leading contributors to the total error. Across all oceanic basins, IMERG exhibits a total positive hit bias, with a maximum over the Indian Ocean (IMERG-F = 36%), followed by the western Pacific (IMERG-F = 32%), and the Atlantic (IMERG-F = 10%) Oceans. However, the eastern Pacific is an exception, as the IMERG-F shows a slight negative hit bias (−2%). This indicates that, except for the eastern Pacific, IMERG predominately overestimates precipitation magnitude from correctly detected events. In terms of error contribution, similar to the previous results, the Indian and western Pacific data have shown a more or less similar error pattern, and so have the Atlantic and eastern Pacific data. Even though the Indian and western Pacific data have a similar total bias (>30%), a significant difference exists in the contributions from individual errors: the false alarms and missed biases over the Indian Ocean are almost twice that of the western Pacific. Nonetheless, the cancellation among the positive hit bias, negative missed bias, and positive false alarm bias leads to a total positive bias similar to the western Pacific.

Fig. 7.
Fig. 7.

Decomposition of the IMERG total bias into hit bias, missed bias, and false-alarm bias (shown as percentages relative to their corresponding buoy-measured precipitation).

Citation: Journal of Hydrometeorology 24, 10; 10.1175/JHM-D-22-0216.1

The error components show a similar pattern of contribution over the Atlantic Ocean, with hit bias and false alarm bias being the major contributors. In contrast, the eastern Pacific Ocean exhibits a different behavior for IMERG-F. Unlike the other oceanic basins, where the hit bias is the major contributor to the total error, the eastern Pacific region shows that false alarm bias (7%) and missed bias (−2%) are the major contributors of IMERG-F’s total bias. For IMERG-E and IMERG-L, hit bias contributes relatively more than other biases, consistent with the pattern observed in other oceanic basins. Moreover, here, all the IMERG runs have greater number of false alarms and missed bias compared to the Indian and western Pacific Oceans. Since the accurate detection of light precipitation has remained a challenging task for the satellite estimates (Li et al. 2021), the dominance of light precipitation over the eastern Pacific and Atlantic Oceans can be attributed to the greater number of false alarms and missed bias over these regions. Moreover, the final IMERG runs outperform the early and late runs, whereas the differences between the latter two are negligible.

h. IMERG performances at extremes

The performance of IMERG-F in detecting different rainfall intensities is also assessed against that of the buoys throughout the tropical oceans (Fig. 8). In terms of POD, IMERG-F performs reasonably well in detecting precipitation events until the 25th percentile (values > 0.8). However, the detection capability of IMERG-F decreases gradually with increasing percentiles. Notably, a clear distinction can be observed between the Indian and western Pacific, and Atlantic and eastern Pacific Oceans. This indicates IMERG-F’s relatively better precipitation detection over the western Pacific and Indian Oceans than over the Atlantic and eastern Pacific Oceans. Additionally, POD scores fall below 0.5 for the Atlantic and eastern Pacific when the precipitation reaches the 75th percentile. For the Indian Ocean, the POD scores remain greater than 0.5 beyond the 95th percentile, and until the 99th percentile for the western Pacific. These findings suggest that IMERG-F performs better at detecting extreme precipitation events over the western Pacific and Indian Oceans compared to the eastern Pacific and Atlantic Oceans.

Fig. 8.
Fig. 8.

Detection capability of IMERG-F at different rainfall thresholds over the tropical oceans.

Citation: Journal of Hydrometeorology 24, 10; 10.1175/JHM-D-22-0216.1

In the case of FAR, precipitation between the 1st and 10th percentiles exhibits the highest variability among the oceanic basins, with the eastern Pacific having the highest FAR, followed by the Atlantic, western Pacific, and the Indian Oceans. However, above the 25th percentile, IMERG-F follows a similar pattern for all the oceanic basins, with a steady increase in FAR with percentile, though the increase for the eastern Pacific is somewhat lower. Subsequently, IMERG-F shows higher false alarms for precipitation percentiles exceeding 75, with FAR values greater than 0.5. This indicates that the capability of IMERG-F to detect extreme precipitation comes at the cost of higher numbers of false alarms, especially for the Indian and western Pacific Oceans. In contrast, for the eastern Pacific, FAR remains below 0.5 until the 90th percentile, indicating that IMERG-F exhibits fewer false alarms for extreme precipitation events in this region than in other oceanic basins.

The CSI scores are also very similar to the POD, at least until the 25th percentile, with the western Pacific performing the best, followed by the Indian, Atlantic, and eastern Pacific. In addition, for the 1st–10th percentiles, the eastern Pacific has a lower CSI, with values around 0.75. However, starting from the 25th percentile, all oceanic basins have a similar pattern: a gradual decline in CSI with percentiles. The values also fall below 0.5 when the precipitation reaches the 75th percentile. Moreover, the CSI values are below 0.25 for all the oceanic basins for the 95th–99th percentiles. Nonetheless, the eastern Pacific displays a slightly better overall CSI score than other basins despite having the lowest POD for high percentiles. This could be due to the IMERG-F’s fewer false alarms for the higher percentiles over the eastern Pacific than the other basins.

4. Discussion

Overall, IMERG-F captures the spatial properties of oceanic precipitation well. Still, some substantial biases appear. Our results indicate that the biases associated with hit and false alarm events are the major contributors to the total error over the tropical oceans. Notably, the hit bias is the leading cause of the total positive bias observed over high-precipitation regions, such as the Indian and the western Pacific Oceans. However, over the eastern Pacific, and especially for the IMERG-F, false alarm bias (−5%) is slightly higher than the hit bias (−2%). This is due to the excessive overestimation of the light precipitation (0.1–2 mm day−1) events over the Atlantic and eastern Pacific Oceans (Fig. 6). Therefore, despite the higher number of false alarms, IMERG-F better estimates the total precipitation over the Atlantic and eastern Pacific Oceans. On the other hand, the overestimation of extreme precipitation frequencies is the main reason for significant overestimation observed over the high-precipitation regions.

IMERG’s issues with overestimation (underestimation) of light (heavy) precipitation are not uncommon and have also been observed in other studies (Ehsani et al. 2022; Li et al. 2022; Tan et al. 2021). In particular, the overestimation of light precipitation has been reported over coastal regions (Derin et al. 2022) and land (Kazamias et al. 2022; Su et al. 2018), in general, especially in arid and semiarid climates where light precipitation often evaporates before reaching the surface. IMERG tends to overdetect the frequency of such light precipitation events in these regions (Guo et al. 2016; Tang et al. 2016). Additionally, despite the point–pixel measurement mismatches, the occurrence of virga (i.e., precipitation that evaporates before reaching the surface) may contribute to IMERG’s overestimation of light precipitation over the tropical oceans (Prakash et al. 2018). Additionally, buoy underestimation due to wind undercatch error, which is more prevalent in light precipitation events, may also be a contributing factor.

One probable reason for the observed biases in precipitation estimates could be the lack of appropriate gauge density to accurately represent the small-scale precipitation. For instance, sometimes extreme or light events are very small in horizontal scale and cannot be captured by the sparse gauge networks, leading to an underestimation of the actual precipitation by the gauges. Another potential explanation could be due to tropical cyclone–related extreme events. For instance, it can also be possible that the high wind speed associated with tropical cyclones further exacerbated the undercatching issue of oceanic buoys. As a result, when compared to the buoys, IMERG may appear to overestimate precipitation rates and total precipitation. This is further confirmed by Prakash et al. (2018), who found that IMERG consistently reported higher amounts of precipitation during three tropical cyclones over the Indian Ocean.

Furthermore, a decreasing trend in IMERG detection scores was also found with an increasing precipitation threshold (Fig. 8). This can be partially attributed to the fact that when the threshold increases, the total number of precipitation events (H + M + F) decreases and is compensated for by an increase in the number of nonprecipitation events or correct negatives (Table S1). Consequently, the number of hits decreases compared to the number of misses and leads to an overall decreasing POD. Similarly, in the case of FAR, the higher number of false alarms than the number of hits leads to an overall higher FAR. The lower detection scores and higher FAR with increasing precipitation threshold can also be due to the inability of the satellite to detect the exact precipitation threshold. In this case, although IMERG may not detect events within the given threshold, it did detect events with a little more or less from the given thresholds, resulting in lower detection scores but contributing to the total amount. These results are in line with those of Manz et al. (2017, using IMERG V05) and Wu et al. (2019), though the study regions are over the tropical Andes and Yangtze River basin in China, respectively. Similar results were also reported by Rojas et al. (2021) over the mountain regions of south-central Chile, and by Retalis et al. (2018, using IMERG V04A) over Cyprus.

An overall overestimation of IMERG V06 monthly estimates was also reported by Bolvin et al. (2021) over the Pacific Ocean. In contrast, Khan and Maggioni (2019) found an overall underestimation by IMERG V05. However, they evaluated IMERG against OceanRAIN, and the results were applied to the entire global ocean instead of specific regions. Similarly, IMERG V06 underestimation of precipitation rate compared with the radar precipitation was also reported over the Kwajalein region of the central Pacific (Wang et al. 2022). In addition, our findings show good agreement with the overestimation of the IMERG-F product over the Pacific Ocean but disagree with the underestimation reported for the Atlantic and Indian Oceans by Wu and Wang (2019). Those authors state that the IMERG-E product aligns well with buoys, although their study was limited to 3 years (April 2014–17) and focused on IMERG V05.

However, most of the abovementioned IMERG underestimations over the ocean were for the previous versions (i.e., earlier than IMERG V06). Compared with the previous versions, IMERG V06 has introduced several changes, for example, in the intercalibration, Kalman filter, and morphing system, etc. (Tan et al. 2019a,b). Although all of these changes are intended to improve IMERG V06 even further, the possibility of causing, instead, an overall overestimation over the tropical oceans cannot be ignored. In fact, Wang et al. (2022) reported an increase in FAR over the central Pacific Ocean in IMERG V06 compared to IMERG V05. This remains an open question as a comparison with the previous IMERG versions is not part of the present study, which focuses on the evaluation only of the last IMERG version. However, the IMERG V06 overestimation is most likely due to the implementation of a new microwave satellite intercalibration scheme, as well as the discovery of swath-dependent precipitation biases in satellite microwave estimates. It is expected that these issues will be addressed in the upcoming IMERG V07. Moreover, the Scheme for Histogram Adjustment with Ranked Precipitation Estimates in the Neighborhood (SHARPEN) is formulated to preserve precipitation rates (Tan et al. 2021), which, in general, is smoothed by the averaging of precipitation fields by the Kalman filter (Rajagopal et al. 2021). Ground validation has shown that the SHARPEN scheme improves performance and increases detection skills, suggesting that its implementation in the IMERG V07 will further reduce false alarms.

Beyond this, we could mention some additional limitations of our study, which pave the road to future research. To begin with, the buoy data used in the analysis were not subject to any wind correction. Since the issue of undercatch in the buoy measurements is well documented in past studies (Wu and Wang 2019; Serra and McPhaden 2003, 2004), it is assumed that the buoys would have slightly underestimated actual precipitation. Even though no specific wind correction formulas are available for the R. M. Young capacitance gauge mounted over open oceanic buoys, few studies have estimated wind correction for other gauges over land (Koschmieder 1934; Yang et al. 1998). For example, Koschmieder (1934) proposed a polynomial equation based on wind speed and found that gauges underestimated precipitation by 10%, 50%, and 70% for wind speeds of 4, 12, and 16 m s−1, respectively. Similar results were also observed by Yang et al. (1998) for the standard National Weather Service gauges. Building upon the Koschmieder (1934) method, Serra et al. (2001) suggested that wind-induced undercatch errors could range from 10% to 50% for the Autonomous Temperature Line Acquisition System (ATLAS) buoys depending on wind speed. In another study, Serra and McPhaden (2003) also reported an increase in monthly percent time of rain (10%), conditional rain rate (14%), and rain accumulation (24%) after applying wind undercatch correction to the ATLAS buoys. Nonetheless, wind undercatch bias is a complex process influenced by various factors, such as wind speed, rainfall intensity, and drop-size distribution (Nešpor and Sevruk 1999), which were not considered by Koschmieder (1934). Although specific information on wind underestimation for the R.M Young capacitance gauge over open oceanic buoys is limited, these studies provide insights into the general effects of wind on gauge measurements. Further research and expert input may be necessary to address this specific issue.

Therefore, applying a wind correction factor to the buoys will reduce noise, which will further improve the IMERG evaluation by reducing the uncertainties associated with the validation results. In addition, it will help identify the main source of error in the IMERG precipitation and its spatial distribution throughout precipitation regimes. Moreover, an appropriate wind correction would provide greater insight into IMERG error characterization, especially for errors associated with light precipitation events. By considering other sources of independent reference precipitation, such as OceanRAIN (Klepp et al. 2018), radar (e.g., Kwajalein polarimetric S-band weather radar), and acoustic rain gauges, could be a means of further confirming the error characteristics of IMERG. This will provide a more accurate representation of the oceanic precipitation and will help address the uncertainties and limitations associated with any single dataset.

In addition, another important limitation could stem from the point- and pixel-based comparison approach (Tian et al. 2018). These discrepancies in the point–pixel approach can lead to a higher false alarm ratio and over- or underestimation of the light and extreme precipitation events. In particular, OceanRAIN would aid in evaluating the IMERG performance on a subdaily scale and is based on precipitation types, that is, stratiform and convective. Nonetheless, while OceanRAIN would be an ideal choice for IMERG evaluation, the ship’s motion poses limitations on the number of samples that can be obtained at a specific IMERG pixel. In particular, the ship travels at a speed of approximately 15 km h−1 (Bumke et al. 2019), resulting in around 3°–4° of latitude coverage in a single day. Furthermore, the spatial coverage of OceanRAIN could be another constraint. For instance, it has limited coverage of the Indian Ocean. Another future direction could be error decomposition and investigating the main source of errors. Moreover, a seasonal evaluation of IMERG will provide further insights into how a buoy’s location and high precipitation ITCZ affect IMERG performance.

5. Conclusions

This study is focused on the tropical oceans and aims to evaluate the IMERG precipitation against the in situ buoys’ precipitation. From this perspective, we quantitatively evaluated the IMERG-E, IMERG-L, and IMERG-F precipitation products with the buoy observations for a common overlapping period of 2001–20 on a daily scale. This evaluation was carried out through both point–pixel and basin-scale approaches. The main conclusions of the study are summarized as follows:

  • Overall, IMERG represents well the spatial pattern of mean daily buoy precipitation throughout the tropical oceans (Fig. 1). However, at the same time, it has shown significant overestimation in total precipitation estimation with its magnitude varying with precipitation regimes.

  • IMERG is better at detecting the occurrence of precipitation over the high-precipitation regions, such as the western Pacific and the Indian Oceans, than the low-precipitation regions of the Atlantic and eastern Pacific Oceans (Fig. 2).

  • The opposite is true in terms of volumetric scores (Fig. 2). IMERG estimates precipitation over the Atlantic and eastern Pacific better than the Indian and western Pacific. This suggests that despite IMERG’s ability to correctly detect precipitation events over the western Pacific and the Indian Oceans, it tends to significantly overestimate precipitation amounts over these two regions. Moreover, IMERG shows a higher estimation error over the high-precipitation regions than the low-precipitation regions, irrespective of the oceanic regions (Fig. 4). However, over the Atlantic and eastern Pacific, relative errors exhibit an opposite trend: high-precipitation regions tend to have lower relative errors compared to low-precipitation regions (Fig. S2).

  • IMERG shows excessive overestimation of the frequency of light precipitation events (0.1–1 mm day−1) over the Atlantic and eastern Pacific Oceans, and heavy precipitation events over the Indian and western Pacific Oceans (>10 mm day−1 onward) (Fig. 6). This excessive overestimation of heavy precipitation events is the primary reason for IMERG’s overall overestimation of precipitation over the Indian and western Pacific Oceans.

  • With regard to the detection of precipitation events, IMERG’s detection capability tends to deteriorate with increasing precipitation rates (Fig. 8). This is strongly evident above the 75th percentile. Moreover, although IMERG exhibits good detection capability for extreme precipitation rates, it also comes with the expense of more false alarms.

  • The error decomposition reveals that IMERG’s positive hit bias and false alarms are the major contributors to IMERG’s overall overestimation throughout the tropical oceans (Fig. 7). This is especially true for the Indian and western Pacific Oceans.

  • We found very slight differences among the different IMERG runs in terms of their performance, with IMERG-F outperforming IMERG-E and IMERG-L.

Our study reveals that it is difficult to conclude whether the reported biases are due to buoy measurement error or IMERG retrieval deficiency. However, it provides a clear overview of the uncertainties encountered and their structural properties. This can offer valuable insight to remote-sensing communities for further research inquiries and methodological improvements. Most importantly, it highlights the need for multisource observation networks over the oceans that will provide a variety of independent sources for evaluation. The combination of buoys, satellite, radar, and ship-based data in a homogeneous network could help us further constrain the uncertainties presented here and get closer to the true estimates of oceanic precipitation globally. In other words, it will provide a more accurate representation of oceanic precipitation and will help address the uncertainties and limitations associated with any single dataset.

Acknowledgments.

R.K.P. was supported by the Internal Grant Agency (Project 2021B0019), Czech University of Life Sciences Prague. We would also like to thank NASA for providing the IMERG datasets and the Global Tropical Moored Buoy Array program for providing the buoy datasets. The authors thank the three anonymous reviewers for their constructive comments, which greatly improved the paper. Finally, the authors would also like to thank Dr. Francesco Marra, Dr. Efthymios I. Nikolopoulos, and Dr. Michael James McPhaden for their useful suggestions and discussion during the course of the work.

Data availability statement.

The IMERG V06 precipitation datasets are available at NASA Goddard Flight Center (https://gpm1.gesdisc.eosdis.nasa.gov/data/GPM_L3/). The buoys datasets are available at https://www.pmel.noaa.gov/gtmba/.

REFERENCES

  • Behrangi, A., and Y. Wen, 2017: On the spatial and temporal sampling errors of remotely sensed precipitation products. Remote Sens., 9, 1127, https://doi.org/10.3390/rs9111127.

    • Search Google Scholar
    • Export Citation
  • Behrangi, A., M. Lebsock, S. Wong, and B. Lambrigtsen, 2012: On the quantification of oceanic rainfall using spaceborne sensors. J. Geophys. Res., 117, D20105, https://doi.org/10.1029/2012JD017979.

    • Search Google Scholar
    • Export Citation
  • Bolvin, D. T., G. J. Huffman, E. J. Nelkin, and J. Tan, 2021: Comparison of monthly IMERG precipitation estimates with PACRAIN atoll observations. J. Hydrometeor., 22, 17451753, https://doi.org/10.1175/JHM-D-20-0202.1.

    • Search Google Scholar
    • Export Citation
  • Bourlès, B., and Coauthors, 2008: The PIRATA program: History, accomplishments, and future directions. Bull. Amer. Meteor. Soc., 89, 11111126, https://doi.org/10.1175/2008BAMS2462.1.

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

    • Search Google Scholar
    • Export Citation
  • Bowman, K. P., J. C. Collier, G. R. North, Q. Wu, E. Ha, and J. Hardin, 2005: Diurnal cycle of tropical precipitation in Tropical Rainfall Measuring Mission (TRMM) satellite and ocean buoy rain gauge data. J. Geophys. Res., 110, D21104, https://doi.org/10.1029/2005JD005763.

    • Search Google Scholar
    • Export Citation
  • Bumke, K., R. Pilch Kedzierski, M. Schröder, C. Klepp, and K. Fennig, 2019: Validation of HOAPS rain retrievals against ocean rain in-situ measurements over the Atlantic Ocean. Atmosphere, 10, 15, https://doi.org/10.3390/atmos10010015.

    • Search Google Scholar
    • Export Citation
  • Chen, C., Q. Chen, Z. Duan, J. Zhang, K. Mo, Z. Li, and G. Tang, 2018: Multiscale comparative evaluation of the GPM IMERG v5 and TRMM 3b42 v7 precipitation products from 2015 to 2017 over a climate transition area of China. Remote Sens., 10, 944, https://doi.org/10.3390/rs10060944.

    • Search Google Scholar
    • Export Citation
  • Derin, Y., P.-E. Kirstetter, N. Brauer, J. J. Gourley, and J. Wang, 2022: Evaluation of IMERG satellite precipitation over the land–coast–ocean continuum. Part II: Quantification. J. Hydrometeor., 23, 12971314, https://doi.org/10.1175/JHM-D-21-0234.1.

    • Search Google Scholar
    • Export Citation
  • Ehsani, M. R., S. Heflin, C. B. Risanto, and A. Behrangi, 2022: How well do satellite and reanalysis precipitation products capture North American monsoon season in Arizona and New Mexico? Wea. Climate Extremes, 38, 100521, https://doi.org/10.1016/j.wace.2022.100521.

    • Search Google Scholar
    • Export Citation
  • Guo, H., S. Chen, A. Bao, A. Behrangi, Y. Hong, F. Ndayisaba, J. Hu, and P. M. Stepanian, 2016: Early assessment of Integrated Multi-satellite Retrievals for Global Precipitation Measurement over China. Atmos. Res., 176, 121133, https://doi.org/10.1016/j.atmosres.2016.02.020.

    • Search Google Scholar
    • Export Citation
  • Hong, Y., K.-L. Hsu, S. Sorooshian, and X. Gao, 2004: Precipitation estimation from remotely sensed imagery using an artificial neural network cloud classification system. J. Appl. Meteor., 43, 18341853, https://doi.org/10.1175/JAM2173.1.

    • Search Google Scholar
    • Export Citation
  • Hou, A., and Coauthors, 2013: The Global Precipitation Measurement (GPM) mission. Bull. Amer. Meteor. Soc., 95, 701722, https://doi.org/10.1175/BAMS-D-13-00164.1.

    • Search Google Scholar
    • Export Citation
  • 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, https://doi.org/10.1175/JHM560.1.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., D. T. Bolvin, D. Braithwaite, K. Hsu, R. Joyce, P. Xie, and S.-H. Yoo, 2015: NASA Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG). Algorithm Theoretical Basis Doc., version 4, 26 pp., https://gpm.nasa.gov/sites/default/files/2020-05/IMERG_ATBD_V06.3.pdf.

  • Huffman, G. J., and Coauthors, 2020: Integrated Multi-satellitE Retrievals for the Global Precipitation Measurement (GPM) mission (IMERG). Satellite Precipitation Measurement, V. Levizzani et al., Eds., Advances in Global Change Research, Vol. 67, Springer, 343353, https://doi.org/10.1007/978-3-030-24568-9_19.

  • Joyce, R. J., and P. Xie, 2011: Kalman filter–based CMORPH. J. Hydrometeor., 12, 15471563, https://doi.org/10.1175/JHM-D-11-022.1.

  • Kazamias, A.-P., M. Sapountzis, and K. Lagouvardos, 2022: Evaluation of GPM-IMERG rainfall estimates at multiple temporal and spatial scales over Greece. Atmos. Res., 269, 106014, https://doi.org/10.1016/j.atmosres.2021.106014.

    • Search Google Scholar
    • Export Citation
  • Khan, S., and V. Maggioni, 2019: Assessment of level-3 gridded Global Precipitation Mission (GPM) products over oceans. Remote Sens., 11, 255, https://doi.org/10.3390/rs11030255.

    • Search Google Scholar
    • Export Citation
  • Klepp, C., and Coauthors, 2018: OceanRAIN, a new in-situ shipboard global ocean surface-reference dataset of all water cycle components. Sci. Data, 5, 180122, https://doi.org/10.1038/sdata.2018.122.

    • Search Google Scholar
    • Export Citation
  • Koschmieder, H., 1934: Methods and results of definite rain measurements: III. Danzig report (1). Mon. Wea. Rev., 62, 57, https://doi.org/10.1175/1520-0493(1934)62<5:MARODR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kucera, P., and C. Klepp, 2018: Evaluation of high resolution IMERG satellite precipitation over the global oceans using OceanRAIN. Geophysical Research Abstracts, Vol. 20, Abstract 17673, https://meetingorganizer.copernicus.org/EGU2018/EGU2018-17673.pdf.

  • Kummerow, C. D., D. L. Randel, M. Kulie, N.-Y. Wang, R. Ferraro, S. Joseph Munchak, and V. Petkovic, 2015: The evolution of the Goddard profiling algorithm to a fully parametric scheme. J. Atmos. Oceanic Technol., 32, 22652280, https://doi.org/10.1175/JTECH-D-15-0039.1.

    • Search Google Scholar
    • Export Citation
  • Li, X., O. Sungmin, N. Wang, I. Liu, and Y. Huang, 2021: Evaluation of the GPM IMERG V06 products for light rain over mainland China. Atmos. Res., 253, 105510, https://doi.org/10.1016/j.atmosres.2021.105510.

    • Search Google Scholar
    • Export Citation
  • Li, Z., G. Tang, P. Kirstetter, S. Gao, J.-L. Li, Y. Wen, and Y. Hong, 2022: Evaluation of GPM IMERG and its constellations in extreme events over the conterminous United States. J. Hydrol., 606, 127357, https://doi.org/10.1016/j.jhydrol.2021.127357.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., 2016: Comparison of Integrated Multisatellite Retrievals for GPM (IMERG) and TRMM Multisatellite Precipitation Analysis (TMPA) monthly precipitation products: Initial results. J. Hydrometeor., 17, 777790, https://doi.org/10.1175/JHM-D-15-0068.1.

    • Search Google Scholar
    • Export Citation
  • Manz, B., S. Páez-Bimos, N. Horna, W. Buytaert, B. Ochoa-Tocachi, W. Lavado-Casimiro, and B. Willems, 2017: Comparative ground validation of IMERG and TMPA at variable spatiotemporal scales in the tropical Andes. J. Hydrometeor., 18, 24692489, https://doi.org/10.1175/JHM-D-16-0277.1.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and Coauthors, 1998: The Tropical Ocean-Global Atmosphere observing system: A decade of progress. J. Geophys. Res., 103, 14 16914 240, https://doi.org/10.1029/97JC02906.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and Coauthors, 2009: RAMA: The Research Moored Array for African–Asian–Australian Monsoon Analysis and Prediction. Bull. Amer. Meteor. Soc., 90, 459480, https://doi.org/10.1175/2008BAMS2608.1.

    • Search Google Scholar
    • Export Citation
  • Nešpor, V., and B. Sevruk, 1999: Estimation of wind-induced error of rainfall gauge measurements using a numerical simulation. J. Atmos. Oceanic Technol., 16, 450464, https://doi.org/10.1175/1520-0426(1999)016<0450:EOWIEO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Pradhan, R. K., and Coauthors, 2022: Review of GPM IMERG performance: A global perspective. Remote Sens. Environ., 268, 112754, https://doi.org/10.1016/j.rse.2021.112754.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., C. Mahesh, R. M. Gairola, and S. Pokhrel, 2011: Surface freshwater flux estimation using TRMM measurements over the tropical oceans. Atmos. Climate Sci., 1, 225234, https://doi.org/10.4236/acs.2011.14025.

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

    • Search Google Scholar
    • Export Citation
  • Prakash, S., M. R. Kumar, S. Mathew, and R. Venkatesan, 2018: How accurate are satellite estimates of precipitation over the north Indian Ocean? Theor. Appl. Climatol., 134, 467475, https://doi.org/10.1007/s00704-017-2287-2.

    • Search Google Scholar
    • Export Citation
  • Rajagopal, M., E. Zipser, G. Huffman, J. Russell, and J. Tan, 2021: Comparisons of IMERG version 06 precipitation at and between passive microwave overpasses in the tropics. J. Hydrometeor., 22, 21172130, https://doi.org/10.1175/JHM-D-20-0226.1.

    • Search Google Scholar
    • Export Citation
  • Retalis, A., D. Katsanos, F. Tymvios, and S. Michaelides, 2018: Validation of the first years of GPM operation over Cyprus. Remote Sens., 10, 1520, https://doi.org/10.3390/rs10101520.

    • Search Google Scholar
    • Export Citation
  • Rojas, Y., J. R. Minder, L. S. Campbell, A. Massmann, and R. Garreaud, 2021: Assessment of GPM IMERG satellite precipitation estimation and its dependence on microphysical rain regimes over the mountains of south-central Chile. Atmos. Res., 253, 105454, https://doi.org/10.1016/j.atmosres.2021.105454.

    • Search Google Scholar
    • Export Citation
  • Sahany, S., V. Venugopal, and R. S. Nanjundiah, 2010: Diurnal-scale signatures of monsoon rainfall over the Indian region from TRMM satellite observations. J. Geophys. Res., 115, D02103, https://doi.org/10.1029/2009JD012644.

    • Search Google Scholar
    • Export Citation
  • Sapiano, M., and P. Arkin, 2009: An intercomparison and validation of high-resolution satellite precipitation estimates with 3-hourly gauge data. J. Hydrometeor., 10, 149166, https://doi.org/10.1175/2008JHM1052.1.

    • Search Google Scholar
    • Export Citation
  • Serra, Y. L., and M. J. McPhaden, 2003: Multiple time- and space-scale comparisons of ATLAS buoy rain gauge measurements with TRMM satellite precipitation measurements. J. Appl. Meteor., 42, 10451059, https://doi.org/10.1175/1520-0450(2003)042<1045:MTASCO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Serra, Y. L., and M. J. McPhaden, 2004: In situ observations of diurnal variability in rainfall over the tropical Pacific and Atlantic Oceans. J. Climate, 17, 34963509, https://doi.org/10.1175/1520-0442(2004)017<3496:ISOODV>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Serra, Y. L., P. A’Hearn, H. P. Freitag, and M. J. McPhaden, 2001: ATLAS self-siphoning rain gauge error estimates. J. Atmos. Oceanic Technol., 18, 19892002, https://doi.org/10.1175/1520-0426(2001)018<1989:ASSRGE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Su, J., H. , Y. Zhu, X. Wang, and G. Wei, 2018: Component analysis of errors in four GPM-based precipitation estimations over mainland China. Remote Sens., 10, 1420, https://doi.org/10.3390/rs10091420.

    • Search Google Scholar
    • Export Citation
  • Tan, J., G. J. Huffman, D. T. Bolvin, and E. J. Nelkin, 2019a: Diurnal cycle of IMERG v06 precipitation. Geophys. Res. Lett., 46, 13 58413 592, https://doi.org/10.1029/2019GL085395.

    • Search Google Scholar
    • Export Citation
  • Tan, J., G. J. Huffman, D. T. Bolvin, and E. J. Nelkin, 2019b: IMERG v06: Changes to the morphing algorithm. J. Atmos. Oceanic Technol., 36, 24712482, https://doi.org/10.1175/JTECH-D-19-0114.1.

    • Search Google Scholar
    • Export Citation
  • Tan, J., G. J. Huffman, D. T. Bolvin, E. J. Nelkin, and M. Rajagopal, 2021: SHARPEN: A scheme to restore the distribution of averaged precipitation fields. J. Hydrometeor., 22, 21052116, https://doi.org/10.1175/JHM-D-20-0225.1.

    • Search Google Scholar
    • Export Citation
  • Tan, M. L., and H. Santo, 2018: Comparison of GPM IMERG, TMPA 3B42 and PERSIANN-CDR satellite precipitation products over Malaysia. Atmos. Res., 202, 6376, https://doi.org/10.1016/j.atmosres.2017.11.006.

    • Search Google Scholar
    • Export Citation
  • Tang, G., Y. Ma, D. Long, L. Zhong, and Y. Hong, 2016: Evaluation of GPM day-1 IMERG and TMPA version-7 legacy products over Mainland China at multiple spatiotemporal scales. J. Hydrol., 533, 152167, https://doi.org/10.1016/j.jhydrol.2015.12.008.

    • Search Google Scholar
    • Export Citation
  • Tian, F., S. Hou, L. Yang, H. Hu, and A. Hou, 2018: How does the evaluation of the GPM IMERG rainfall product depend on gauge density and rainfall intensity? J. Hydrometeor., 19, 339349, https://doi.org/10.1175/JHM-D-17-0161.1.

    • Search Google Scholar
    • Export Citation
  • Tian, Y., and C. D. Peters-Lidard, 2007: Systematic anomalies over inland water bodies in satellite-based precipitation estimates. Geophys. Res. Lett., 34, L14403, https://doi.org/10.1029/2007GL030787.

    • Search Google Scholar
    • Export Citation
  • Tian, Y., and Coauthors, 2009: Component analysis of errors in satellite-based precipitation estimates. J. Geophys. Res., 114, D24101, https://doi.org/10.1029/2009JD011949.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., L. Smith, T. Qian, A. Dai, and J. Fasullo, 2007: Estimates of the global water budget and its annual cycle using observational and model data. J. Hydrometeor., 8, 758769, https://doi.org/10.1175/JHM600.1.

    • Search Google Scholar
    • Export Citation
  • Vargas Godoy, M. R., Y. Markonis, M. Hanel, J. Kyselỳ, and S. M. Papalexiou, 2021: The global water cycle budget: A chronological review. Surv. Geophys., 42, 10751107, https://doi.org/10.1007/s10712-021-09652-6.

    • Search Google Scholar
    • Export Citation
  • Wang, J., D. B. Wolff, J. Tan, D. A. Marks, J. L. Pippitt, and G. J. Huffman, 2022: Validation of IMERG oceanic precipitation over Kwajalein. Remote Sens., 14, 3753, https://doi.org/10.3390/rs14153753.

    • Search Google Scholar
    • Export Citation
  • Wang, Z., R. Zhong, C. Lai, and J. Chen, 2017: Evaluation of the GPM IMERG satellite-based precipitation products and the hydrological utility. Atmos. Res., 196, 151163, https://doi.org/10.1016/j.atmosres.2017.06.020.

    • Search Google Scholar
    • Export Citation
  • Wu, L., Y. Xu, and S. Wang, 2018: Comparison of TMPA-3B42RT legacy product and the equivalent IMERG products over mainland China. Remote Sens., 10, 1778, https://doi.org/10.3390/rs10111778.

    • Search Google Scholar
    • Export Citation
  • Wu, Q., and Y. Wang, 2019: Comparison of oceanic multisatellite precipitation data from tropical rainfall measurement mission and Global Precipitation Measurement mission datasets with rain gauge data from ocean buoys. J. Atmos. Oceanic Technol., 36, 903920, https://doi.org/10.1175/JTECH-D-18-0152.1.

    • Search Google Scholar
    • Export Citation
  • Wu, Y., Z. Zhang, Y. Huang, Q. Jin, X. Chen, and J. Chang, 2019: Evaluation of the GPM IMERG v5 and TRMM 3B42 v7 precipitation products in the Yangtze River basin, China. Water, 11, 1459, https://doi.org/10.3390/w11071459.

    • Search Google Scholar
    • Export Citation
  • Xie, P., R. Joyce, S. Wu, S.-H. Yoo, Y. Yarosh, F. Sun, and R. Lin, 2017: Reprocessed, bias-corrected CMORPH global high-resolution precipitation estimates from 1998. J. Hydrometeor., 18, 16171641, https://doi.org/10.1175/JHM-D-16-0168.1.

    • Search Google Scholar
    • Export Citation
  • Xie, W., S. Yi, C. Leng, D. Xia, M. Li, Z. Zhong, and J. Ye, 2022: The evaluation of IMERG and ERA5-land daily precipitation over China with considering the influence of gauge data bias. Sci. Rep., 12, 8085, https://doi.org/10.1038/s41598-022-12307-0.

    • Search Google Scholar
    • Export Citation
  • Yang, D., B. E. Goodison, J. R. Metcalfe, V. S. Golubev, R. Bates, T. Pangburn, and C. L. Hanson, 1998: Accuracy of NWS 8″ standard nonrecording precipitation gauge: Results and application of WMO intercomparison. J. Atmos. Oceanic Technol., 15, 5468, https://doi.org/10.1175/1520-0426(1998)015<0054:AONSNP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

Supplementary Materials

Save
  • Behrangi, A., and Y. Wen, 2017: On the spatial and temporal sampling errors of remotely sensed precipitation products. Remote Sens., 9, 1127, https://doi.org/10.3390/rs9111127.

    • Search Google Scholar
    • Export Citation
  • Behrangi, A., M. Lebsock, S. Wong, and B. Lambrigtsen, 2012: On the quantification of oceanic rainfall using spaceborne sensors. J. Geophys. Res., 117, D20105, https://doi.org/10.1029/2012JD017979.

    • Search Google Scholar
    • Export Citation
  • Bolvin, D. T., G. J. Huffman, E. J. Nelkin, and J. Tan, 2021: Comparison of monthly IMERG precipitation estimates with PACRAIN atoll observations. J. Hydrometeor., 22, 17451753, https://doi.org/10.1175/JHM-D-20-0202.1.

    • Search Google Scholar
    • Export Citation
  • Bourlès, B., and Coauthors, 2008: The PIRATA program: History, accomplishments, and future directions. Bull. Amer. Meteor. Soc., 89, 11111126, https://doi.org/10.1175/2008BAMS2462.1.

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

    • Search Google Scholar
    • Export Citation
  • Bowman, K. P., J. C. Collier, G. R. North, Q. Wu, E. Ha, and J. Hardin, 2005: Diurnal cycle of tropical precipitation in Tropical Rainfall Measuring Mission (TRMM) satellite and ocean buoy rain gauge data. J. Geophys. Res., 110, D21104, https://doi.org/10.1029/2005JD005763.

    • Search Google Scholar
    • Export Citation
  • Bumke, K., R. Pilch Kedzierski, M. Schröder, C. Klepp, and K. Fennig, 2019: Validation of HOAPS rain retrievals against ocean rain in-situ measurements over the Atlantic Ocean. Atmosphere, 10, 15, https://doi.org/10.3390/atmos10010015.

    • Search Google Scholar
    • Export Citation
  • Chen, C., Q. Chen, Z. Duan, J. Zhang, K. Mo, Z. Li, and G. Tang, 2018: Multiscale comparative evaluation of the GPM IMERG v5 and TRMM 3b42 v7 precipitation products from 2015 to 2017 over a climate transition area of China. Remote Sens., 10, 944, https://doi.org/10.3390/rs10060944.

    • Search Google Scholar
    • Export Citation
  • Derin, Y., P.-E. Kirstetter, N. Brauer, J. J. Gourley, and J. Wang, 2022: Evaluation of IMERG satellite precipitation over the land–coast–ocean continuum. Part II: Quantification. J. Hydrometeor., 23, 12971314, https://doi.org/10.1175/JHM-D-21-0234.1.

    • Search Google Scholar
    • Export Citation
  • Ehsani, M. R., S. Heflin, C. B. Risanto, and A. Behrangi, 2022: How well do satellite and reanalysis precipitation products capture North American monsoon season in Arizona and New Mexico? Wea. Climate Extremes, 38, 100521, https://doi.org/10.1016/j.wace.2022.100521.

    • Search Google Scholar
    • Export Citation
  • Guo, H., S. Chen, A. Bao, A. Behrangi, Y. Hong, F. Ndayisaba, J. Hu, and P. M. Stepanian, 2016: Early assessment of Integrated Multi-satellite Retrievals for Global Precipitation Measurement over China. Atmos. Res., 176, 121133, https://doi.org/10.1016/j.atmosres.2016.02.020.

    • Search Google Scholar
    • Export Citation
  • Hong, Y., K.-L. Hsu, S. Sorooshian, and X. Gao, 2004: Precipitation estimation from remotely sensed imagery using an artificial neural network cloud classification system. J. Appl. Meteor., 43, 18341853, https://doi.org/10.1175/JAM2173.1.

    • Search Google Scholar
    • Export Citation
  • Hou, A., and Coauthors, 2013: The Global Precipitation Measurement (GPM) mission. Bull. Amer. Meteor. Soc., 95, 701722, https://doi.org/10.1175/BAMS-D-13-00164.1.

    • Search Google Scholar
    • Export Citation
  • 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, https://doi.org/10.1175/JHM560.1.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., D. T. Bolvin, D. Braithwaite, K. Hsu, R. Joyce, P. Xie, and S.-H. Yoo, 2015: NASA Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG). Algorithm Theoretical Basis Doc., version 4, 26 pp., https://gpm.nasa.gov/sites/default/files/2020-05/IMERG_ATBD_V06.3.pdf.

  • Huffman, G. J., and Coauthors, 2020: Integrated Multi-satellitE Retrievals for the Global Precipitation Measurement (GPM) mission (IMERG). Satellite Precipitation Measurement, V. Levizzani et al., Eds., Advances in Global Change Research, Vol. 67, Springer, 343353, https://doi.org/10.1007/978-3-030-24568-9_19.

  • Joyce, R. J., and P. Xie, 2011: Kalman filter–based CMORPH. J. Hydrometeor., 12, 15471563, https://doi.org/10.1175/JHM-D-11-022.1.

  • Kazamias, A.-P., M. Sapountzis, and K. Lagouvardos, 2022: Evaluation of GPM-IMERG rainfall estimates at multiple temporal and spatial scales over Greece. Atmos. Res., 269, 106014, https://doi.org/10.1016/j.atmosres.2021.106014.

    • Search Google Scholar
    • Export Citation
  • Khan, S., and V. Maggioni, 2019: Assessment of level-3 gridded Global Precipitation Mission (GPM) products over oceans. Remote Sens., 11, 255, https://doi.org/10.3390/rs11030255.

    • Search Google Scholar
    • Export Citation
  • Klepp, C., and Coauthors, 2018: OceanRAIN, a new in-situ shipboard global ocean surface-reference dataset of all water cycle components. Sci. Data, 5, 180122, https://doi.org/10.1038/sdata.2018.122.

    • Search Google Scholar
    • Export Citation
  • Koschmieder, H., 1934: Methods and results of definite rain measurements: III. Danzig report (1). Mon. Wea. Rev., 62, 57, https://doi.org/10.1175/1520-0493(1934)62<5:MARODR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kucera, P., and C. Klepp, 2018: Evaluation of high resolution IMERG satellite precipitation over the global oceans using OceanRAIN. Geophysical Research Abstracts, Vol. 20, Abstract 17673, https://meetingorganizer.copernicus.org/EGU2018/EGU2018-17673.pdf.

  • Kummerow, C. D., D. L. Randel, M. Kulie, N.-Y. Wang, R. Ferraro, S. Joseph Munchak, and V. Petkovic, 2015: The evolution of the Goddard profiling algorithm to a fully parametric scheme. J. Atmos. Oceanic Technol., 32, 22652280, https://doi.org/10.1175/JTECH-D-15-0039.1.

    • Search Google Scholar
    • Export Citation
  • Li, X., O. Sungmin, N. Wang, I. Liu, and Y. Huang, 2021: Evaluation of the GPM IMERG V06 products for light rain over mainland China. Atmos. Res., 253, 105510, https://doi.org/10.1016/j.atmosres.2021.105510.

    • Search Google Scholar
    • Export Citation
  • Li, Z., G. Tang, P. Kirstetter, S. Gao, J.-L. Li, Y. Wen, and Y. Hong, 2022: Evaluation of GPM IMERG and its constellations in extreme events over the conterminous United States. J. Hydrol., 606, 127357, https://doi.org/10.1016/j.jhydrol.2021.127357.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., 2016: Comparison of Integrated Multisatellite Retrievals for GPM (IMERG) and TRMM Multisatellite Precipitation Analysis (TMPA) monthly precipitation products: Initial results. J. Hydrometeor., 17, 777790, https://doi.org/10.1175/JHM-D-15-0068.1.

    • Search Google Scholar
    • Export Citation
  • Manz, B., S. Páez-Bimos, N. Horna, W. Buytaert, B. Ochoa-Tocachi, W. Lavado-Casimiro, and B. Willems, 2017: Comparative ground validation of IMERG and TMPA at variable spatiotemporal scales in the tropical Andes. J. Hydrometeor., 18, 24692489, https://doi.org/10.1175/JHM-D-16-0277.1.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and Coauthors, 1998: The Tropical Ocean-Global Atmosphere observing system: A decade of progress. J. Geophys. Res., 103, 14 16914 240, https://doi.org/10.1029/97JC02906.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and Coauthors, 2009: RAMA: The Research Moored Array for African–Asian–Australian Monsoon Analysis and Prediction. Bull. Amer. Meteor. Soc., 90, 459480, https://doi.org/10.1175/2008BAMS2608.1.

    • Search Google Scholar
    • Export Citation
  • Nešpor, V., and B. Sevruk, 1999: Estimation of wind-induced error of rainfall gauge measurements using a numerical simulation. J. Atmos. Oceanic Technol., 16, 450464, https://doi.org/10.1175/1520-0426(1999)016<0450:EOWIEO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Pradhan, R. K., and Coauthors, 2022: Review of GPM IMERG performance: A global perspective. Remote Sens. Environ., 268, 112754, https://doi.org/10.1016/j.rse.2021.112754.

    • Search Google Scholar
    • Export Citation
  • Prakash, S., C. Mahesh, R. M. Gairola, and S. Pokhrel, 2011: Surface freshwater flux estimation using TRMM measurements over the tropical oceans. Atmos. Climate Sci., 1, 225234, https://doi.org/10.4236/acs.2011.14025.

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

    • Search Google Scholar
    • Export Citation
  • Prakash, S., M. R. Kumar, S. Mathew, and R. Venkatesan, 2018: How accurate are satellite estimates of precipitation over the north Indian Ocean? Theor. Appl. Climatol., 134, 467475, https://doi.org/10.1007/s00704-017-2287-2.

    • Search Google Scholar
    • Export Citation
  • Rajagopal, M., E. Zipser, G. Huffman, J. Russell, and J. Tan, 2021: Comparisons of IMERG version 06 precipitation at and between passive microwave overpasses in the tropics. J. Hydrometeor., 22, 21172130, https://doi.org/10.1175/JHM-D-20-0226.1.

    • Search Google Scholar
    • Export Citation
  • Retalis, A., D. Katsanos, F. Tymvios, and S. Michaelides, 2018: Validation of the first years of GPM operation over Cyprus. Remote Sens., 10, 1520, https://doi.org/10.3390/rs10101520.

    • Search Google Scholar
    • Export Citation
  • Rojas, Y., J. R. Minder, L. S. Campbell, A. Massmann, and R. Garreaud, 2021: Assessment of GPM IMERG satellite precipitation estimation and its dependence on microphysical rain regimes over the mountains of south-central Chile. Atmos. Res., 253, 105454, https://doi.org/10.1016/j.atmosres.2021.105454.

    • Search Google Scholar
    • Export Citation
  • Sahany, S., V. Venugopal, and R. S. Nanjundiah, 2010: Diurnal-scale signatures of monsoon rainfall over the Indian region from TRMM satellite observations. J. Geophys. Res., 115, D02103, https://doi.org/10.1029/2009JD012644.

    • Search Google Scholar
    • Export Citation
  • Sapiano, M., and P. Arkin, 2009: An intercomparison and validation of high-resolution satellite precipitation estimates with 3-hourly gauge data. J. Hydrometeor., 10, 149166, https://doi.org/10.1175/2008JHM1052.1.

    • Search Google Scholar
    • Export Citation
  • Serra, Y. L., and M. J. McPhaden, 2003: Multiple time- and space-scale comparisons of ATLAS buoy rain gauge measurements with TRMM satellite precipitation measurements. J. Appl. Meteor., 42, 10451059, https://doi.org/10.1175/1520-0450(2003)042<1045:MTASCO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Serra, Y. L., and M. J. McPhaden, 2004: In situ observations of diurnal variability in rainfall over the tropical Pacific and Atlantic Oceans. J. Climate, 17, 34963509, https://doi.org/10.1175/1520-0442(2004)017<3496:ISOODV>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Serra, Y. L., P. A’Hearn, H. P. Freitag, and M. J. McPhaden, 2001: ATLAS self-siphoning rain gauge error estimates. J. Atmos. Oceanic Technol., 18, 19892002, https://doi.org/10.1175/1520-0426(2001)018<1989:ASSRGE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Su, J., H. , Y. Zhu, X. Wang, and G. Wei, 2018: Component analysis of errors in four GPM-based precipitation estimations over mainland China. Remote Sens., 10, 1420, https://doi.org/10.3390/rs10091420.

    • Search Google Scholar
    • Export Citation
  • Tan, J., G. J. Huffman, D. T. Bolvin, and E. J. Nelkin, 2019a: Diurnal cycle of IMERG v06 precipitation. Geophys. Res. Lett., 46, 13 58413 592, https://doi.org/10.1029/2019GL085395.

    • Search Google Scholar
    • Export Citation
  • Tan, J., G. J. Huffman, D. T. Bolvin, and E. J. Nelkin, 2019b: IMERG v06: Changes to the morphing algorithm. J. Atmos. Oceanic Technol., 36, 24712482, https://doi.org/10.1175/JTECH-D-19-0114.1.

    • Search Google Scholar
    • Export Citation
  • Tan, J., G. J. Huffman, D. T. Bolvin, E. J. Nelkin, and M. Rajagopal, 2021: SHARPEN: A scheme to restore the distribution of averaged precipitation fields. J. Hydrometeor., 22, 21052116, https://doi.org/10.1175/JHM-D-20-0225.1.

    • Search Google Scholar
    • Export Citation
  • Tan, M. L., and H. Santo, 2018: Comparison of GPM IMERG, TMPA 3B42 and PERSIANN-CDR satellite precipitation products over Malaysia. Atmos. Res., 202, 6376, https://doi.org/10.1016/j.atmosres.2017.11.006.

    • Search Google Scholar
    • Export Citation
  • Tang, G., Y. Ma, D. Long, L. Zhong, and Y. Hong, 2016: Evaluation of GPM day-1 IMERG and TMPA version-7 legacy products over Mainland China at multiple spatiotemporal scales. J. Hydrol., 533, 152167, https://doi.org/10.1016/j.jhydrol.2015.12.008.

    • Search Google Scholar
    • Export Citation
  • Tian, F., S. Hou, L. Yang, H. Hu, and A. Hou, 2018: How does the evaluation of the GPM IMERG rainfall product depend on gauge density and rainfall intensity? J. Hydrometeor., 19, 339349, https://doi.org/10.1175/JHM-D-17-0161.1.

    • Search Google Scholar
    • Export Citation
  • Tian, Y., and C. D. Peters-Lidard, 2007: Systematic anomalies over inland water bodies in satellite-based precipitation estimates. Geophys. Res. Lett., 34, L14403, https://doi.org/10.1029/2007GL030787.

    • Search Google Scholar
    • Export Citation
  • Tian, Y., and Coauthors, 2009: Component analysis of errors in satellite-based precipitation estimates. J. Geophys. Res., 114, D24101, https://doi.org/10.1029/2009JD011949.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., L. Smith, T. Qian, A. Dai, and J. Fasullo, 2007: Estimates of the global water budget and its annual cycle using observational and model data. J. Hydrometeor., 8, 758769, https://doi.org/10.1175/JHM600.1.

    • Search Google Scholar
    • Export Citation
  • Vargas Godoy, M. R., Y. Markonis, M. Hanel, J. Kyselỳ, and S. M. Papalexiou, 2021: The global water cycle budget: A chronological review. Surv. Geophys., 42, 10751107, https://doi.org/10.1007/s10712-021-09652-6.

    • Search Google Scholar
    • Export Citation
  • Wang, J., D. B. Wolff, J. Tan, D. A. Marks, J. L. Pippitt, and G. J. Huffman, 2022: Validation of IMERG oceanic precipitation over Kwajalein. Remote Sens., 14, 3753, https://doi.org/10.3390/rs14153753.

    • Search Google Scholar
    • Export Citation
  • Wang, Z., R. Zhong, C. Lai, and J. Chen, 2017: Evaluation of the GPM IMERG satellite-based precipitation products and the hydrological utility. Atmos. Res., 196, 151163, https://doi.org/10.1016/j.atmosres.2017.06.020.

    • Search Google Scholar
    • Export Citation
  • Wu, L., Y. Xu, and S. Wang, 2018: Comparison of TMPA-3B42RT legacy product and the equivalent IMERG products over mainland China. Remote Sens., 10, 1778, https://doi.org/10.3390/rs10111778.

    • Search Google Scholar
    • Export Citation
  • Wu, Q., and Y. Wang, 2019: Comparison of oceanic multisatellite precipitation data from tropical rainfall measurement mission and Global Precipitation Measurement mission datasets with rain gauge data from ocean buoys. J. Atmos. Oceanic Technol., 36, 903920, https://doi.org/10.1175/JTECH-D-18-0152.1.

    • Search Google Scholar
    • Export Citation
  • Wu, Y., Z. Zhang, Y. Huang, Q. Jin, X. Chen, and J. Chang, 2019: Evaluation of the GPM IMERG v5 and TRMM 3B42 v7 precipitation products in the Yangtze River basin, China. Water, 11, 1459, https://doi.org/10.3390/w11071459.

    • Search Google Scholar
    • Export Citation