1. Introduction
Precipitation is one of the most important elements in global and regional hydrologic and energy cycles (Kidd and Huffman 2011; Wu et al. 2013; Yong et al. 2015) and is also a critical parameter used in ecological modeling to assess the relationship among vegetation, climate, and human activities and to solve important ecological problems, such as ecological program construction and regional water resource allocation. Therefore, accurate and high-resolution precipitation distributions are crucial for ecological engineering and the redistribution of regional water resources (AghaKouchak et al. 2011; Golian et al. 2015; Sharifi et al. 2016; Mahmoud et al. 2019).
Currently, interpolated gauge observation data and ground-based weather radar and satellite precipitation products are the main methods for obtaining precipitation data, which have been widely used. Interpolated gauge data provide a conventional and reliable method for precipitation quantification via direct measurement. However, this method experiences limitations in areas where gauges are unevenly distributed or the distance increases to hundreds of kilometers (Zhang et al. 2004; Michaelides et al. 2009). In addition, observation stations tend to be located in strategically determined operational areas and do not optimally represent the environments in which afforestation (or reforestation) could occur. These conditions lead to inaccurately interpolated gauge observation estimates of precipitation. Radar measurements can provide real-time high-resolution monitoring over a large area (Germann et al. 2006). Nevertheless, low radar network coverage rates occur in many areas around the globe (Tang et al. 2016), and radar measurements experience various sources of errors, including range-dependent systematic errors, mean-field systematic errors and random errors (Dinku et al. 2002). Satellite precipitation products constitute one of the most important data sources for estimating the spatiotemporal characteristics of precipitation in recent years, especially in complex terrain regions with limited observations (Yong et al. 2012; Guo et al. 2016). Moreover, numerous satellite precipitation missions have become operational, with freely available satellite data to the public, such as the Climate Precipitation Center morphing (CMORPH) technique (Joyce et al. 2004), Global Satellite Mapping of Precipitation (GSMaP), Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG; Huffman et al. 2020), Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks (PERSIANN; Sorooshian et al. 2000), and Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA; Huffman et al. 2020).
The GPM Core Observatory satellite, as the successor of the TRMM satellite, was launched on 28 February 2014 by the National Aeronautics and Space Administration (NASA) and the Japan Aerospace Exploration Agency (JAXA). The GPM satellite carried the first spaceborne dual-frequency precipitation radar (DPR) and a conical-scanning multichannel GPM Microwave Imager (GMI), estimating low-intensity precipitation (<0.5 mm h−1) and solid precipitation more accurately than the TRMM (Hou et al. 2014; Caracciolo et al. 2018). GPM-IMERG is a new-generation multisatellite precipitation algorithm that was released by NASA. This dataset integrates the advantages of other products, such as CMORPH and PERSIANN. To date, numerous studies have been devoted to the confirmation and verification of satellite-based meteorological observation data in various countries and regions, such as India (Prakash et al. 2016), Arabia (Mahmoud et al. 2019), Singapore (Tan and Duan 2017), the northeastern region of Iran (Shirmohammadi-Aliakbarkhani and Akbari 2020), the two major Mediterranean islands (Caracciolo et al. 2018), and China (Tang et al. 2016; Lu et al. 2018; Wei et al. 2018; Liu et al. 2020; Zhou et al. 2020).
China has conducted ecological restoration programs (including reforestation programs) to address climate change and improve the regional ecological environment. These programs require high-resolution precipitation data on annual, seasonal, and monthly scales. However, meteorological stations are sparse, and China’s terrain is complex and diverse, with different climatic zones. Therefore, GPM-IMERG V06 Final Run products (herein, GPM-IMERG) provide the possibility to obtain high-resolution spatiotemporal precipitation data for China’s ecological restoration programs. However, most previous studies analyzing the GPM-IMERG product performance have focused on arid and semiarid zones of China, compared GPM-IMERG data with other data (e.g., TRMM), and analyzed the GPM-IMERG product performance on a daily scale (e.g., light precipitation) (Table 1). No studies have quantified the strengths and weaknesses of the GPM mission with multiple uniform standards in areas of notable disparity on monthly, quarterly, and annual scales.
Summary of the related literature.
Therefore, the main objectives of this study were to 1) evaluate the performance of GPM-IMERG in terms of estimating monthly, seasonal, and annual precipitation over mainland China and 2) analyze the error structure and features of GPM-IMERG and determine the impacts of climate and geography on the retrieval errors. Our research strived to demonstrate whether GPM-IMERG can offset the deficiency of site observations.
2. Study area, dataset, and methodology
a. Study area
Mainland China, which is located on the eastern Eurasian continent and west of the Pacific Ocean between 73°33′ and 135°05′E and between 3°51′ and 53°33′N, encompasses a variety of climatic conditions and complex terrain. The climate of mainland China is influenced by various factors, such as the Tibetan Plateau, the western Pacific subtropical high pressure, the East Asian monsoon circulation system, and El Niño–Southern Oscillation on a large scale, as well as small-scale factors, such as regional-scale topographic forcing and mesoscale eddies. Based on topographic and climatic elements, mainland China can be divided into eight geographical regions: 1) North China (NC) includes Beijing, Tianjin, Shanxi, Hebei, and Inner Mongolia. The climate of NC is mainly a temperate monsoon climate, with annual average temperatures of 8°–13°C and annual precipitation of approximately 400–1000 mm. However, the precipitation in Inner Mongolia is less than 400 mm, rendering it a semiarid region (Dong et al. 2020). 2) Northeast China (NeC) includes Liaoning, Jilin, and Heilongjiang. NeC is the highest-latitude region of China, with a cold climate in winter. Satellite precipitation products are prone to high uncertainties in these regions. 3) Northwest China (NwC) includes Xinjiang, Qinghai, Gansu, Ningxia, and Shanxi and contains vast areas prone to drought and water shortages and widespread deserts, mountains, and basins (Lu et al. 2019). 4) Southwest China (SwC) includes Sichuan, Yunnan, Guizhou, and Chongqing, is hot and humid in summer, and is mild and dry in winter (Wei et al. 2018). 5) Central China (CC) includes Henan, Hubei, Hunan, and Jiangxi Provinces. 6) East China (EC) includes Anhui, Jiangsu, and Zhejiang Provinces. 7) South China (SC) includes Guangxi, Guangdong, Fijian, Hainan, and Taiwan. The most distinctive climatic feature of this subregion is the East Asian monsoon (Chen et al. 2010). The climate in 8) Tibet (Tib) varies greatly from north to south. It is cold and dry in the northwest and warm and humid in the southeast, with a complex climate and topography (Yao et al. 2012). The annual precipitation amount varies across Tib (Zhu et al. 2018) (Fig. 1; Table 2).
(a) Meteorological station locations and digital elevation model of mainland China, and (b) the eight geographical divisions and representative stations.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
Climate and topography in the different regions.
b. Dataset
1) Meteorological observation data
Monthly and annual precipitation observations data (for all phases—liquid, mixed, and solid phases) were obtained from 2014 to 2020 covering 699 meteorological stations in China, and these observations were used as benchmarks for GPM-IMERG product evaluation (Fig. 1). These observations are provided by the China Meteorological Administration (http://data.cma.cn/), and the stations are unevenly distributed in China. Of these stations, 15.77% are situated in NC, 13.14% are situated in NeC, 19.40% are situated in NwC, 16.02% are situated in SwC, 14.27% are situated in CC, 8.64% are situated in EC, 10.26% are situated in SC, and 2.50% are situated in Tib.
2) GPM-IMERG
The GPM mission comprises one Core Observatory and approximately 10 constellation satellites. The GPM Core Observatory (GPMCO) introduced a combination of two advanced instruments, a DPR (the Ku band is 13.6 GHz, and the Ka band is 35.5 GHz) and a microwave radiometer, namely, the GMI, which provides a frequency range of 10 to 183 GHz (Ning et al. 2017; Tang et al. 2016). Therefore, the GPM can detect light precipitation (<0.5 mm h−1) and solid precipitation more accurately (Hou et al. 2014). The GPM-IMERG products offer a relatively fine spatial resolution of 0.1° × 0.1° and a high temporal resolution of half an hour (Huffman et al. 2017, 2020; Zhang et al. 2021; Jiang et al. 2023). The GPM-IMERG algorithm first estimates precipitation system motions similar to CMORPH for passive microwave (PMW) data and then uses a Kalman filter to incorporate the shifted PMW data with infrared-based estimates for missing data.
Currently, IMERG offers three types of satellite precipitation products: Early Run, Late Run and Final Run products. Although the Early and Late products have provided high value in terms of disaster warning (they are produced in near–real time), most assessments indicate that the Final run products generally outperform the other two and better capture the temporal and spatial distributions of precipitation. The Final products are calibrated by the Global Precipitation Climatology Centre’s (GPCC) full and monthly monitoring products and are calibrated by GPCC monitoring products, and the data source is the geostationary tether satellite (GTS), with approximately 7000 stations worldwide (Schneider et al. 2014). In this study, we used GPM satellite data retrieved from version 6 of the IMERG Final product over China (IMERG V06 Final) during the GPM era (2014–20) (available at https://pmm.nasa.gov/data-access/downloads/gpm).
c. Methodology
1) Evaluation metrics
To evaluate the satellite estimation performance, the GPM-IMERG products were compared with observations data from meteorological stations. Several statistical indicators were used to quantify the consistency between the satellite precipitation and gauge data, including Pearson’s correlation coefficient R, bias, relative bias (RBias), root-mean-square error (RMSE), and relative root-mean-square error (RRMSE) (Table 3). The R indicates the degree of linear correlation between the satellite products and gauge observations, bias and Rbias were used to examine the overestimation or underestimation degree of the satellite precipitation products, and the RMSE reflects the magnitude of the average error, which measures sample dispersion to ensure the reliability of the satellite data (Liu et al. 2015; Kim et al. 2022). The RRMSE evaluates the accuracy of GPM-IMERG by considering geographic or seasonal differences (Chen and Li 2016). The closer the R value is to 1 and the closer the values of bias, Rbias, and RMSE are to 0, the better the GPM-IMERG product performance is. This estimate is considered to be reliable when the RRMSE value is less than 50%. In contrast, when the RRMSE is equal to or higher than 50%, the estimate is considered to be unreliable for the region and specific season.
Evaluation-index (as defined in the text) equations for GPM-IMERG satellite precipitation. Note that n is the number of samples, xi is the precipitation observation at meteorological stations,
We selected the monthly quantile bias (MQB) to analyze extreme precipitation on monthly scales (AghaKouchak et al. 2011). The MQB was defined as the monthly ratio of the satellite-based precipitation accumulation to the reference precipitation accumulation above a given threshold (Table 3). In addition, three categorical scores were considered, including the probability of detection (POD), false alert ratio (FAR), and critical success index (CSI) (Xie et al. 2022). The POD represents the proportion of correctly detected precipitation events, the FAR represents the proportion of precipitation events not detected by the meteorological station but detected by the satellite, and the CSI indicates the overall ratio of precipitation events correctly detected by the satellite.
2) Error decomposition
The above statistical metrics quantify the magnitude of the error but cannot indicate the error source. As well, these error metrics are incomplete, interdependent, and indirect, and they assume linear, additive, and Gaussian errors (Tian et al. 2016). Therefore, when evaluating the applicability of satellite precipitation products, it is important to perform error decomposition. We employed effective error decomposition strategies, thus decomposing the bias into three components: the hit bias, missed precipitation, and false precipitation (Tian et al. 2009). This section used a daily precipitation rate threshold of 0.25 mm to better identify error sources (Zhou et al. 2020).
The hit bias was defined as the difference between nonzero satellite data and nonzero gauge observations. Missed precipitation represents a nonzero gauge observation when the precipitation event was not detected by the satellite, and false precipitation represents nonzero satellite data when the precipitation event was not detected by the meteorological station. The three error components were calculated, as listed in Table 3.
To examine the factors influencing the accuracy of GPM-IMERG in terms of precipitation estimation, we analyzed the relationship between the elevation and precipitation differences between GPM-IMERG and the meteorological station observations, as well as the effects of elevation on the above precipitation evaluation metrics. We chose to divide mainland China into seven elevation ranges: 1) <500, 2) 500–1000, 3) 1000–1500, 4) 1500–2000, 5) 2000–3000, 6) 3000–4000, and 7) >4000 m.
3. Results and discussion
a. Evaluation on the monthly scale
At the monthly scale, GPM-IMERG agreed with the meteorological station precipitation observations, with average R, bias, Rbias, and RMSE values of 0.93, 6.36 mm, 38.53%, and 22.79 mm, respectively (Fig. 2a). The R values are actually quite high and have very small variations. GPM-IMERG generally reflects the monthly variation trend of precipitation in China, and it performed better from April to September than from October to March (Fig. 2b). This conclusion is similar to that of Guo et al. (2016), and it may be related to the lower capability of the DPR and GMI algorithms in quantifying solid precipitation (snowfall). The RMSE and Rbias values exhibited obvious periodic changes; the RMSE values were lower from October to March of the following year and gradually increased from April, which is similar to the result obtained by Zhao et al. (2021) in Yunnan, China, and the Rbias values were lower from April to September. This periodic variation in the RMSE was closely related to the seasonal distribution of precipitation in the study area, which mostly occurred in June, July, and August. The main flood season was the period with the highest precipitation intensity and precipitation in a given year, and September represented the gradual transition to the end of the rainy season. With decreasing precipitation, the RMSE decreased and reached its lowest value in December. The total precipitation in China was relatively low from October to March, resulting in low RMSE and high Rbias and R values. China experienced high precipitation during the rest of the year; the RMSE values were large, but the Rbias and R values were relatively poor (Fig. 2b).
(a) Scatterplot and fitting functions of GPM-IMERG against the meteorological station observations at all stations at the monthly scale (the blue and red lines are 1:1 and fitting lines, respectively), and (b) the annual cycle of the evaluation indices.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
For the performance of GPM-IMERG in the different zones on the monthly scale, we selected a representative precipitation station in each zone to compare the relationship between GPM-IMERG and the meteorological station data (Fig. 3). Influenced by the monsoon climate, the precipitation at all meteorological stations began to increase in March every year, and the peak appeared in June and July. Then, the precipitation decreased in August. After October, the precipitation returned to a lower level, which is consistent with the actual precipitation pattern in China. In the different zones, we found that GPM-IMERG performed relatively well in SwC, CC, EC, and SC, with bias values of 7.12, 6.11, 6.04, and 9.30 mm, respectively, and Rbias values of 23.31%, 15.04%, 12.98%, and 13.57%, respectively. GPM-IMERG precipitation estimation was adequately performed in NC and NeC. NwC and Tib exhibited the worst estimation effect, with R values of 0.89 and 0.91, respectively, and Rbias values of 108.21% and 87.45%, respectively. In addition, we found that GPM-IMERG usually overestimated precipitation in most regions and months and that it overestimated the monthly precipitation in most parts of China.
Comparison of the monthly precipitation determined by GPM-IMERG and the observation data in the different regions from 2014 to 2020 (note the differences in the y-axis range in the various panels).
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
Regarding extreme precipitation on a monthly scale, GPM-IMERG exhibited overestimation (MQB > 1) for all data, but the overestimation degree of precipitation in the warm months, including May, June, July, and August, was lower (Fig. 4). With increasing threshold value, GPM-IMERG tended to underestimate the extreme values. At a threshold of 75%, the MQB was closest to 1, while at a threshold of 95%, GPM-IMERG substantially underestimated precipitation, especially during the warm months. The MQB revealed that GPM-IMERG tended to overestimate the extremely low values and underestimate the extremely high values.
MQB with respect to all data and the 75th-, 90th-, and 95th-percentile thresholds of the observation data.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
b. Evaluation on the seasonal scale
Precipitation is highly seasonally dependent given that most precipitation over China is monsoon-driven precipitation. Therefore, we analyzed the accuracy of GPM-IMERG at the seasonal scale (spring: March–May; summer: June–August; autumn: September–November; and winter: December–February). GPM-IMERG yielded high R values of 0.82, 0.90, 0.88, and 0.87 for spring, summer, autumn, and winter, respectively (Fig. 5). These results showed that GPM-IMERG could effectively capture the seasonal precipitation patterns in mainland China. GPM-IMERG attained higher R (R = 0.9) values, and lower bias (bias = 5.07 mm) and Rbias (Rbias = 22.67%) values in summer, which are better than those during the other seasons. IMERG V06 Final produced more overestimations in winter, with R = 0.87 and Rbias = 46.22%.
Scatterplot and fitting functions of GPM-IMERG against the gauge observations at all stations at the seasonal scale: (top left) spring, (top right) summer, (bottom left) autumn, and (bottom right) winter.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
The performance of GPM-IMERG at the seasonal scale varied widely across regions (Fig. 6); the observation accuracy of GPM-IMERG in NwC and Tib was poorer than that in mainland China, and there were many stations with R < 0.4 and Rbias > 1. According to the analysis of Rbias during the four seasons, the estimation of winter precipitation in China by GPM-IMERG was substantially worse, and the number of stations with Rbias > 1 was substantially larger than that during the other seasons. Moreover, the RMSE determined for GPM-IMERG and the observation stations showed an increase from NwC to SwC during all four seasons.
Statistical indicators (R, RBias, and RMSE) between the GPM-IMERG data and gauge observations in terms of seasonal precipitation.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
The regional study of the RRMSE showed that among all seasons, the summer precipitation estimates were the most reliable and the winter precipitation estimates were the least reliable (Fig. 7). In summer, the RRMSE values in all subregions except NwC were the lowest during all four seasons. One of the main functions of the GPM mission was to improve snow cover estimation accuracy. However, in winter, GPM-IMERG did not substantially improve the estimation accuracy in NC and NeC. In CC, EC, SC, and SwC, the GPM-IMERG precipitation estimates were relatively reliable throughout the year. In NC and NeC, the GPM-IMERG estimates were not reliable in winter, which was due to the persistent snow cover affecting the estimation accuracy. In NwC and Tib, the regional GPM-IMERG precipitation estimates were relatively unreliable during all seasons due to the complex topographic conditions.
RRMSE values for the subregions on a seasonal scale.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
c. Quantitative evaluation of the GPM-IMERG-derived annual precipitation
We compared the GPM-IMERG data with the annual precipitation observations of stations and found that their ability for monitoring the annual precipitation was better than that for the monthly precipitation, with an R value of 0.97 and an Rbias value of 19.24% from 2014 to 2020 (Fig. 8). Because of the high interannual variability in precipitation, GPM-IMERG performed better in 2014, 2015, and 2017, with lower bias (54.80, 73.63, and 65.23 mm, respectively) and RMSE values (139.75, 161.14, and 164.45 mm, respectively). The performance was relatively poor in 2019 and 2020, with higher bias (85.26 and 98.39 mm, respectively) and RMSE values (173.48 and 197.61 mm, respectively). Based on the scatterplots and statistical indices, the best correlation between the GPM-IMERG precipitation estimates and the gauge observations was observed in 2016, with an R value of 0.98, a bias of 62.37 mm yr−1, and RBias of 14.66% (Table 4).
Scatterplot and statistical indices of the GPM-IMERG data against the meteorological station observation annual precipitation values at all stations in China.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
Evaluation indices of the GPM-IMERG precipitation products in the different years.
To illustrate the annual-scale GPM-IMERG performance, we compared the kriging interpolation values of the meteorological stations with the mean of the satellite estimates. We found that GPM-IMERG could describe the spatial distribution characteristics of the annual precipitation with an R value of 0.97, a bias of 95.17 mm yr−1, RBias of 19.24%, and RMSE of 186.13 mm yr−1. Our result is similar to the results of Anjum et al. (2019), who reported significant spatial variations in the GPM-IMERG estimation skill. The GPM-IMERG performance varied widely across the different regions; the annual precipitation derived in CC, EC, and SC was much higher than that derived in Tib and NwC (Fig. 9), and the results are similar to those of Guo et al. (2016). GPM-IMERG overestimated the annual precipitation in China in most areas of China, but there are a few underestimates in the high-precipitation areas, such as SwC, CC, EC, and SC.
(a) Annual average precipitation map derived from kriging interpolation of the gauge observations, (b) GPM-IMERG data for China from 2014 to 2020, and (c) the precipitation difference between the satellite estimates and gauge observations.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
The R values could exceed 0.9 in most regions of China, indicating a high correlation with the observed values of the meteorological stations (Fig. 10a). The GPM-IMERG RBias value in most regions of China ranged from 0% to 20%, and RBias was greater than 100% only in the NwC and Tib regions (Fig. 10b), which may be due to the complex terrain and sparse precipitation with a notable contribution of snow. In CC, EC, and SC, with high annual precipitation levels, the RMSE value significantly exceeded the overall value of China. The RMSE exhibited a decreasing trend from southeast to northwest, which is consistent with the precipitation variation pattern, and the RMSE value decreased with decreasing precipitation (Fig. 10c). The alternating pattern of overestimation and underestimation for GPM-IMERG in NwC may be related to the sparsely and unevenly spaced gauge stations, the kriging interpolation technique, or the complex terrain, as low-level topographic precipitation is difficult to detect with PMW and infrared satellites (Karaseva et al. 2012; Taniguchi et al. 2013). Research by Wang et al. (2019) showed that IMERG does not show substantial improvement in terrain fluctuations and was even worse in estimating precipitation.
Statistical indices [(a) R, (b) RBias, and (c) RMSE] between the GPM-IMERG data and gauge observations in terms of the annual precipitation.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
The average values of the statistical indices in the eight regions are provided in Table 5. At the annual scale, GPM-IMERG performed well in SwC, CC, EC, and SC, with lower RBias and RRMSE values than those in the other regions in China. Next, GPM-IMERG performed relatively well in NeC and NC. However, in NwC and Tib, GPM-IMERG performed poorly in regard to precipitation estimation. These conclusions are consistent with those at the monthly scale. The estimation error in NwC and Tib may be attributed at least in part to the complex terrain and sparsely distributed measurement stations.
Annual evaluation index of the GPM-IMERG precipitation products in the different regions.
d. GPM-IMERG error sources
In areas with an annual precipitation ranging from 0 to 500 mm, false precipitation was the main factor affecting the accuracy of GPM-IMERG estimation, and the maximum value of false precipitation occurred in this precipitation area (Fig. 11d). In areas with an annual precipitation ranging 500–1000 mm, missed precipitation was the main error source, and the maximum error also occurred in this area (Fig. 11c). In areas with an annual precipitation of >2000 mm, the main source of the GPM-IMERG estimation error was the hit bias, the missed precipitation value was almost 0, and the false precipitation value was minor. Overall, the main error affecting the GPM-IMERG data in China was the hit bias. Missed precipitation and false precipitation yielded values of mostly 0 for each range of precipitation values.
Relationship between bias and the annual precipitation: (a) total bias, (b) hit bias, (c) miss precipitation, and (d) false precipitation.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
In addition, the false precipitation values mostly ranged from 0 to 1 mm, followed by 1–5 mm, and they were mainly located in areas with complex terrain. The precipitation missed by GPM-IMERG was mainly concentrated in NeC and NC, which was the direct cause of the estimation bias (Fig. 12d). This result indicates that GPM-IMERG provides an insufficient capacity to monitor light precipitation. False precipitation was the main bias source of GPM-IMERG in NwC (Fig. 12c). In CC, SC, and EC, which experience high precipitation levels, the main reason for the GPM-IMERG observation bias was the hit bias, followed by false precipitation, and there was almost no bias caused by missed precipitation (Figs. 12b–d). Our findings are basically consistent with those of Li et al. (2021).
Average error components of the GPM-IMERG precipitation products in 2016: (a) bias, (b) hit bias, (c) false precipitation, and (d) missed precipitation.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
The error of the satellite precipitation products was also affected by terrain, and several previous studies have reported the dependence of the GPM precipitation product performance on elevation (e.g., Ma et al. 2016; Xu et al. 2017). Therefore, we selected several representative indicators, including evaluation indices (i.e., R and RBias), error components (i.e., false precipitation and missed precipitation), and categorical indices (i.e., the POD, FAR, and CSI), and we calculated their changes with elevation. The R value between the GPM-IMERG estimates and the gauge observations decreased, and the RBias value increased with increasing elevation (Figs. 13a,b). Additionally, with increasing elevation, false precipitation and missed precipitation substantially increased (Figs. 13c,d). Although the R values of these indicators were small, the linear fit between elevation and each selected index passed the 0.05 level significance test, which indicated a weak effect of terrain on IMERG.
Scatterplots between the statistical indicators and elevation based on the annual precipitation from March 2014 to December 2020: (a) R, (b) RBias, (c) false precipitation, and (d) missed precipitation.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
Combined with the results shown in Fig. 14, the GPM-IMERG precipitation estimation error in China increased with increasing elevation. We found that GPM-IMERG is highly correlated with the meteorological observations at elevations of <1500 m (R ≥ 0.95). However, the GPM-IMERG and meteorological observation data exhibited the worst relationship at elevations > 4000 m (R = 0.85). At elevations of <500 m, GPM-IMERG substantially overestimated precipitation in China, with bias = 98.33 mm and RMSE = 174.29 mm, which were the highest among the seven elevation ranges. At elevations of 1500–2000 and 3000–4000 m, GPM-IMERG exhibited a favorable performance in precipitation evaluation, with lower Bias and RBias values. At elevations > 4000 m, GPM-IMERG attained a poor precipitation accuracy evaluation performance, resulting in the lowest R value and highest RBias (40.18%) and RMSE (246.93 mm) values with respect to the gauge observations. Among the seven elevation intervals, the estimated value of GPM-IMERG was higher than the observed value, the R values were similar, and the RBias values were very different. The results showed that the lowest RBias (6.51%) and bias values (13.39 mm) were obtained at elevations of 1500–3000 m in China (Fig. 14); this finding is similar to the results obtained by Ma et al. (2019) in a study comparing the precipitation estimated by GPM-IMERG with instrumentally observed values. In addition, the dispersion of R for GPM-IMERG was greater at elevations of 2000–3000 m and that of RBias was greater at elevations of 1000–1500 and >4000 m. The bias and RMSE values tended to decrease with increasing elevation, which may occur because high-elevation areas are mainly distributed in areas with less precipitation in China (Fig. 15).
Density scatterplots of the GPM-IMERG data and gauge observations within different elevation ranges: (a) <500, (b) 500–1000, (c) 1000–1500, (d) 1500–2000, (e) 2000–3000, (f) 3000–4000, and (g) e>4000 m. The blue lines are 1:1 lines, and the red lines are the fitted lines.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
Statistical indicators within different elevation ranges for GPM-IMERG: x-axis labels a–g indicate <500, 500–1000, 1000–1500, 1500–2000, 2000–3000, 3000–4000, and >4000 m; the horizontal line in the box is the median, and the edges of the box are the 25th and 75th percentiles. The blue line denotes a perfect match for each indicator.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
By analyzing the seasonal indicators of GPM-IMERG, we found that the POD values were the smallest in winter and decreased with increasing elevation. The POD values were the largest in summer and decreased in the elevation range of >4000 m (Fig. 16a). At each elevation range, FAR values were lower in summer, and CSI was higher than that during the other seasons (Figs. 16b,c).
Categorical index values of the GPM-IMERG data and gauge observations in different seasons and elevation ranges for (a) POD, (b) FAR, and (c) CSI. The elevation ranges are labeled as in Fig. 15.
Citation: Journal of Applied Meteorology and Climatology 62, 8; 10.1175/JAMC-D-22-0110.1
4. Conclusions
In this study, we compared GPM-IMERG, version 6, with gauge measurements from 2014 to 2020 retrieved from the National Meteorological Information Center of China, which is the first study to evaluate the spatial performance of the GPM-IMERG system under difficult climatic and terrain conditions. The main findings are as follows:
-
GPM-IMERG captured the spatial and temporal distributions of the monthly, seasonal, and annual precipitation levels in China and performed better at the annual scale than at the monthly scale. GPM-IMERG performed significantly better in summer than in the other seasons.
-
The climate in China played a leading role in the GPM-IMERG accuracy. GPM-IMERG performed well in humid zones and relatively poorly in dry zones, and it performed best in summer. Missed precipitation and false precipitation were the main sources of GPM-IMERG deviation in the arid and semiarid regions.
-
In addition to climatic conditions, the topography impacted the accuracy of GPM-IMERG. With increasing elevation, the influences of missed precipitation and false precipitation on the bias increased, the probability of detecting precipitation substantially decreased, and the GPM-IMERG precipitation estimation accuracy decreased.
Despite the substantial bias in precipitation estimation in the dry and high-elevation areas, GPM-IMERG generally effectively reflected the spatial-temporal distribution and total amount of precipitation in China, demonstrating a high application potential for supplementing gauge observation data, especially in areas with an uneven distribution or an insufficient number of observation stations. We hope that our results can facilitate GPM application in water resource management, afforestation (or reforestation) projects, and so on, in areas with scarce meteorological stations.
Acknowledgments.
This work was financially supported by the National Key Research and Development Program of China (Grant 2022YFF1302501-04) and the National Natural Science Foundation of China (Grant 31770758). We thank the editor and anonymous reviewers for their constructive comments, which helped us to improve the paper.
Data availability statement.
GPM-IMERG V06 satellite datasets were used in this study and are freely available. The satellite precipitation data were obtained online (https://pmm.nasa.gov/data-access/downloads/gpm). Meteorological station observation data were also used, and they were freely downloaded from the China Meteorological Administration (http://data.cma.cn/).
REFERENCES
AghaKouchak, A., A. Behrangi, S. Sorooshian, K. Hsu, and E. Amitai, 2011: Evaluation of satellite-retrieved extreme precipitation rates across the central United States. J. Geophys. Res., 116, D02115, https://doi.org/10.1029/2010JD014741.
Anjum, M. N., and Coauthors, 2019: Assessment of IMERG-V06 precipitation product over different hydro-climatic regimes in the Tianshan Mountains, north-western China. Remote Sens., 11, 2314, https://doi.org/10.3390/rs11192314.
Caracciolo, D., A. Francipane, F. Viola, L. V. Noto, and R. Deidda, 2018: Performances of GPM satellite precipitation over the two major Mediterranean islands. Atmos. Res., 213, 309–322, https://doi.org/10.1016/j.atmosres.2018.06.010.
Chen, F., and X. Li, 2016: Evaluation of IMERG and TRMM 3B43 monthly precipitation products over mainland China. Remote Sens., 8, 472, https://doi.org/10.3390/rs8060472.
Chen, F., and H. Huang, 2018: Comparisons of gauge, TMPA and IMERG products for monsoon and tropical cyclone precipitation in southern China. Pure Appl. Geophys., 176, 1767–1784, https://doi.org/10.1007/s00024-018-2038-z.
Chen, W., Z. Jiang, L. Li, and P. Yiou, 2010: Simulation of regional climate change under the IPCC A2 scenario in southeast China. Climate Dyn., 36, 491–507, https://doi.org/10.1007/s00382-010-0910-3.
Dinku, T., E. N. Anagnostou, and M. Borga, 2002: Improving radar-based estimation of rainfall over complex terrain. J. Appl. Meteor., 41, 1163–1178, https://doi.org/10.1175/1520-0450(2002)041<1163:IRBEOR>2.0.CO;2.
Dong, Y., J. Li, J. Guo, Z. Jiang, Y. Chu, L. Chang, Y. Yang, and H. Liao, 2020: The impact of synoptic patterns on summertime ozone pollution in the North China Plain. Sci. Total Environ., 735, 139559, https://doi.org/10.1016/j.scitotenv.2020.139559.
Germann, U., G. Galli, M. Boscacci, and M. Bolliger, 2006: Radar precipitation measurement in a mountainous region. Quart. J. Roy. Meteor. Soc., 132, 1669–1692, https://doi.org/10.1256/qj.05.190.
Golian, S., S. Moazami, P.-E. Kirstetter, and Y. Hong, 2015: Evaluating the performance of merged multi-satellite precipitation products over a complex terrain. Water Resour. Manage., 29, 4885–4901, https://doi.org/10.1007/s11269-015-1096-6.
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., 176–177, 121–133, https://doi.org/10.1016/j.atmosres.2016.02.020.
Hou, A. Y., and Coauthors, 2014: The Global Precipitation Measurement Mission. Bull. Amer. Meteor. Soc., 95, 701–722, https://doi.org/10.1175/BAMS-D-13-00164.1.
Huffman, G. J., D. T. Bolvin, and E. J. Nelkin, 2017: Integrated Multi-satellitE Retrievals for GPM (IMERG) technical documentation. NASA Tech. Doc., 54 pp., https://pmm.nasa.gov/sites/default/files/document_files/IMERG_technical_doc_3_22_17.pdf.
Huffman, G. J., and Coauthors, 2020: Integrated Multi-satellite Retrievals for the Global Precipitation Measurement (GPM) Mission (IMERG). Satellite Precipitation Measurement, Springer, 343–353, https://doi.org/10.1007/978-3-030-24568-9_19.
Jiang, S.-h., L.-y. Wei, L.-l. Ren, L.-q. Zhang, M.-h. Wang, and H. Cui, 2023: Evaluation of IMERG, TMPA, ERA5, and CPC precipitation products over mainland China: Spatiotemporal patterns and extremes. Water Sci. Eng., 16, 45–56, https://doi.org/10.1016/j.wse.2022.05.001.
Joyce, R. J., J. E. Janowiak, P. A. Arkin, and P. Xie, 2004: CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hydrometeor., 5, 487–503, https://doi.org/10.1175/1525-7541(2004)005<0487:CAMTPG>2.0.CO;2.
Karaseva, M. O., S. Prakash, and R. M. Gairola, 2012: Validation of high-resolution TRMM-3B43 precipitation product using rain gauge measurements over Kyrgyzstan. Theor. Appl. Climatol., 108, 147–157, https://doi.org/10.1007/s00704-011-0509-6.
Kidd, C., and G. Huffman, 2011: Global precipitation measurement. Meteor. Appl., 18, 334–353, https://doi.org/10.1002/met.284.
Kim, T., T. Yang, L. Zhang, and Y. Hong, 2022: Near real-time hurricane rainfall forecasting using convolutional neural network models with Integrated Multi-satellitE Retrievals for GPM (IMERG) product. Atmos. Res., 270, 106037, https://doi.org/10.1016/j.atmosres.2022.106037.
Li, X., S. O, N. Wang, L. 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.
Liu, J., J. Du, Y. Yang, and Y. Wang, 2020: Evaluating extreme precipitation estimations based on the GPM IMERG products over the Yangtze River Basin, China. Geomatics Nat. Hazards Risk, 11, 601–618, https://doi.org/10.1080/19475705.2020.1734103.
Liu, M., X. Xu, A. Y. Sun, K. Wang, Y. Yue, X. Tong, and W. Liu, 2015: Evaluation of high-resolution satellite rainfall products using rain gauge data over complex terrain in southwest China. Theor. Appl. Climatol., 119, 203–219, https://doi.org/10.1007/s00704-014-1092-4.
Lu, D., and B. Yong, 2018: Evaluation and hydrological utility of the latest GPM IMERG V5 and GSMaP V7 precipitation products over the Tibetan Plateau. Remote Sens., 10, 2022, https://doi.org/10.3390/rs10122022.
Lu, X., M. Wei, G. Tang, and Y. Zhang, 2018: Evaluation and correction of the TRMM 3B43V7 and GPM 3IMERGM satellite precipitation products by use of ground-based data over Xinjiang, China. Environ. Earth Sci., 77, 209, https://doi.org/10.1007/s12665-018-7378-6.
Lu, X., G. Tang, X. Wang, Y. Liu, L. Jia, G. Xie, S. Li, and Y. Zhang, 2019: Correcting GPM IMERG precipitation data over the Tianshan Mountains in China. J. Hydrol., 575, 1239–1252, https://doi.org/10.1016/j.jhydrol.2019.06.019.
Ma, L., L. Zhao, L.-m. Tian, L.-m. Yuan, Y. Xiao, L.-l. Zhang, D.-f. Zou, and Y.-p. Qiao, 2019: Evaluation of the Integrated Multi-satellitE Retrievals for Global Precipitation Measurement over the Tibetan Plateau. J. Mt. Sci., 16, 1500–1514, https://doi.org/10.1007/s11629-018-5158-0.
Ma, Y., G. Tang, D. Long, B. Yong, L. Zhong, W. Wan, and Y. Hong, 2016: Similarity and error intercomparison of the GPM and its predecessor—TRMM multisatellite precipitation analysis using the best available hourly gauge network over the Tibetan Plateau. Remote Sens., 8, 569, https://doi.org/10.3390/rs8070569.
Mahmoud, M. T., M. A. Hamouda, and M. M. Mohamed, 2019: Spatiotemporal evaluation of the GPM satellite precipitation products over the United Arab Emirates. Atmos. Res., 219, 200–212, https://doi.org/10.1016/j.atmosres.2018.12.029.
Michaelides, S., V. Levizzani, E. Anagnostou, P. Bauer, T. Kasparis, and J. E. Lane, 2009: Precipitation: Measurement, remote sensing, climatology and modeling. Atmos. Res., 94, 512–533, https://doi.org/10.1016/j.atmosres.2009.08.017.
Ning, S., F. Song, P. Udmale, J. Jin, B. R. Thapa, and H. Ishidaira, 2017: Error analysis and evaluation of the latest GSMap and IMERG precipitation products over eastern China. Adv. Meteor., 2017, 1803492, https://doi.org/10.1155/2017/1803492.
Prakash, S., A. K. Mitra, D. S. Pai, and A. AghaKouchak, 2016: From TRMM to GPM: How well can heavy rainfall be detected from space? Adv. Water Resour., 88, 1–7, https://doi.org/10.1016/j.advwatres.2015.11.008.
Schneider, U., A. Becker, P. Finger, A. Meyer-Christoffer, M. Ziese, and B. Rudolf, 2014: GPCC’s new land surface precipitation climatology based on quality-controlled in situ data and its role in quantifying the global water cycle. Theor. Appl. Climatol., 115, 15–40, https://doi.org/10.1007/s00704-013-0860-x.
Sharifi, E., R. Steinacker, and B. Saghafian, 2016: Assessment of GPM-IMERG and other precipitation products against gauge data under different topographic and climatic conditions in Iran: Preliminary results. Remote Sens., 8, 135, https://doi.org/10.3390/rs8020135.
Shirmohammadi-Aliakbarkhani, Z., and A. Akbari, 2020: Ground validation of diurnal TRMM 3B42 V7 and GPM precipitation products over the northeast of Iran. Theor. Appl. Climatol., 142, 1413–1423, https://doi.org/10.1007/s00704-020-03392-0.
Sorooshian, S., K.-L. Hsu, X. Gao, H. V. Gupta, B. Imam, and D. Braithwaite, 2000: Evaluation of PERSIANN system satellite-based estimates of tropical rainfall. Bull. Amer. Meteor. Soc., 81, 2035–2046, https://doi.org/10.1175/1520-0477(2000)081<2035:EOPSSE>2.3.CO;2.
Tan, M. L., and Z. Duan, 2017: Assessment of GPM and TRMM precipitation products over Singapore. Remote Sens., 9, 720, https://doi.org/10.3390/rs9070720.
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, 152–167, https://doi.org/10.1016/j.jhydrol.2015.12.008.
Taniguchi, A., and Coauthors, 2013: Improvement of high-resolution satellite rainfall product for Typhoon Morakot (2009) over Taiwan. J. Hydrometeor., 14, 1859–1871, https://doi.org/10.1175/JHM-D-13-047.1.
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.
Tian, Y., G. S. Nearing, C. D. Peters-Lidard, K. W. Harrison, and L. Tang, 2016: Performance metrics, error modeling, and uncertainty quantification. Mon. Wea. Rev., 144, 607–613, https://doi.org/10.1175/MWR-D-15-0087.1.
Wang, X., Y. Ding, C. Zhao, and J. Wang, 2019: Similarities and improvements of GPM IMERG upon TRMM 3B42 precipitation product under complex topographic and climatic conditions over Hexi region, northeastern Tibetan Plateau. Atmos. Res., 218, 347–363, https://doi.org/10.1016/j.atmosres.2018.12.011.
Wei, G., H. Lü, W. T. Crow, Y. Zhu, J. Wang, and J. Su, 2018: Comprehensive evaluation of GPM-IMERG, CMORPH, and TMPA precipitation products with gauged rainfall over Mainland China. Adv. Meteor., 2018, 3024190, https://doi.org/10.1155/2018/3024190.
Wu, P., N. Christidis, and P. Stott, 2013: Anthropogenic impact on Earth’s hydrological cycle. Nat. Climate Change, 3, 807–810, https://doi.org/10.1038/nclimate1932.
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.
Xu, R., F. Tian, L. Yang, H. Hu, H. Lu, and A. Hou, 2017: Ground validation of GPM IMERG and TRMM 3B42V7 rainfall products over southern Tibetan Plateau based on a high-density rain gauge network. J. Geophys. Res. Atmos., 122, 910–924, https://doi.org/10.1002/2016JD025418.
Yao, T., and Coauthors, 2012: Third Pole Environment (TPE). Environ. Dev., 3, 52–64, https://doi.org/10.1016/j.envdev.2012.04.002.
Yong, B., Y. Hong, L. L. Ren, J. J. Gourley, G. J. Huffman, X. Chen, W. Wang, and S. I. Khan, 2012: Assessment of evolving TRMM-based multisatellite real-time precipitation estimation methods and their impacts on hydrologic prediction in a high latitude basin. J. Geophys. Res., 117, D09108, https://doi.org/10.1029/2011JD017069.
Yong, B., D. Liu, J. J. Gourley, Y. Tian, G. J. Huffman, L. Ren, and Y. Hong, 2015: Global view of real-time TRMM multisatellite precipitation analysis: Implications for its successor Global Precipitation Measurement Mission. Bull. Amer. Meteor. Soc., 96, 283–296, https://doi.org/10.1175/BAMS-D-14-00017.1.
Zhang, L., T. Kim, T. Yang, Y. Hong, and Q. Zhu, 2021: Evaluation of Subseasonal-to-Seasonal (S2S) precipitation forecast from the North American Multi-Model ensemble phase II (NMME-2) over the contiguous U.S. J. Hydrol., 603, 127058, https://doi.org/10.1016/j.jhydrol.2021.127058.
Zhang, Y., T. Ohata, D. Yang, and G. Davaa, 2004: Bias correction of daily precipitation measurements for Mongolia. Hydrol. Processes, 18, 2991–3005, https://doi.org/10.1002/hyp.5745.
Zhao, P., B. Li, J. Wang, H. Yang, P. Guo, and L. Gong, 2021: Accuracy evaluation and comparison of GPM IMERG and ERA5 precipitation products over complex terrain of Yunnan. Meteor. Sci. Technol., 49, 114–123.
Zhou, Z., B. Guo, W. Xing, J. Zhou, F. Xu, and Y. Xu, 2020: Comprehensive evaluation of latest GPM era IMERG and GSMaP precipitation products over mainland China. Atmos. Res., 246, 105132, https://doi.org/10.1016/j.atmosres.2020.105132.
Zhu, H., S. Chen, Z. Li, L. Gao, and X. Li, 2022: Comparison of satellite precipitation products: IMERG and GSMaP with rain gauge observations in northern China. Remote Sens., 14, 4748, https://doi.org/10.3390/rs14194748.
Zhu, Z., B. Yong, L. Ke, G. Wang, L. Ren, and X. Chen, 2018: Tracing the error sources of Global Satellite Mapping of Precipitation for GPM (GPM-GSMaP) over the Tibetan Plateau, China. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 11, 2181–2191, https://doi.org/10.1109/JSTARS.2018.2825336.
