1. Introduction
Numerous studies have consistently demonstrated that the intensity and frequency of extreme precipitation have been steadily increasing since 1950 (Seneviratne et al. 2012), with the potential for robust enhancement trends in the future, exerting profound impacts on both human society and natural ecosystems (Fischer et al. 2013; Kharin et al. 2013). Owing to high population densities, inadequate drainage systems, and unsustainable land use practices, the repercussions of extreme precipitation–induced torrential rain and flooding disasters will be notably pronounced in developing nations such as China (Zhang et al. 2005). Hence, a precise assessment of extreme precipitation exerts paramount importance for comprehending extreme climate events, rationalizing regional water resource management, conducting watershed hydrological simulation analyses, and devising disaster mitigation policies (Ren et al. 2022).
To delineate the features of extreme events, the Expert Team on Climate Change Detection and Indices (ETCCDI) has defined 27 extreme climate indices (Zhao et al. 2020), encompassing 10 extreme precipitation indices including maximum 1-day precipitation (RX1day) and maximum 5-day precipitation (RX5day). Due to its statistical advantages of low noise, strong comparability, and high significance, these indices have gained widespread international adoption (Frich et al. 2002). Prior research outcomes underscore an increasing trend of intense precipitation events across most global land regions (Seneviratne et al. 2021). On a continental scale, the intensity of extreme precipitation in Europe and North America generally displays an ascending trend (Zhou and Qian 2021), while in East Asia it exhibits a decreasing pattern from the southeast coastal regions toward the northwest interior, with insignificant interannual variations (Burton et al. 2012). In China, numerous scholars have conducted investigations into the spatial distribution and temporal changes of extreme precipitation indices utilizing observations from meteorological stations. Results indicate distinct regional variations in the characteristics of extreme precipitation events (Zhai et al. 2005; Zhou et al. 2016). Most of extreme precipitation intensity indices show a decreasing trend in a southwest–northeast belt from southwest China to northeast China, while they exhibit an increasing trend in eastern China and northwestern China.
Although the China Meteorological Administration has provided ground meteorological observations from over 2400 meteorological stations, and the spatial coverage of observational data has also significantly increased on a global scale, an uneven spatial distribution of stations, insufficient numbers of stations, and discontinuous record of stations continue to be prevalent issues. Alexander et al. (2006) indicates that global precipitation observation stations are densely concentrated in the midlatitude regions of Asia, Europe, and North America. However, significant gaps exist in meteorological observations across the entirety of Africa, the Brazilian plateau region in South America, and the central regions of Oceania. Bador et al. (2020) suggest that the spatial heterogeneity of meteorological station distribution could introduce errors in understanding regional extreme precipitation events. Simultaneously, due to limitations in data collection capabilities, different countries have varying update frequencies for observational data. Moreover, the number of gauge stations was reported to be smaller (Sun et al. 2018). Alexander et al. (2019) point out that even routine global monthly observation data used in creating Climate Hazards group Infrared Precipitation with Stations (CHIRPS-2.0) products have significantly decreased by approximately two-thirds since the 1980s (https://data.chc.ucsb.edu/products/CHIRPS-2.0/diagnostics/stations-perMonth-byRegion/pngs/all.station.count.CHIRPS-v2.0.png). This reduction is attributed to the erosion of observation networks and the limited reporting of extant observations in other countries, such as Africa and Central and South America. Lorenz and Kunstmann (2012) found that in South America, the number of active measurement stations decreased from 4267 to 390 between 1989 and 2006. Consequently, observational data have become insufficient for meeting research requirements on finer time scales. Therefore, given the limitations of observational data, the development of precipitation datasets with long time series and high spatial accuracy is deemed essential for climate change studies.
With the advancement of technology, a plethora of high spatiotemporal resolution global precipitation datasets applicable to hydrological and climatic research have been utilized. These datasets can be broadly categorized into three types: surface observation–based interpolated gridded precipitation datasets, satellite precipitation datasets, and reanalysis datasets. Due to distinct data sources and creation principles, these precipitation datasets possess varying degrees of applicability. High-resolution gridded data products obtained through the interpolation of ground-based precipitation data can achieve better spatial coverage. However, the utilization of interpolation methods introduces certain disparities among different grid results, and the extension of scale inherently carries considerable uncertainty (Li et al. 2022). High-resolution gridded precipitation datasets generated through satellite remote sensing inversion algorithms help compensate for the deficiencies in precipitation data in regions with complex terrains. These datasets provide globally extensive spatially homogeneous and temporally complete data. Nevertheless, due to the relatively late launch times of satellites, the temporal record length of satellite-based datasets is comparatively shorter than that of other precipitation datasets (Zhang et al. 2021). Reanalysis precipitation data products exhibit a longer temporal sequence than satellite data, rendering them widely applied in precipitation observation studies (Sun et al. 2018). In contrast to ground and satellite observations, reanalysis precipitation data are generated through the assimilation of diverse observational datasets by numerical weather prediction models, introducing large uncertainty in accuracy. Research indicates that reanalysis products manifest significant polarization in tropical regions, with portions being substantially drier or wetter. These products also exhibit poor modeling performance in tropical areas, arid regions, and regions with complex terrains (Alexander et al. 2020). Such deficiencies are likely attributed to changes in reanalysis product forecasting models and uncertainties in data assimilation (Inoue and Matsumoto 2004). Consequently, investigating the performance of various precipitation datasets in simulating extreme precipitation provides a foundational basis for accurately predicting flood risks brought about by extreme events. This exploration holds significant implications for the fusion and development of multisource precipitation datasets.
China, with its diverse climate and complex topography, serves as a typical region for assessing the applicability of different precipitation datasets (Peng et al. 2021). Various studies have accumulated results regarding the suitability assessment of different precipitation datasets for monitoring extreme precipitation in China. L. Zhang et al. (2022) compared the performance differences of 4-h scale datasets: Integrated Multi-satellitE Retrievals for GPM (IMERG), Global Satellite Mapping of Precipitation (GSMaP), the Modern-Era Retrospective Analysis for Research and Applications (MERRA), and the Climate Forecast System Reanalysis (CFSR) in mainland China across different time scales. They pointed out that reanalysis datasets exhibit inferior performance in analyzing extreme precipitation due to their inadequate representation of convective precipitation, compared to remote sensing datasets. Q. Wang et al. (2021) evaluated the accuracy of Climate Hazards Infrared Precipitation with Station data (CHIRPS), TRMM Multisatellite Precipitation Analysis (TMPA), the CPC morphing technique (CMORPH), and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) in capturing extreme precipitation in the humid southern regions of China. The results showed that CMORPH and TMPA can capture most instances of heavy and extreme precipitation. Liu et al. (2019) selected six commonly used satellite precipitation datasets and employed extreme precipitation indices along with statistical evaluation methods to compare their performance differences in the semiarid to semihumid climate transition zone of China. They found that TRMM 3B42V7, PERSIANN-CDR, and CMORPH-CRT exhibit better consistency with reference data at daily scale. At monthly or yearly scales, TRMM 3B42V7 outperformed other products. Evidently, different precipitation datasets exhibit varying capabilities in different regions of China, influencing the choice of datasets in various regions. However, current research primarily focuses on individual precipitation datasets or single-source datasets. Different studies adopt diverse precipitation indices, research methods, and spatiotemporal ranges, making it challenging to directly compare evaluation results across datasets. Furthermore, many studies employ relatively singular evaluation metrics, with most not encompassing the intensity, frequency, and duration of extreme precipitation events. Thus, these metrics fall short in providing a comprehensive description of extreme event characteristics.
Therefore, based on multisource precipitation datasets and the 10 defining extreme precipitation indices as defined by ETCCDI, this paper employs the spatiotemporal distribution results of extreme precipitation obtained from observations at over 2400 stations as ground truth. Through regional analysis, it examines the spatiotemporal characteristics and long-term trends of extreme precipitation in China. Furthermore, using the aforementioned outcomes as reference, it assesses the applicability of other multisource datasets and investigates the reasons behind the disparities in simulation performance among different data sources. The research findings can provide academic reference for the selection of precipitation datasets in various climatic regions of China and offer scientific references for improving precipitation dataset algorithms and data fusion among products.
2. Study area and data
a. Study area
China’s vast territory, intricate topography, and diverse climate lead to variations in the distribution of extreme precipitation (Guan et al. 2017), which in turn influences the accuracy of various precipitation datasets (Wei et al. 2018). However, this also positions China as a critical region for evaluating the quality of precipitation datasets. Focusing on mainland China, this study adopts the geographic division method proposed by S. Chen et al. (2013). Based on considerations of topography and precipitation characteristics, China is divided into eight major geographic regions: the inland Xinjiang region (XJ), Northwestern China (NW), Northeastern China (NE), Northern China (NC), the plain region of the Changjiang (Yangtze) River (CJ), Southeastern China (SE), Southwestern China (SW), and the Qinghai–Tibetan Plateau (TP). The climatic distribution of each region is illustrated in Fig. 1. The majority of meteorological stations are located in eastern and southern China. However, regions with complex terrain and sparse station coverage, such as the Qinghai–Tibetan Plateau characterized by high-altitude mountainous climate, the Xinjiang region dominated by the arid and semiarid climate due to the presence of the Tian Shan mountains, and the northeastern region with its harsh winter coldness, introduce higher uncertainties in precipitation datasets. Additionally, given the limitations of reference dataset resources from observation stations, the study does not include the Hong Kong Special Administrative Region, the Macau Special Administrative Region, and Taiwan.
The subregions of China and the distribution of the 2416 national meteorological stations utilized in the development of the daily gridded meteorological dataset of China (named CN05.1). China is divided into eight major geographic regions: the inland Xinjiang region (XJ), Northwestern China (NW), Northeastern China (NE), Northern China (NC), the plain region of the Changjiang (Yangtze) River (CJ), Southeastern China (SE), Southwestern China (SW), and the Qinghai–Tibetan Plateau (TP).
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
b. Data
The daily gridded meteorological dataset of China, referred to as CN05.1, is used as the reference dataset, which is developed based on the interpolation of observational data from over 2400 stations in China (Wu and Gao 2013). CN05.1 is interpolated using the thin plate spline function and angular distance weighting method. The thin plate spline function is a mathematical interpolation technique used for creating smooth surfaces from given data points, while the angular distance weighting method involves assigning weights to neighboring points based on their angular separation, commonly utilized in spatial interpolation to account for directional relationships between data points. The specific interpolation process of CN05.1 is as follows. The researchers initially utilized the ANUSPLIN software, incorporating longitude and latitude as variables in the thin-plate spline function, with elevation as a covariate for interpolating the climatological field. Subsequently, interpolation of the anomaly field was conducted, taking into account the angular and distance weights of the stations concerning the grid points. Ultimately, the two results were superimposed to yield the desired outcome. The highest spatial resolution of the grid is 0.25° × 0.25°, making it a representative regional gridded near-surface meteorological field dataset in China. The Multi-Source Weighted-Ensemble Precipitation (MSWEP) fusion product is developed by merging precipitation data from two ground observations, three reanalyses, and five satellites. It capitalizes on the advantages of multiple data sources, offering long temporal scales, short latency, and high spatial resolution. It has been widely applied globally. Additionally, the selected 11 precipitation datasets are all sourced from the Frequent Rainfall Observations on Grids (FROGS) dataset (Roca et al. 2019). This dataset comprises surface observations, satellite, and reanalysis precipitation datasets, all interpolated onto a 1° × 1° spatial grid. The time resolution is unified to daily scale, facilitating comparisons and evaluations among the datasets. For detailed information about the precipitation datasets, refer to Table 1. We can observe that the common time span covered by these datasets is from 2001 to 2019, with the lowest spatial resolution of the datasets being 1° × 1°.
Basic information of precipitation products.
Due to the differing time spans and spatial resolutions of various precipitation datasets, to ensure consistency in comparative analysis, the common time period covered by each precipitation dataset, 2001–19, is adopted as the study period. Prior to calculating extreme precipitation indices, both the CN05.1 and MSWEP datasets are interpolated to a 1° × 1° spatial grid. This adjustment facilitates subsequent comparative research.
3. Methodology
a. Extreme precipitation indices
The Extreme Temperature and Climate Indices Expert Team (ETCCDI)-defined extreme precipitation intensity indices (PRCPTOT, SDII, RX1day, RX5day, R95p, R99p) were selected to characterize extreme precipitation patterns in China. The specific definitions of these indices are presented in Table 2.
Definition of extreme precipitation indices.
b. Evaluation indicators of precipitation products
Conducting a comprehensive assessment of the applicability of precipitation datasets is important and necessary. In this study, the error metric “distance between indices of simulation and observation” (DISO) developed by Lei et al. (2022) was adopted to quantitatively and comprehensively evaluate the datasets. DISO comprises four commonly used statistical indicators for accuracy assessment: normalized root-mean-square error (NRMSE), normalized mean absolute error (NMAE), relative bias (RB), and the Pearson correlation coefficient (CC). Detailed calculation formulas for four statistical indicators and DISO are presented in Tables 3 and 4, respectively.
Evaluation indicators of precipitation products. In the formulas, n represents the sample size, Ei represents the calculated results of extreme precipitation indices for various precipitation datasets, Oi represents the calculated results of extreme precipitation indices for the reference dataset CN05.1, and
Distance between indices of simulation and observation (DISO) indicator.
Among these, NRMSE, NMAE, RB, and other indicators can assess the level of error between precipitation datasets and reference datasets, with values of 0 indicating no error, and larger values indicating greater deviations. On the other hand, CC evaluates the linear correlation between precipitation datasets and reference datasets, with a value of 0 indicating no correlation, and a value closer to 1 indicating a stronger correlation between the two datasets. When the values of the conventional statistical indicators are smaller than 0.5, DISO < 1.0, indicating higher accuracy of the precipitation dataset’s simulation of extreme precipitation. Conversely, when the values of these indicators are greater than 1.0, DISO > 2.0, suggesting a very low reliability in simulating extreme precipitation in the precipitation dataset. Hence, DISO = 1.0 serves as the threshold for assessing the applicability of precipitation datasets. When DISO < 1.0, the dataset is considered to have good applicability, and as DISO approaches 0, the applicability improves. When DISO > 1.0, the dataset is considered to have poor applicability, and as DISO values increase, the applicability becomes less ideal.
c. The kernel density estimation
d. Calculation of precipitation characteristics
To better understand the precipitation characteristics in different intensities represented by multiple datasets, two indicators, namely the annual accumulated precipitation amount and the occurrence frequency of precipitation events of varying intensities, are employed. This method follows the approach utilized in S. Ma et al. (2015) and Ma et al. (2017). First, a precipitation event is defined as a day with daily precipitation equal to or greater than 0.1 mm day−1, then for a given grid and year, daily precipitation is divided into 100 intensity bins, each with a size of 1 mm day−1. Subsequently, the accumulated precipitation and frequency at each bin are calculated. The accumulated precipitation in each intensity bin is the accumulated precipitation amount over the precipitation days within the corresponding intensity range, and the precipitation frequency in each intensity bin is the ratio (in %) of the number of days with daily precipitation rates falling within the respective intensity range to the total number of days with available data. Finally, the multiyear average yearly accumulated amount and frequency are calculated for each grid. The regional averages for each intensity bin were calculated by averaging all the grid cells within the region, using area as the weight.
e. Trend analysis and significance test
This study employs Sen’s slope method to detect the changing trends of extreme precipitation indices from 2001 to 2019 (Sen 1968). The nonparametric Mann–Kendall statistical method is used to test the significance of the changing trends at a significance level of 5% (Mann 1945). Since the Mann–Kendall method does not require assumptions about data distribution, it exhibits strong adaptability to nonnormally distributed meteorological data and has been widely applied in climate research (Perkins-Kirkpatrick and Lewis 2020; Wang et al. 2020). The formula can be found in the aforementioned studies.
4. Results
a. Spatial distribution of extreme precipitation intensity indices in China from 2001 to 2019
1) Spatial distribution of multiyear averages for various extreme precipitation indices
The spatial distribution characteristics of various intensity indices are generally similar, with high values primarily distributed in southern China, especially along the southeastern coast. The high values gradually decrease from the southeastern coast to the northwestern inland regions.
Figure 2 illustrates the spatial distribution of the multiyear average precipitation amount (PRCPTOT) from the reference dataset CN05.1 and the deviations of other datasets relative to CN05.1. From Fig. 2, it can be observed that in comparison to CN05.1, other station-based observation datasets such as CPC and GPCC generally underestimate the precipitation amount (PRCPTOT), with median biases of −29.36 and −11.82 mm throughout China, respectively. The bias spatial distribution of remote sensing datasets exhibits significant variations. CMORPH and GSMaP show negative biases in the Tianshan Mountains of Xinjiang region, the Qinghai–Tibetan Plateau, Southwestern China, Southeastern China, and the plain region of the Changjiang (Yangtze) River, while displaying notable positive biases in the southern part of Xinjiang region and Northern China. GPCP, IMERG, and PERSIANN all detect high bias values in the southern part of the Qinghai–Tibetan Plateau. The multisource fusion dataset MSWEP generally presents low bias values in the plain region of the Changjiang (Yangtze) River and Southeastern China. The remaining reanalysis datasets exhibit pronounced high bias values in most regions across the country. Among them, CFSR and MERRA-2 significantly overestimate precipitation in Southwestern China, Southeastern China, and the plain region of the Changjiang (Yangtze) River, with bias values exceeding 400 mm.
(a) Multiyear mean PRCPTOT in the CN05.1 dataset and (b)–(m) bias in PRCPTOT from the CN05.1 dataset in other rainfall datasets. The number inserted in each panel is the median bias in China for respective datasets. bias is calculated by subtracting the mean indices in the reference dataset (CN05.1) from those in other datasets.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
Figure 3 illustrates the multiyear average precipitation intensity (SDII) from the reference dataset CN05.1 and the deviations of various datasets from CN05.1. Due to the varying detection capabilities of precipitation datasets, there are significant differences in the SDII detection results among the datasets compared to PRCPTOT. From Fig. 3, it can be observed that among the observation datasets, CPC mainly overestimates SDII in the inland Xinjiang region and the Qinghai–Tibetan Plateau, while underestimating it in the Northern and Southwestern China regions. GPCC generally exhibits high bias values across the country. The high bias value regions of remote sensing datasets are mostly located in the southern part of the Qinghai–Tibetan Plateau, Southwestern China, and Southeastern China regions, while the low bias value regions are primarily distributed in Northern China. In addition, CMORPH and GSMaP also show strong high bias values in the Northeastern China region, and PERSIANN’s low bias value regions are widespread in Xinjiang region, Northwestern China, Northeastern China, Northern China, and the plain region of the Changjiang (Yangtze) River. MSWEP and reanalysis datasets generally present low bias values in Northern China, while the high bias value regions in the inland Xinjiang region and Northwestern China areas are broader than those in remote sensing datasets. It is worth noting that MERRA-2 exhibits particularly strong overestimation characteristics in the inland Xinjiang region, the Qinghai–Tibetan Plateau, Northwestern China, and the plain region of the Changjiang (Yangtze) River.
(a) Multiyear mean SDII in the CN05.1 dataset and (b)–(m) bias in SDII from the CN05.1 dataset in other rainfall datasets. The number inserted in each panel is the median bias in China for respective datasets. bias is calculated by subtracting the mean indices in the reference dataset (CN05.1) from those in other datasets.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
RX1DAY is an important index describing annual extreme rainfall values. Frequency analysis is conducted based on RX1DAY, and the optimal probability distribution function is fitted. This forms the foundation for extreme rainfall design and risk analysis (Gu et al. 2022). RX5DAY, on the other hand, holds significance in terms of accumulated extreme rainfall. Figure 4 indicates that for RX1DAY, the deviation patterns of various precipitation datasets exhibit significant spatial differences. Among the observational datasets, CPC and GPCC’s overestimation and underestimation areas are uniformly distributed across the country. Among the remote sensing datasets, GPCP and IMERG exhibit underestimation in the inland Xinjiang region, Northwestern China, and Northern China regions, while they show significant overestimation in the southern part of the Qinghai–Tibetan Plateau, Southeastern China, and Southwestern China regions. CMORPH and GSMaP show notable underestimation in Northern China, while they generally indicate overestimation in the Xinjiang, Qinghai–Tibetan Plateau, Southwestern China, and Southeastern China regions. PERSIANN differs from other datasets, showing a wide underestimation region encompassing the Xinjiang region, Northwestern China, Northeastern China, Northern China, and the plain region of the Changjiang (Yangtze) River, with a concentrated overestimation in the Qinghai–Tibetan Plateau. The low estimation region of the multisource fusion dataset MSWEP is mainly distributed in Northern China, the plain region of the Changjiang (Yangtze) River, and the Southeastern China region, while the high estimation area is concentrated in the eastern part of the Qinghai–Tibetan Plateau and the Northeastern China region. The underestimation region of reanalysis datasets is relatively smaller compared to MSWEP. Among them, JRA-55 exhibits the most similar distribution pattern to MSWEP, while other datasets generally show significant overestimation across the country. The spatial pattern of RX5DAY, shown in Fig. S1 in the online supplemental material, shares highly consistent bias spatial distribution characteristics with RX1DAY, but the bias values of reanalysis datasets are significantly higher compared to RX1DAY.
(a) Multiyear mean value of RX1DAY in the CN05.1 dataset and (b)–(m) bias in RX1DAY from the CN05.1 dataset in other rainfall datasets. The number inserted in each panel is the median bias in China for respective datasets. bias is calculated by subtracting the mean indices in the reference dataset (CN05.1) from those in other datasets.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
Although RX1DAY and RX5DAY are widely used in the field of hydrometeorology, their approach of selecting annual extremes neglects interannual variations caused by climate change. This might lead to the loss of valuable information in extreme precipitation sequences (Ding et al. 2011). Setting rainfall thresholds can reflect the statistical characteristics of precipitation sequences and improve extreme precipitation sampling methods (Liu et al. 2017). R95P and R99P are the most commonly used threshold indices. Figure 5 and Fig. S2 show that due to their similar response to extreme precipitation, the spatial distribution patterns of R95P and R99P in various precipitation datasets resemble those of RX1DAY and RX5DAY. However, the degree of deviation between different precipitation datasets has increased, especially in the positive bias of reanalysis products, particularly CFSR and MERRA-2, compared to other datasets.
(a) Multiyear mean value of R95P in the CN05.1 dataset and (b)–(m) bias in R95P from the CN05.1 dataset in other rainfall datasets. The number inserted in each panel is the median bias in China for respective datasets. bias is calculated by subtracting the mean indices in the reference dataset (CN05.1) from those in other datasets.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
2) Spatial distribution of the error evaluation index DISO
By combining the index error indicator DISO of observations and reference datasets with the previous research results, a more intuitive reflection of the dataset’s applicability to various indices can be achieved. In the article, the DISO value of each grid point represents the mean value for the period 2001–19. Overall, most datasets exhibit higher detection errors for the inland Xinjiang region and the Qinghai–Tibetan Plateau. Observational datasets demonstrate better applicability to most indices, while the applicability of other datasets varies depending on the index.
As shown in Fig. 6, for PRCPTOT, the mean DISO values of observational datasets, remote sensing datasets, and multisource fusion datasets are generally less than 1, indicating a good level of applicability. Among them, CMORPH and MSWEP have the lowest mean DISO values among all datasets, demonstrating relatively good applicability across most regions of the country. Particularly, they exhibit good applicability in Northwestern China areas and the central part of the Qinghai–Tibetan Plateau. However, in the marginal areas of the Xinjiang region and the Qinghai–Tibetan Plateau, there are prominent grid points with DISO values exceeding 2. On the other hand, CPC, GPCC, GPCP, IMERG, and PERSIANN exhibit similar spatial distribution patterns of DISO indicators. High DISO values are found only in the marginal areas of the inland Xinjiang region and the Qinghai–Tibetan Plateau, while DISO values in other areas are around 1. GSMAP differs from other remote sensing datasets, with a significant concentration of grid points with DISO values above 2.5 in the southern part of the inland Xinjiang region. Reanalysis datasets show particularly high DISO values in the inland Xinjiang region and the Qinghai–Tibetan Plateau.
Spatial distribution of DISO for PRCPTOT over China. The number inserted in each panel is the median DISO in China for the respective datasets.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
Compared with PRCPTOT, the low DISO value area of SDII has significantly expanded (Fig. 7). The remote sensing dataset CMORPH performs the best in this regard, exhibiting DISO values consistently below 1 across most regions of the country. The majority of other datasets also show relatively high applicability nationwide. CPC, GPCC, MSWEP, and GPCP only exhibit partial DISO high values in the marginal areas of the Qinghai–Tibetan Plateau, indicating that the errors of most datasets in relation to SDII fall within an applicable range. For the remote sensing dataset GSMaP, a prominent and concentrated DISO high value area appears in the northwestern part of the Qinghai–Tibetan Plateau. Reanalysis datasets, especially CFSR and MERRA-2, show DISO high values in the Northwestern China regions, the inland Xinjiang region, and the Qinghai–Tibetan Plateau, suggesting that these datasets are less applicable for research in the Northwestern China, the inland Xinjiang region, and the Qinghai–Tibetan Plateau of China.
Spatial distribution of DISO for SDII over China. The number inserted in each panel is the median DISO in China for the respective datasets.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
RX1DAY (Fig. 8) and RX5DAY (Fig. S3) exhibit similar spatial patterns in DISO distribution. Except for the CFSR and MERRA-2 datasets, most datasets still perform well in the Northeastern China, Northern China, the plain region of the Changjiang (Yangtze) River, and Southeastern China regions of China. However, the CFSR dataset shows widespread DISO high values in the inland Xinjiang region, the Qinghai–Tibetan Plateau, Southwestern China, and Southeastern China regions, while MERRA-2 demonstrates DISO high values greater than 1.5 in most parts of the country.
Spatial distribution of DISO for RX1DAY over China. The number inserted in each panel is the median DISO in China for the respective datasets.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
Compared with RX1DAY and RX5DAY, the applicability of various datasets significantly decreases for the R95P and R99P indices. For R95P (Fig. 9), except for the CFSR and MERRA-2 datasets, most datasets can generally demonstrate good performance levels in the Northeastern China, Northern China, the plain region of the Changjiang (Yangtze) River, and Southeastern China. However, for R99P (Fig. S4), all datasets show significant precipitation errors, and the simulation performance of R99P is unsatisfactory across the country.
Spatial distribution of DISO for R95P over China. The number inserted in each panel is the median DISO in China for the respective datasets.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
The calculation results of the skill score shown in Fig. 10 confirm the conclusions of the DISO. Clearly, there is a significant gap in the performance of reanalysis datasets compared to rain gauge–based and satellite-based datasets, especially when calculating PRCPTOT, R95P, and R99P, where skill scores are relatively low. Satellite-based datasets exhibit higher skill scores compared to rain gauge–based datasets, with some satellite-based datasets such as CMORPH, GPCP, and IMERG even outperforming the two rain gauge–based datasets in skill scores. This outcome, similar to the DISO results, may suggest that satellite-based datasets share similar distribution and variability characteristics with the reference datasets.
The skill scores of 12 precipitation datasets.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
In summary, observational datasets, remote sensing datasets, and MSWEP all exhibit good performance for PRCPTOT and SDII. Among them, CMORPH performs best for PRCPTOT and SDII. Except for CFSR and MERRA-2, most datasets show higher applicability for the RX1DAY, RX5DAY, and R95P indices in the Northeastern China, Northern China, the plain region of the Changjiang (Yangtze) River, and Southeastern China regions of China. However, all datasets perform poorly for R99P.
b. Spatial distribution of extreme precipitation intensity index trends in China from 2001 to 2019
1) Spatial distribution of extreme precipitation trends
Figure 11 indicates that, for the trend of PRCPTOT changes, CN05.1 results show a widespread increasing trend in annual precipitation across China. Similar spatial distributions are observed in other datasets, with a general increasing trend in the Northeastern China region. Most grid points exhibit significant trends. Similarly increasing trends are also evident in the plain region of the Changjiang (Yangtze) River, Southeastern China, and the eastern edge of the Qinghai–Tibetan Plateau. Among them, the increase in annual precipitation in the plain region of the Changjiang (Yangtze) River may be primarily attributed to the midscale convective systems during the rainy season (Li et al. 2023). Conversely, the western part of the Southwestern China region and the southern part of the Northern China region show decreasing trends. In the case of reanalysis data, CFSR and MERRA-2 exhibit significant trends in most areas, while ERA5 shows less distinct trends. As seen in Fig. 12, the spatial distribution of SDII trends is similar to that of PRCPTOT, but with smaller trend coefficients. The increasing trend region in the plain region of the Changjiang (Yangtze) River has also diminished. Unlike PRCPTOT, SDII demonstrates a weaker decreasing trend in the Southeastern China region in most datasets.
Spatial distribution of trend coefficient for PRCPTOT over China; the solid dots indicate the trends are significant at the 95% level. The numerical value at the bottom left represents the proportion of grids that show a decreasing or increasing trend.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
Spatial distribution of trend coefficient for SDII over China; the solid dots indicate the trends are significant at the 95% level. The numerical value in the bottom left represents the proportion of grids that show a decreasing or increasing trend.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
As depicted in Fig. 13 and Fig. S5, the analysis results of RX1DAY for various precipitation datasets reveal cross-distributed increasing and decreasing regions nationwide. Prominent increasing regions are mainly distributed in the Northeastern China and the plain region of the Changjiang (Yangtze) River. GSMaP, CFSR, and MERRA-2 exhibit strong increasing trends in the plain region of the Changjiang (Yangtze) River, Southeastern China, and Southwestern China regions. On the other hand, ERA5 and MSWEP primarily exhibit decreasing trends in the Northwestern China, Northern China, the plain region of the Changjiang (Yangtze) River, and the Southwestern China region. The trend patterns of other datasets are similar to the reference dataset. The trends for RX5DAY are similar to those for RX1DAY, with a higher number of significantly significant points compared to RX1DAY.
Spatial distribution of trend coefficient for RX1DAY over China; the solid dots indicate the trends are significant at the 95% level. The numerical value in the bottom left represents the proportion of grids that show a decreasing or increasing trend.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
From Fig. 14 and Fig. S6, it can be observed that the trends of R95P are consistent in spatial distribution across datasets, generally showing increasing trends in Northeastern China and the plain region of the Changjiang (Yangtze) River, and decreasing trends in Northern China and the southern part of Southwestern China. For R99P, a significant number of datasets show regions with no trend, while CFSR and MERRA-2 still exhibit significant increasing trends in the plain region of the Changjiang (Yangtze) River and the Southeastern China region.
Spatial distribution of trend coefficient for R95P over China; the solid dots indicate the trends are significant at the 95% level. The numerical value in the bottom left represents the proportion of grids that show a decreasing or increasing trend.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
2) PDFs
Probability density functions can provide a clearer and more intuitive representation of the spatial distribution information of index trend results for each dataset, as shown in Fig. 15.
Probability density functions of trend coefficient of extreme precipitation indices for all grid points in China. The horizontal axis represents the trend values range of each dataset across all grids, while the vertical axis represents the probability density of the corresponding values.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
From the PDFs results of PRCPTOT trend coefficients, CPC, GPCC, and GPCP show relatively good trend estimation capabilities. Except for ERA5, the remaining datasets exhibit right-skewed features and significant overestimation compared to the reference dataset. The overestimation is most pronounced in MERRA-2 and CFSR. Regarding SDII, only PERSIANN, JRA-55, ERA5, and the reference data have well-fitting probability density functions. The other datasets have larger errors, particularly CFSR and MERRA-2, which exhibit a pronounced right-skewed distribution.
The PDFs results of various datasets for RX1DAY and RX5DAY are relatively consistent. Remote sensing datasets like PERSIANN and GPCP provide a more accurate grasp of the distribution characteristics of the PDFs. Other datasets tend to underestimate around the zero value. CMORPH, GSMaP, CFSR, and MERRA-2 exhibit significant simulation deviations and their distribution curves are notably right-skewed, indicating substantial overestimation.
The PDFs of R95P and R99P show significant differences. For R95P, besides the gauge-based datasets, PERSIANN also effectively captures the trend changes of R95P. Reanalysis datasets like ERA5 and MSWEP show left-skewed patterns, indicating general underestimation. Other datasets maintain right-skewed features, implying widespread overestimation. In the case of R99P, ERA5, IMERG, and GPCP exhibit good performance, while GPCC, CFSR, and MERRA-2 have trend values more dispersed along the coordinate axis, resulting in flatter probability density functions compared to the reference data.
In summary, except for R99P, intensity indices generally exhibit strong increasing trends in the Northeastern China and Yangtze River middle–lower reaches regions, while showing strong decreasing trends in the southern parts of Northern China and the Southwestern China region. Gauge-based datasets perform well for PRCPTOT and R95P. Additionally, GPCP exhibits good performance for PRCPTOT, RX1DAY, and RX5DAY indices, ERA5 performs well for SDII and R99P indices, and PERSIANN shows good performance for SDII, RX1DAY, RX5DAY, and R95P indices.
c. Time series of extreme precipitation intensity indices in China from 2001 to 2019
Figure 16 displays the time series of various extreme precipitation indices in China for the period of 2001–19. The calculated long-term trends of each time series are presented in Fig. 17. CN05.1’s results indicate that during 2001–19, all extreme precipitation indices exhibit clear increasing trends. Other datasets can reproduce the varying trends of different indices during this period, but with varying degrees of overestimation or underestimation.
Temporal variation of extreme precipitation indices for different precipitation products from 2001 to 2019. The annual precipitation of each dataset from 2001 to 2019 is calculated using area-weighted average method.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
Trends of extreme precipitation indices for different precipitation products from 2001 to 2019. Annual values are calculated based on the annual area-weighted average method. The crosses indicate the trends are significant at the 90% level, while the asterisks indicate the trends are significant at the 95% level.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
Overall, gauge-based datasets such as CPC and GPCC, along with the reference dataset CN05.1, show the best agreement and can better capture the interannual variability of extreme precipitation indices. Reanalysis datasets generally exhibit overestimation for all indices, while the overestimation or underestimation characteristics of remote sensing datasets vary depending on the specific index.
For PRCPTOT, the interannual variations of various datasets are relatively consistent with the reference dataset. Notably, the gauge-based datasets, remote sensing datasets, multisource fusion datasets, and CN05.1 exhibit strong agreement, while the reanalysis datasets show varying degrees of overestimation. The overestimation is particularly severe in MERRA-2 and CFSR, and relatively lighter in ERA5 and JRA-55.
Regarding SDII, the overestimation or underestimation characteristics among datasets are less distinct compared to other indices. MSWEP shows good agreement with CN05.1, while both CPC and GPCC exhibit overestimation. Among the remote sensing datasets, except for PERSIANN’s underestimation, the others show varying degrees of overestimation with similar temporal variations. GSMaP demonstrates the most pronounced overestimation, deviating from the reference data’s fluctuation trend during 2006–10 and 2014–19. Reanalysis datasets are generally characterized by overestimation, with MERRA-2 showing the most severe overestimation.
RX1DAY and RX5DAY exhibit similar variation patterns. CPC and MSWEP align closely with the reference dataset in terms of interannual variations. Among the remote sensing datasets, most exhibit varying degrees of overestimation except for PERSIANN, with GSMaP showing the most noticeable overestimation. It exhibits notably high values during 2007–10 and 2015–19. Among the reanalysis datasets, ERA5, CFSR, and MERRA-2 tend to overestimate the indices significantly.
For R95P and R99P, the temporal variation curves are similarly shaped. Apart from reanalysis datasets, the others closely match the variation curve of CN05.1. Among the reanalysis datasets, JRA-55 closely resembles the reference data’s changes, with ERA5 showing slight overestimation. CFSR and MERRA-2 generally exceed other curves and exhibit an increasing deviation from the reference dataset since 2008.
The prevalent overestimation of reanalysis datasets can also be observed in the multiyear average histogram of annual accumulated precipitation amount and occurrence frequency (Fig. 18). For annual accumulated precipitation amount, almost all datasets resemble a gamma distribution, with a peak near the intensity of 2–4 mm day−1 and a long tail toward high precipitation rates. However, there are significant performance differences among different types of datasets. Except for PERSIANN, both rain gauge–based and satellite-based datasets generally underestimate precipitation below the intensity of 15 mm day−1, while reanalysis datasets, especially MERRA-2 and CFSR, tend to produce too much heavy precipitation. Similarly, for occurrence frequency, MERRA-2 and CFSR also overestimate the probability of heavy precipitation events.
Multiyear average (2001–19) histograms of yearly (a) accumulated precipitation amount (mm) and (b) frequency of occurrence (%) as a function of daily precipitation intensity from 0.1 to 100 mm day−1 (bin size is 1 mm day−1), averaged over China.
Citation: Journal of Hydrometeorology 25, 3; 10.1175/JHM-D-23-0162.1
5. Discussion
a. Comparison with previous research
The research results have been compared with previous works. Regarding spatial distribution, the computed results of intensity indices are consistent with Li et al. (2022) and Yu et al. (2015). In terms of trend analysis, the computed results align closely with Shi et al. (2018), Xiao et al. (2017), while showing notable enhancements in the increasing trends of intensity indices PRCPTOT, RX5DAY, and R95P for the plain region of the Changjiang (Yangtze) River and Southeastern China regions. This consistency with the findings of Ning and Qian (2009) and Ren et al. (2015) suggests that this might be influenced by the strengthening of the South China Sea summer monsoon, which enhances convection over southern China. Concerning interannual variability, the fluctuations in various indices are generally in line with the research results of Q. Wang et al. (2021), Zhou et al. (2016), and Fu et al. (2013). However, there exist significant differences in the indices for RX1DAY and RX5DAY. Previous research indicated significant upward fluctuations in RX1DAY and RX5DAY indices, while this study’s results indicate a dominant downward fluctuation trend between 2001 and 2014, followed by a pronounced increase after 2014. These contrasting conclusions are primarily due to the differences in the selected time periods among the studies, highlighting the importance of not only focusing on overall study period results but also paying attention to the characteristics of changes in different periods.
b. Error factors of precipitation datasets in complex terrain regions
From a spatial distribution perspective, significant errors exist in the extreme precipitation index monitoring results of various datasets for the inland Xinjiang region and the Qinghai–Tibetan Plateau region. In fact, apart from China, research in regions with complex terrains such as Pakistan (Nadeem et al. 2022), India (Prakash et al. 2015), and the Andes Mountains in Peru (Mantas et al. 2015) has also indicated that extreme precipitation monitoring results are more accurate in plain areas compared to mountainous regions. This is generally attributed to several reasons: first, the terrain-induced complexity of rainfall increases the difficulty of detecting precipitation in datasets (Liu et al. 2019); second, weather radar can be obstructed by terrain, and the reflectivity sensitivity of satellite sensors might significantly decrease in intricate terrains (Hu et al. 2013); third, due to the challenges and high costs associated with establishing meteorological stations in mountainous areas, along with sparse station distribution, there is insufficient capacity to calibrate remote sensing and reanalysis datasets in the absence of observational data (Tan and Duan 2017). However, the physical mechanisms through which terrain influences the accuracy of precipitation datasets remain unclear, and while precipitation dataset algorithms have improved with version iterations, the challenges posed by complex terrains have yet to be resolved (Ren et al. 2019), necessitating further research.
c. Exploring error factors in precipitation dataset index monitoring
Each dataset demonstrates varying levels of applicability across different indices. Further analysis of the significant performance differences among various remote sensing datasets suggests that the causes behind this phenomenon might be related to the sensors and retrieval algorithms, considering that the remote sensing datasets used in the article generally undergo bias correction. Taking PERSIANN as an example, in intensity indices, PERSIANN often exhibits larger deviations compared to the reference datasets. This is because PERSIANN primarily relies on infrared sensors rather than microwave data to detect precipitation, where the principle of infrared sensors is to capture the relationship between cloud-top temperature and precipitation. This is less direct than the microwave method sensitive to precipitation particles (Xia et al. 2022), which is commonly used by the other datasets that combine both infrared and microwave estimations. Additionally, PERSIANN employs an artificial neural network method for retrieval, with the training dataset starting from the end of 2011. The model parameters derived from the neural network remain constant during the inversion process (Chen and Gao 2018), which might not favor accurate rainfall estimation. Many studies also corroborate PERSIANN’s larger errors through comparisons (Zhu et al. 2016). Furthermore, the absence of available verification rain gauge data could also contribute to the errors in satellite precipitation datasets (Chen et al. 2020).
Reanalysis datasets generally exhibit an overestimation characteristic. In fact, numerous studies by scholars have indicated that reanalysis datasets tend to overestimate precipitation (Ding et al. 2022; Song et al. 2021). However, attributing the precipitation differences shown in reanalysis data to their own assimilation techniques and sea surface temperature (SST) boundary conditions is challenging (Jones et al. 2021; Lin et al. 2014). Currently, it is mostly believed that the substantial system differences within the hydrological cycle lead to errors in reanalysis data (Hua et al. 2019; Quagraine et al. 2020). In terms of trend analysis, although ERA5 demonstrates good performance in the trend coefficients of SDII and R99P indices, most reanalysis datasets generally exhibit significant errors in estimating the trends of extreme precipitation indices. In terms of time series analysis, the CFSR and MERRA-2 datasets exhibit substantial increases around 2008. The pronounced ascending extreme indices observed in these two datasets are consistent with findings from a global-scale study (Alexander et al. 2020), whereas other precipitation products fail to capture these trends. It is plausible to attribute these trends to data artifacts, possibly contributing to the identification of nonstationary systematic errors as reported in Funk et al. (2019). Consequently, in practical applications, reanalysis datasets are often considered unsuitable for trend analysis (Trenberth et al. 2008). The introduction of ground-based observational data for calibration is crucial to enhancing the accuracy of reanalysis data. In addition, reanalysis datasets have merged irregular observations and models that encompass many physical and dynamical processes (Sun et al. 2018), the uncertainties of the assimilating observations and the retrieval, downscaling, and data fusion algorithms significantly affect the accuracy of reanalysis products (Miao et al. 2021). Therefore, when using reanalysis datasets, it is important to be aware of the significant uncertainty inherent in them.
Furthermore, we have also noted the good monitoring capability of the fusion dataset MSWEP in capturing the temporal variations of extreme precipitation indices. In real-life scenarios, multisource data fusion is also flourishing as a new direction for spatial precipitation estimation (T. Zhang et al. 2022). Many researchers are generating precipitation datasets with superior performance based on the study area’s requirements. For instance, Sun et al. (2020) combined CPC, CMORPH, and MERRA-2 with observed temperature and precipitation datasets of different time scales to create the high spatiotemporal resolution dataset CLDAS covering all of China. Evaluation indicates that this dataset can fulfill the needs of hydrological modeling research. Considering the generally better performance of fusion datasets compared to single-source data, they have considerable potential for finer hydrological forecasting. However, inherent issues also exist. On one hand, fusion datasets inherit the deficiencies of individual precipitation datasets while combining their advantages. Taking MSWEP as an example, although this dataset exhibits good representation of index temporal variations in the text, due to the fusion of reanalysis precipitation dataset ERA5, it shares the consistent overestimation characteristics in the spatial distribution of intensity indices with the reanalysis dataset. On the other hand, the diversity of data sources and fusion techniques significantly increases the uncertainty of fusion data products (Xiong et al. 2021). Thus, the theoretical advantages of multisource fusion datasets require substantial future research efforts for validation.
In conclusion, there is no single precipitation dataset that can perform well for all regions and various extreme precipitation indices. Therefore, in extreme precipitation research, researchers need to select precipitation datasets based on the applicability of different datasets to different regions’ extreme precipitation indices.
d. Uncertainty in observational datasets
Observational datasets of site-measured data are typically considered as “truth” and are used for accuracy assessment of precipitation datasets. However, in real-world scenarios, due to instrument design and geographical location, site-measured data can easily be influenced by various sources of error (Lanza and Vuerich 2009). Among these, wind introduces particularly evident uncertainty in precipitation measurements (Shedekar et al. 2016). On one hand, wind causes vortices and turbulence around the instrument, leading to undercatchment of precipitation or adherence to the instrument’s inner walls. On the other hand, for raindrops adhering to the instrument’s inner walls, wind can accelerate the evaporation of this moisture, resulting in underestimation of precipitation by the instrument. Further research by Sieck et al. (2007) suggests that the impact of wind on measured data depends on environmental wind speed, raindrop size, and instrument design.
Many studies also acknowledge the underestimation of precipitation by rain gauges. Groisman et al. (1991) state that each recorded rain measurement in the Russian region might be associated with a loss of around 0.2 mm, with an average loss of 0.15 mm for rain–snow mixed precipitation. Research by Yang and Ohata (2001) indicates that the corrected annual precipitation for Siberia is 30–330 mm higher than the original data. A study on China (Ye et al. 2004) reveals an underestimation of 19% in actual precipitation. Similarly, Y. Ma et al. (2015) suggest that annual precipitation in Tibet and its surrounding areas is underestimated by 27%. These findings to some extent undermine the reliability of precipitation dataset evaluation results. Considering the potential for the underestimation of observational data to further misguide the direction of improvement in precipitation datasets, it is necessary to conduct in-depth research on observational data errors and error corrections in the future.
e. Limitation and improvements
It is evident that we have not addressed some issues in this paper. In terms of the spatial match of the data, we broadly assume that the indices calculated from each product are meaningful and directly comparable between the product types. But in fact, various products may not match in space. The rain gauge–based datasets we use are interpolated based on point data, assuming that point-scale precipitation data are equivalent to grid-scale precipitation data. This assumption may be valid in areas with relatively uniform precipitation patterns and flat terrain, but may not hold in areas with high spatial heterogeneity of precipitation and complex terrain (Xie et al. 2022). Remote sensing and reanalysis products provide gridded data, which differs significantly from point data (Tang et al. 2020). Additionally, biases may exist between gridded data from different products. For instance, there is an approximately 0.05° difference in grid center locations between IMERG-final and GSMaP-Gauge compared to ERA5 and ERA5-Land (Xu et al. 2022; Xie et al. 2022). Indirect hydrological modeling might be an effective approach to address the scale mismatch issue. Specifically, different datasets can be employed as forcings for hydrological models, and their performance can be evaluated against observed streamflow records (Tang et al. 2020). However, the reliability of this method still needs validation.
In terms of the temporal length of the data, due to the limitation of remote sensing data, we can only compare the climatology and trends of extreme precipitation indices for the period 2001–19 among various datasets. It is undeniable that using 20 years of data to represent precipitation trends may have certain limitations. The analysis of precipitation trends typically requires a longer time span because the climate system exhibits long-term cyclic changes, and short-term precipitation variations may be influenced by decadal climate variations, seasonal changes, and stochastic variations. Although our study, constrained by data availability, cannot simulate trends over a longer period, nonetheless the analysis of short-term trends in the paper provides an understanding of the characteristics of existing datasets. This approach validates the quality and consistency of various datasets, aids in monitoring potential errors or anomalies in the datasets, and prompts further product improvements. If it is possible to obtain longer time series data in the future, we hope to re-examine the modeling capabilities of various datasets for trends, to validate the findings of the study.
6. Conclusions
The study evaluated the applicability of 12 precipitation datasets for monitoring extreme precipitation indices in China using CN05.1 as the reference dataset. It specifically explored the ability of each dataset to capture the spatial distribution of multiyear averages, spatial distribution of trend values, and temporal changes of extreme precipitation intensity indices PRCPTOT, SDII, RX1DAY, RX5DAY, R95P, and R99P. The main conclusions are as follows:
-
Regarding the spatial distribution of means, the performance of various datasets in monitoring extreme precipitation indices in regions with complex terrain is generally unsatisfactory. Gauge-based datasets consistently exhibit good performance in index monitoring. Additionally, CMORPH stands out for its monitoring capabilities of PRCPTOT and SDII. However, for the RX1DAY, RX5DAY, and R95P indices, datasets perform well only in the Northeastern China, Northern China, the plain region of the Changjiang (Yangtze) River, and Southeastern China regions. For R99P, the performance of all datasets is inadequate.
-
In terms of the spatial distribution of trend values, gauge-based datasets show strong applicability for PRCPTOT and R95P. Some remote sensing datasets also exhibit good performance, such as GPCP for the PRCPTOT, RX1DAY, and RX5DAY indices; PERSIANN for the SDII, RX1DAY, RX5DAY, and R95P indices; and IMERG for the R99P index, demonstrating superior capabilities in characterizing extreme precipitation change trends. Among the reanalysis datasets, ERA5 stands out for its performance in SDII and R99P, while the rest of the datasets show significant errors.
-
From a temporal perspective, the multisource fusion dataset MSWEP closely resembles CN05.1’s time series for all indices, effectively capturing the temporal changes of extreme precipitation indices. Gauge-based datasets, apart from exhibiting a clear overestimation in monitoring the SDII index, generally present similar temporal patterns to CN05.1 for most indices. The temporal error in remote sensing datasets mainly comes from some datasets, particularly GSMaP, which displays abnormally high estimates in certain years. The temporal error in reanalysis datasets primarily arises from significant overestimation of indices in various years, with CFSR and MERRA-2 displaying particularly pronounced overestimation, showing an increasing trend in overestimation values over the years.
In summary, gauge-based datasets exhibit high reliability in terms of temporal variation, spatial distribution, and trend analysis. The fusion dataset MSWEP effectively captures the temporal patterns of various extreme precipitation indices. Remote sensing datasets perform well in certain extreme precipitation index monitoring cases, such as CMORPH’s performance comparable to observational datasets for PRCPTOT and SDII indices, and GPCP’s ability to capture the trend changes of PRCPTOT, RX1DAY, and RX5DAY. However, caution is needed for some datasets that exhibit exceptionally high values in interannual variation. Reanalysis datasets generally overestimate extreme precipitation indices, with CFSR and MERRA-2 exhibiting particularly severe overestimation. These research findings contribute to guiding the application and improvement of global precipitation datasets to some extent.
Acknowledgments.
This research is funded by the National Basic Research Program of China (2020YFA0608201). This research is also funded by the National Basic Research Program of China (2022YFF0801301) and National Natural Science Foundation of China (42005127).
Data availability statement.
Publicly available datasets were analyzed in this study. CN05.1 data can be downloaded from https://www.gleam.eu/. CPC data can be downloaded from https://downloads.psl.noaa.gov/Datasets/cpc_global_precip/. GPCC data can be downloaded from https://psl.noaa.gov/data/gridded/data.gpcc.html. GSMaP data can be downloaded from https://sharaku.eorc.jaxa.jp/GSMaP/. PERSIANN data can be downloaded from https://www.ncei.noaa.gov/metadata/geoportal/rest/metadata/item/gov.noaa.ncdc:C00854/html. CMORPH data can be downloaded from https://www.ncei.noaa.gov/products/climate-data-records/precipitation-cmorph. GPCP data can be downloaded from https://www.ncei.noaa.gov/products/climate-data-records/precipitation-gpcp-daily. IMERG data can be downloaded from https://gpm.nasa.gov/data. ERA5 data can be downloaded from https://www.ecmwf.int/. CFSR data can be downloaded from https://cfs.ncep.noaa.gov/. JRA-55 data can be downloaded from https://jra.kishou.go.jp/JRA-55/index_en.html. MERRA-2 data can be downloaded from https://gmao.gsfc.nasa.gov/. MSWEP data can be downloaded from http://www.gloh2o.org/mswep/. The Frequent Rainfall Observations on Grids (FROGS) dataset comprises all of these data, interpolated onto a 1° × 1° spatial grid. FROGS can be downloaded from https://frogs.ipsl.fr/.
REFERENCES
Alexander, L. V., and Coauthors, 2006: Global observed changes in daily climate extremes of temperature and precipitation. J. Geophys. Res., 111, D05109, https://doi.org/10.1029/2005JD006290.
Alexander, L. V., and Coauthors, 2019: On the use of indices to study extreme precipitation on sub-daily and daily timescales. Environ. Res. Lett., 14, 125008, https://doi.org/10.1088/1748-9326/ab51b6.
Alexander, L. V., M. Bador, R. Roca, S. Contractor, M. G. Donat, and P. L. Nguyen, 2020: Intercomparison of annual precipitation indices and extremes over global land areas from in situ, space-based and reanalysis products. Environ. Res. Lett., 15, 055002, https://doi.org/10.1088/1748-9326/ab79e2.
Ashouri, H., K.-L. Hsu, S. Sorooshian, D. K. Braithwaite, K. R. Knapp, L. D. Cecil, B. R. Nelson, and O. P. Prat, 2015: PERSIANN-CDR: Daily precipitation climate data record from multisatellite observations for hydrological and climate studies. Bull. Amer. Meteor. Soc., 96, 69–83, https://doi.org/10.1175/BAMS-D-13-00068.1.
Bador, M., L. V. Alexander, S. Contractor, and R. Roca, 2020: Diverse estimates of annual maxima daily precipitation in 22 state-of-the-art quasi-global land observation datasets. Environ. Res. Lett., 15, 035005, https://doi.org/10.1088/1748-9326/ab6a22.
Beck, H. E., E. F. Wood, M. Pan, C. K. Fisher, D. G. Miralles, A. I. J. M. van Dijk, T. R. McVicar, and R. F. Adler, 2019: MSWEP V2 global 3-hourly 0.1° precipitation: Methodology and quantitative assessment. Bull. Amer. Meteor. Soc., 100, 473–500, https://doi.org/10.1175/BAMS-D-17-0138.1.
Burton, I., and Coauthors, 2012: Managing the risks: International level and integration across scales. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation, C. B. Field et al., Eds., Cambridge University Press, 393–436.
Chen, C., Z. Li, Y. Song, Z. Duan, K. Mo, Z. Wang, and Q. Chen, 2020: Performance of multiple satellite precipitation estimates over a typical arid mountainous area of China: Spatiotemporal patterns and extremes. J. Hydrometeor., 21, 533–550, https://doi.org/10.1175/JHM-D-19-0167.1.
Chen, F., and Y. Gao, 2018: Evaluation of precipitation trends from high-resolution satellite precipitation products over mainland China. Climate Dyn., 51, 3311–3331, https://doi.org/10.1007/s00382-018-4080-z.
Chen, L., Y. Yu, and D.-Z. Sun, 2013: Cloud and water vapor feedbacks to the El Niño warming: Are they still biased in CMIP5 models? J. Climate, 26, 4947–4961, https://doi.org/10.1175/JCLI-D-12-00575.1.
Chen, S., and Coauthors, 2013: Similarity and difference of the two successive V6 and V7 TRMM Multisatellite Precipitation Analysis performance over China. J. Geophys. Res. Atmos., 118, 13 060–13 074, https://doi.org/10.1002/2013JD019964.
Ding, G., J. Guo, Z. Tang, H. Zhang, and H. Guo, 2022: Evaluation on applicability of various reanalysis datasets of precipitation in Changjiang River Basin. Renmin Changjiang, 53, 72–79, https://doi.org/10.16232/j.cnki.1001-4179.2022.09.012.
Ding, Y., J. Li, Z. Jiang, and J. Yu, 2011: Advances in extremes statistics and their application to climate change study. Climate Change Res., 7, 248–252, https://doi.org/10.3969/j.issn.1673-1719.2011.04.003.
Fischer, E. M., U. Beyerle, and R. Knutti, 2013: Robust spatially aggregated projections of climate extremes. Nat. Climate Change, 3, 1033–1038, https://doi.org/10.1038/nclimate2051.
Frich, P., L. V. Alexander, P. Della-Marta, B. Gleason, M. Haylock, A. M. G. K. Tank, and T. Peterson, 2002: Observed coherent changes in climatic extremes during the second half of the twentieth century. Climate Res., 19, 193–212, https://doi.org/10.3354/cr019193.
Fu, G., J. Yu, X. Yu, R. Ouyang, Y. Zhang, P. Wang, W. Liu, and L. Min, 2013: Temporal variation of extreme rainfall events in China, 1961–2009. J. Hydrol., 487, 48–59, https://doi.org/10.1016/j.jhydrol.2013.02.021.
Funk, C., L. Harrison, L. Alexander, P. Peterson, A. Behrangi, and G. Husak, 2019: Exploring trends in wet-season precipitation and drought indices in wet, humid and dry regions. Environ. Res. Lett., 14, 115002, https://doi.org/10.1088/1748-9326/ab4a6c.
Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 5419–5454, https://doi.org/10.1175/JCLI-D-16-0758.1.
Groisman, P. Y., V. V. Koknaeva, T. A. Belokrylova, and T. R. Karl, 1991: Overcoming biases of precipitation measurement: A history of the USSR experience. Bull. Amer. Meteor. Soc., 72, 1725–1733, https://doi.org/10.1175/1520-0477(1991)072<1725:OBOPMA>2.0.CO;2.
Gu, X., L. Ye, Q. Xin, C. Zhang, F. Zeng, S. D. Nerantzaki, and S. M. Papalexiou, 2022: Extreme precipitation in China: A review on statistical methods and applications. Adv. Water Resour., 163, 104144, https://doi.org/10.1016/j.advwatres.2022.104144.
Guan, Y., F. Zheng, X. Zhang, and B. Wang, 2017: Trends and variability of daily precipitation and extremes during 1960–2012 in the Yangtze River Basin, China. Int. J. Climatol., 37, 1282–1298, https://doi.org/10.1002/joc.4776.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hu, Q., D. Yang, Y. Wang, and H. Yang, 2013: Accuracy and spatio-temporal variation of high resolution satellite rainfall estimate over the Ganjiang River Basin. Sci. China Technol. Sci., 56, 853–865, https://doi.org/10.1007/s11431-013-5176-7.
Hua, W., L. Zhou, S. E. Nicholson, H. Chen, and M. Qin, 2019: Assessing reanalysis data for understanding rainfall climatology and variability over central equatorial Africa. Climate Dyn., 53, 651–669, https://doi.org/10.1007/s00382-018-04604-0.
Huffman, G. J., R. F. Adler, M. M. Morrissey, D. T. Bolvin, S. Curtis, R. Joyce, B. McGavock, and J. Susskind, 2001: Global precipitation at one-degree daily resolution from multisatellite observations. J. Hydrometeor., 2, 36–50, https://doi.org/10.1175/1525-7541(2001)002<0036:GPAODD>2.0.CO;2.
Huffman, G. J., E. F. Stocker, D. T. Bolvin, E. J. Nelkin, and J. Tan, 2019: GPM IMERG final precipitation L3 half hourly 0.1 degree × 0.1 degree V06. Goddard Earth Sciences Data and Information Services Center (GES DISC), accessed 10 December 2022, https://doi.org/10.5067/GPM/IMERG/3B-HH/06.
Inoue, T., and J. Matsumoto, 2004: A comparison of summer sea level pressure over East Eurasia between NCEP-NCAR reanalysis and ERA-40 for the period 1960–99. J. Meteor. Soc. Japan, 82, 951–958, https://doi.org/10.2151/jmsj.2004.951.
Jones, E., A. A. Wing, and R. Parfitt, 2021: A global perspective of tropical cyclone precipitation in reanalyses. J. Climate, 34, 8461–8480, https://doi.org/10.1175/JCLI-D-20-0892.1.
Kharin, V. V., F. W. Zwiers, X. Zhang, and M. Wehner, 2013: Changes in temperature and precipitation extremes in the CMIP5 ensemble. Climatic Change, 119, 345–357, https://doi.org/10.1007/s10584-013-0705-8.
Kobayashi, S., and Coauthors, 2015: The JRA-55 reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 5–48, https://doi.org/10.2151/jmsj.2015-001.
Kubota, T., and Coauthors, 2007: Global precipitation map using satellite-borne microwave radiometers by the GSMaP project: Production and validation. IEEE Trans. Geosci. Remote Sens., 45, 2259–2275, https://doi.org/10.1109/TGRS.2007.895337.
Lanza, L. G., and E. Vuerich, 2009: The WMO field intercomparison of rain intensity gauges. Atmos. Res., 94, 534–543, https://doi.org/10.1016/j.atmosres.2009.06.012.
Lei, X., and Coauthors, 2022: How well does the ERA5 reanalysis capture the extreme climate events over China? Part I: Extreme precipitation. Front. Environ. Sci., 10, 921658, https://doi.org/10.3389/fenvs.2022.921658.
Li, P., and Coauthors, 2023: Intensification of mesoscale convective systems in the East Asian rainband over the past two decades. Geophys. Res. Lett., 50, e2023GL103595, https://doi.org/10.1029/2023GL103595.
Li, Y., J. Zhuang, P. Bai, Y. Zeng, Y. Xing, and Z. Yang, 2022: Improving and comprehensive evaluation precipitation estimation by merging multi-satellite precipitation products in different climate regions of China. Geogr. Res., 41, 3335–3351, https://doi.org/10.11821/dlyj020220206.
Lin, R., T. Zhou, and Y. Qian, 2014: Evaluation of global monsoon precipitation changes based on five reanalysis datasets. J. Climate, 27, 1271–1289, https://doi.org/10.1175/JCLI-D-13-00215.1.
Liu, B., X. Chen, J. Chen, and X. Chen, 2017: Impacts of different threshold definition methods on analyzing temporal-spatial features of extreme precipitation in the Pearl River Basin. Stochastic Environ. Res. Risk Assess., 31, 1241–1252, https://doi.org/10.1007/s00477-016-1284-9.
Liu, J., J. Xia, D. She, L. Li, Q. Wang, and L. Zou, 2019: Evaluation of six satellite-based precipitation products and their ability for capturing characteristics of extreme precipitation events over a climate transition area in China. Remote Sens., 11, 1477, https://doi.org/10.3390/rs11121477.
Lorenz, C., and H. Kunstmann, 2012: The hydrological cycle in three state-of-the-art reanalyses: Intercomparison and performance analysis. J. Hydrometeor., 13, 1397–1420, https://doi.org/10.1175/JHM-D-11-088.1.
Ma, S., T. Zhou, A. Dai, and Z. Han, 2015: Observed changes in the distributions of daily precipitation frequency and amount over China from 1960 to 2013. J. Climate, 28, 6960–6978, https://doi.org/10.1175/JCLI-D-15-0011.1.
Ma, S., and Coauthors, 2017: Detectable anthropogenic shift toward heavy precipitation over eastern China. J. Climate, 30, 1381–1396, https://doi.org/10.1175/JCLI-D-16-0311.1.
Ma, Y., Y. Zhang, D. Yang, and S. B. Farhan, 2015: Precipitation bias variability versus various gauges under different climatic conditions over the Third Pole Environment (TPE) region. Int. J. Climatol., 35, 1201–1211, https://doi.org/10.1002/joc.4045.
Mann, H. B., 1945: Nonparametric tests against trend. Econometrica, 13, 245–259, https://doi.org/10.2307/1907187.
Mantas, V. M., Z. Liu, C. Caro, and A. J. S. C. Pereira, 2015: Validation of TRMM Multi-Satellite Precipitation Analysis (TMPA) products in the Peruvian Andes. Atmos. Res., 163, 132–145, https://doi.org/10.1016/j.atmosres.2014.11.012.
Miao, Y., R. Liu, Q. Wang, L. Jiao, Y. Wang, L. Li, and L. Cao, 2021: Study of uncertainty of satellite and reanalysis precipitation products and their impact on hydrological simulation. Environ. Sci. Pollut. Res., 28, 60 935–60 953, https://doi.org/10.1007/s11356-021-14847-w.
Nadeem, M. U., A. A. J. Ghanim, M. N. Anjum, D. Shangguan, G. Rasool, M. Irfan, U. M. Niazi, and S. Hassan, 2022: Multiscale ground validation of satellite and reanalysis precipitation products over diverse climatic and topographic conditions. Remote Sens., 14, 4680, https://doi.org/10.3390/rs14184680.
Ning, L., and Y. Qian, 2009: Interdecadal change in extreme precipitation over South China and its mechanism. Adv. Atmos. Sci., 26, 109–118, https://doi.org/10.1007/s00376-009-0109-x.
Peng, Z., Y. Li, W. Yu, Y. Xing, A. Feng, and S. Du, 2021: Research on the applicability of remote sensing precipitation products in different climatic regions of China. J. Geo-Inf. Sci., 23, 1296–1311, https://doi.org/10.12082/dqxxkx.2021.200348.
Perkins-Kirkpatrick, S. E., and S. C. Lewis, 2020: Increasing trends in regional heatwaves. Nat. Commun., 11, 3357, https://doi.org/10.1038/s41467-020-16970-7.
Prakash, S., A. K. Mitra, A. AghaKouchak, and D. S. Pai, 2015: Error characterization of TRMM Multisatellite Precipitation Analysis (TMPA-3B42) products over India for different seasons. J. Hydrol., 529, 1302–1312, https://doi.org/10.1016/j.jhydrol.2015.08.062.
Quagraine, K. A., F. Nkrumah, C. Klein, N. A. B. Klutse, and K. T. Quagraine, 2020: West African summer monsoon precipitation variability as represented by reanalysis datasets. Climate, 8, 111, https://doi.org/10.3390/cli8100111.
Ren, Y., B. Yong, D. Lu, and H. Chen, 2019: Evaluation of the Integrated Multi-satellitE Retrievals (IMERG) for Global Precipitation Measurement (GPM) mission over the Mainland China at multiple scales (in Chinese). Hupo Kexue, 31, 560–572, https://doi.org/10.18307/2019.0224.
Ren, Y., J. Zhang, J. Fu, S. Peng, and Z. Li, 2022: Spatiotemporally varied extreme precipitation events simultaneously controlled by multiple circulation factors in China’s Loess Plateau. Int. J. Climatol., 42, 6351–6372, https://doi.org/10.1002/joc.7593.
Ren, Z., M. Zhang, S. Wang, F. Qiang, X. Zhu, and L. Dong, 2015: Changes in daily extreme precipitation events in South China from 1961 to 2011. J. Geogr. Sci., 25, 58–68, https://doi.org/10.1007/s11442-015-1153-3.
Roca, R., L. V. Alexander, G. Potter, M. Bador, R. Jucá, S. Contractor, M. G. Bosilovich, and S. Cloché, 2019: FROGS: A daily 1° × 1° gridded precipitation database of rain gauge, satellite and reanalysis products. Earth Syst. Sci. Data, 11, 1017–1035, https://doi.org/10.5194/essd-11-1017-2019.
Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 1015–1058, https://doi.org/10.1175/2010BAMS3001.1.
Schamm, K., M. Ziese, A. Becker, P. Finger, A. Meyer-Christoffer, U. Schneider, M. Schröder, and P. Stender, 2014: Global gridded precipitation over land: A description of the new GPCC first guess daily product. Earth Syst. Sci. Data, 6, 49–60, https://doi.org/10.5194/essd-6-49-2014.
Sen, P. K., 1968: Estimates of the regression coefficient based on Kendall’s tau. J. Amer. Stat. Assoc., 63, 1379–1389, https://doi.org/10.1080/01621459.1968.10480934.
Seneviratne, S. I., and Coauthors, 2012: Changes in climate extremes and their impacts on the natural physical environment. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation, C. B. Field et al., Eds., Cambridge University Press, 109–230.
Seneviratne, S. I., and Coauthors, 2021: Weather and climate extreme events in a changing climate. Climate Change 2021: The Physical Science Basis, V. Masson-Delmotte et al., Eds., Cambridge University Press, 1513–1766.
Shedekar, V. S., K. W. King, N. R. Fausey, A. B. O. Soboyejo, R. D. Harmel, and L. C. Brown, 2016: Assessment of measurement errors and dynamic calibration methods for three different tipping bucket rain gauges. Atmos. Res., 178–179, 445–458, https://doi.org/10.1016/j.atmosres.2016.04.016.
Shi, J., L. Cui, K. Wen, Z. Tian, P. Wei, and B. Zhang, 2018: Trends in the consecutive days of temperature and precipitation extremes in China during 1961–2015. Environ. Res., 161, 381–391, https://doi.org/10.1016/j.envres.2017.11.037.
Sieck, L. C., S. J. Burges, and M. Steiner, 2007: Challenges in obtaining reliable measurements of point rainfall. Water Resour. Res., 43, W01420, https://doi.org/10.1029/2005WR004519.
Song, F., and T. Zhou, 2014: Interannual variability of East Asian summer monsoon simulated by CMIP3 and CMIP5 AGCMs: Skill dependence on Indian Ocean–western Pacific anticyclone teleconnection. J. Climate, 27, 1679–1697, https://doi.org/10.1175/JCLI-D-13-00248.1.
Song, H., Z. Zhu, and Y. Li, 2021: Validation of land data assimilation and reanalysis precipitation datasets over inner Mongolia. Arid Zone Res., 38, 1624–1636, https://doi.org/10.13866/j.azr.2021.06.14.
Sun, Q., C. Miao, Q. Duan, H. Ashouri, S. Sorooshian, and K.-L. Hsu, 2018: A review of global precipitation data sets: Data sources, estimation, and intercomparisons. Rev. Geophys., 56, 79–107, https://doi.org/10.1002/2017RG000574.
Sun, S., C. Shi, Y. Pan, L. Bai, B. Xu, T. Zhang, S. Han, and L. Jiang, 2020: Applicability assessment of the 1998–2018 CLDAS multi-source precipitation fusion dataset over China. J. Meteor. Res., 34, 879–892, https://doi.org/10.1007/s13351-020-9101-2.
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., M. P. Clark, S. M. Papalexiou, Z. Ma, and Y. Hong, 2020: Have satellite precipitation products improved over last two decades? A comprehensive comparison of GPM IMERG with nine satellite and reanalysis datasets. Remote Sens. Environ., 240, 111697, https://doi.org/10.1016/j.rse.2020.111697.
Trenberth, K. E., T. Koike, and K. Onogi, 2008: Progress and prospects for reanalysis for weather and climate. Eos, Trans. Amer. Geophys. Union, 89, 234–235, https://doi.org/10.1029/2008EO260002.
Wang, J., Y. Chen, S. F. B. Tett, Z. Yan, P. Zhai, J. Feng, and J. Xia, 2020: Anthropogenically-driven increases in the risks of summertime compound hot extremes. Nat. Commun., 11, 528, https://doi.org/10.1038/s41467-019-14233-8.
Wang, Q., J. Xia, D. She, X. Zhang, J. Liu, and Y. Zhang, 2021: Assessment of four latest long-term satellite-based precipitation products in capturing the extreme precipitation and streamflow across a humid region of southern China. Atmos. Res., 257, 105554, https://doi.org/10.1016/j.atmosres.2021.105554.
Wang, X., Y. Li, M. Wang, Y. Li, X. Gong, Y. Chen, Y. Chen, and W. Cao, 2021: Changes in daily extreme temperature and precipitation events in mainland China from 1960 to 2016 under global warming. Int. J. Climatol., 41, 1465–1483, https://doi.org/10.1002/joc.6865.
Wang, Z., J. Wu, S. Bai, F. Shi, and Z. Han, 2022: Spatiotemporal changes of sloping farmland resources and its soil erosion effects in Yan’an City. Res. Soil Water Conserv., 29, 1–11, https://doi.org/10.13869/j.cnki.rswc.2022.03.019.
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, J., and X.-J. Gao, 2013: A gridded daily observation dataset over China region and comparison with the other datasets. Chin. J. Geophys., 56, 1102–1111, https://doi.org/10.6038/cjg20130406.
Xia, X., Y. Tian, W. Tan, and Shao, 2022: Evaluation of satellite precipitation products in China. Water Resour. Hydropower Eng., 53, 29–40, https://doi.org/10.13928/j.cnki.wrahe.2022.08.003.
Xiao, M., Q. Zhang, and V. P. Singh, 2017: Spatiotemporal variations of extreme precipitation regimes during 1961–2010 and possible teleconnections with climate indices across China. Int. J. Climatol., 37, 468–479, https://doi.org/10.1002/joc.4719.
Xie, P., M. Chen, S. Yang, A. Yatagai, T. Hayasaka, Y. Fukushima, and C. Liu, 2007: A gauge-based analysis of daily precipitation over East Asia. J. Hydrometeor., 8, 607–626, https://doi.org/10.1175/JHM583.1.
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, 1617–1641, https://doi.org/10.1175/JHM-D-16-0168.1.
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.
Xiong, L., C. Liu, S. Chen, X. Zha, and Q. Ma, 2021: Review of post-processing research for remote-sensing precipitation products. Adv. Water Sci., 32, 627–637, https://doi.org/10.14042/j.cnki.32.1309.2021.04.014.
Xu, J., Z. Ma, S. Yan, and J. Peng, 2022: Do ERA5 and ERA5-land precipitation estimates outperform satellite-based precipitation products? A comprehensive comparison between state-of-the-art model-based and satellite-based precipitation products over mainland China. J. Hydrol., 605, 127353, https://doi.org/10.1016/j.jhydrol.2021.127353.
Yang, D., and T. Ohata, 2001: A bias-corrected Siberian regional precipitation climatology. J. Hydrometeor., 2, 122–139, https://doi.org/10.1175/1525-7541(2001)002<0122:ABCSRP>2.0.CO;2.
Ye, B., D. Yang, Y. Ding, T. Han, and T. Koike, 2004: A bias-corrected precipitation climatology for China. J. Hydrometeor., 5, 1147–1160, https://doi.org/10.1175/JHM-366.1.
Yu, E., J. Sun, H. Chen, and W. Xiang, 2015: Evaluation of a high-resolution historical simulation over China: Climatology and extremes. Climate Dyn., 45, 2013–2031, https://doi.org/10.1007/s00382-014-2452-6.
Zhai, P., X. Zhang, H. Wan, and X. Pan, 2005: Trends in total precipitation and frequency of daily precipitation extremes over China. J. Climate, 18, 1096–1108, https://doi.org/10.1175/JCLI-3318.1.
Zhang, L., P. Lan, G. Qin, C. R. Mello, E. W. Boyer, P. Luo, and L. Guo, 2021: Evaluation of three gridded precipitation products to quantify water inputs over complex mountainous terrain of western China. Remote Sens., 13, 3795, https://doi.org/10.3390/rs13193795.
Zhang, L., X. Chen, R. Lai, and Z. Zhu, 2022: Performance of satellite-based and reanalysis precipitation products under multi-temporal scales and extreme weather in mainland China. J. Hydrol., 605, 127389, https://doi.org/10.1016/j.jhydrol.2021.127389.
Zhang, T., Y. Li, J. Li, and P. Feng, 2022: Research progress on multi-source rainfall data fusion and their hydrological application. Ziran Zaihai Xuebao, 31, 15–28, https://doi.org/10.13577/j.jnd.2022.0102.
Zhang, X., and Coauthors, 2005: Trends in Middle East climate extreme indices from 1950 to 2003. J. Geophys. Res., 110, D22104, https://doi.org/10.1029/2005JD006181.
Zhao, C., J. Gong, H. Wang, S. Wei, Q. Song, and Y. Zhou, 2020: Changes of temperature and precipitation extremes in a typical arid and semiarid zone: Observations and multi-model ensemble projections. Int. J. Climatol., 40, 5128–5153, https://doi.org/10.1002/joc.6510.
Zhou, B., Y. Xu, J. Wu, S. Dong, and Y. Shi, 2016: Changes in temperature and precipitation extreme indices over China: Analysis of a high-resolution grid dataset. Int. J. Climatol., 36, 1051–1066, https://doi.org/10.1002/joc.4400.
Zhou, B.-T., and J. Qian, 2021: Changes of weather and climate extremes in the IPCC AR6. Climate Change Res., 17, 713–718, https://doi.org/10.12006/j.issn.1673-1719.2021.167.
Zhu, Q., W. Xuan, L. Liu, and Y.-P. Xu, 2016: Evaluation and hydrological application of precipitation estimates derived from PERSIANN-CDR, TRMM 3B42V7, and NCEP-CFSR over humid regions in China. Hydrol. Processes, 30, 3061–3083, https://doi.org/10.1002/hyp.10846.
