Accurate long-term precipitation data (e.g., intensity and accumulation) is critical for understanding the mechanisms behind how precipitation couples with Earth’s water fluxes, energy balances, and biogeochemical cycles across space–time scales under the changing climate (Zhang et al. 2018; Grill et al. 2019). Given the limitations of ground observations over both land and ocean, the common view is that monitoring global precipitation is likely only possible from the vantage point of space. Therefore, two international constellation-based satellite missions have been launched, TRMM (for abbreviations, see the appendix) in 1997 and GPM in 2014 (Kummerow et al. 1998; Hou et al. 2014). The objective of the GPM mission is to not only provide accurate and frequent measurements of precipitation-affected radiances but also ultimately to enable the generation of the Level 4 global precipitation products through assimilation of the satellite-based observations into numerical models (Hou et al. 2014). Currently, most state-of-the-art global and regional precipitation products are obtained through ground-based measurements (gauges and radars); satellite-based retrievals from infrared, microwave, and radar observations; and atmospheric retrospective-analysis models, as well as through the merging of their products by the scientific community (Beck et al. 2017, 2019; Zhu et al. 2021; for details of the mainstream datasets, see Table 1).
List of the state-of-the-art satellite-based (S), gauge-based (G), and reanalysis (R) precipitation products.
Recently, benefiting from a decade of advances in model physics, core dynamics, and data assimilation, ECMWF released its fifth generation of global atmospheric reanalysis, ERA5 (0.25°, 1-hourly, 1950–present), including consistent “maps without gaps” of the total precipitation, which assimilates observations collected from a rate of approximately 0.75 million day−1 on average in 1979 to ∼24 million day−1 by January 2019 from, e.g., more than over 200 satellite instruments (Hersbach et al. 2018, 2020). Meanwhile, through a single simulation driven by near-surface atmospheric fields from ERA5 and with thermodynamical orographic adjustment of temperature, a downscaled land product has also been developed: ERA5-Land (0.1°, 1-hourly). The first batch (1981–present) of the ERA5-Land dataset is now available from the Climate Data Store (last update 12 July 2019), and the back extension (1950–80) has been updated for public release in November 2021 (Muñoz-Sabater 2019a,b). Several other popular global atmospheric reanalysis products are also available, including MERRA-2 reanalysis from NASA GMAO (Gelaro et al. 2017), JRA-55 from JMA (Kobayashi et al. 2015), and CFSv2 from NCEP (Saha et al. 2014). These reanalysis products inherently contain critical references for GPM to generate the Level 4 precipitation product, and explorations of the error characteristics of these reanalysis precipitation products have indicated that they are excellent values for designing novel schemes targeting the foreseeable potential drawbacks in the GPM Level 4 global precipitation product.
Under the implementations of TRMM and GPM, various global satellite-based retrieving precipitation products with reasonable resolutions have been reported, including PERSIANN (Hsu et al. 1997), PERSIANN-CCS (Hong et al. 2004), PERSIANN-CDR (Ashouri et al. 2015), CHIRPS (Funk et al. 2015), CMORPH (Joyce et al. 2004), GSMaP (Ushio et al. 2009), TMPA (Huffman et al. 2007), and IMERG (Huffman et al. 2019a). In general, the satellite-based precipitation retrievals have mainly been focused on accurately measuring the precipitation information from 2000 to present, providing limited ability to meet the requirements of long-term hydrologically and climatologically related studies given that the WMO has stipulated that a minimum of 30 years of historical data are needed (Burroughs 2003). Although great efforts have been devoted to generating long-term climate data records spanning 30 years, e.g., PERSIANN-CDR and CHIRPS, they are limited in spatiotemporal resolutions and temporal extent because accurate satellite measurements suitable for precipitation measurements did not exist before 1980. In addition, these satellite-based precipitation products have substantial regional and systematic biases as well as substantial random errors (Shen et al. 2014; Xu et al. 2019; Lu et al. 2020). Consequently, gauge-based analysis precipitation data with high quality and high resolutions still play critical roles, usually as a benchmark, for calibrating and anchoring the satellite-based precipitation retrievals, especially those for precipitation over land (Huffman et al. 2019a; Ma et al. 2020f).
Various gauge-based precipitation analysis datasets have been developed under a series of international projects, e.g., GPCP SG (2.5°, monthly; Adler et al. 2018), CRU-TS (0.5°, monthly; Harris et al. 2020), GPCC (1.0°, daily; Markus et al. 2018), CPC_Gauge (0.5°, daily; Xie et al. 2007), and a Monthly Climatic Precipitation Dataset in China (0.5°, monthly; Peng et al. 2021). The most popular regional gauge analysis dataset with the relative highest resolutions and highest-quality data for Asia, APHRODITE (0.25°, daily) (Yatagai et al. 2012), which covers 1951–2015, provides valuable information for investigating remote and mountainous areas with complex topography, e.g., in and around the Tibetan Plateau (Tan et al. 2020). To eliminate the regional and systematic biases and the random errors inherent in the satellite precipitation product IMERG, Ma et al. (2020f) calibrated the IMERG-Final dataset by considering the APHRODITE dataset as the “ground truth” at the daily scale and provided a new dataset with higher quality: AIMERG (0.1°, half-hourly, 2000–15, Asia). Unfortunately, the temporal span of AIMERG is limited to 2000–15 because the spans of APHRODITE and IMERG-Final are 1951–2015 and 2000–present, respectively.
Preliminary investigations have found that, compared with the quality of ERA-Interim in terms of the global-mean correlations against the monthly mean GPCP, the quality of ERA5 increased substantially, from 67% to 77%. These investigations also found that the ERA5 could outperform satellite precipitation estimates and other reanalysis products (e.g., the ERA-15, ERA-40, and ERA-Interim) in cold seasons and at high latitudes (Hersbach et al. 2018, 2020; Tang et al. 2020). However, the reanalysis products (including ERA5) generally performed worse than the GPM IMERG at the hourly scale and in diurnal cycles in Mainland China; they also performed poorly in western China because of the rugged topography, especially along the boundary of the TP, where the slope is steep, which means that the drastic elevation uplift and complex topography have notable negative effects on the qualities of reanalysis products (Tang et al. 2020). Meanwhile, as pointed by Hersbach et al. (2020), the reanalysis data (e.g., ERA5) generally exhibit nonzero, and often substantial, random errors and biases because of the interactions between the models and the evolving observing system. The emergence of ERA5-Land (0.1°, 1-hourly, 1950–present) provides excellent opportunities for exploring the finer details in the long-term evolution of the precipitation events (Muñoz-Sabater 2019a). However, its quality is still unclear, and whether it could be used for generating a long-term precipitation dataset over Asia to extend the limited span of the AIMERG dataset (2000–15) warrants exploration, especially for the critical Asian regions, e.g., the TP (Ma et al. 2017, 2020f).
The objectives of the present study are 1) to explore the error characteristics of ERA5-Land; 2) to propose a novel framework for improving ERA5-Land, in combination with the APHRODITE product; and 3) to provide a new long-term precipitation dataset with higher quality over Asia (0.1°, 1-hourly, 1951–2015, Asia) for Asian applications. In addition, this study aims to provide references and possible calibration schemes using gauge analysis for generating the future GPM Level 4 global precipitation products, targeting its foreseeable potential drawbacks.
Data
ERA5-Land.
The ERA5-Land dataset (0.1°, 1-hourly) is a new generation of global land reanalysis by the European Centre for Medium-Range Weather Forecasts (ECMWF). It is also a land product downscaled through a single simulation driven by near-surface atmospheric fields from ERA5 with thermodynamical orographic adjustment of temperature, of which the first batch (1981–present) is now available from the Climate Data Store (last update 12 July 2019); the back extension (1950–80) has been updated for public release in November 2021 (Muñoz-Sabater 2019a,b; Hersbach et al. 2018, 2020). The ERA5-Land dataset provides much more information at finer spatial resolution than the ERA5 dataset (0.25°). Meanwhile, to some extent, the ERA5-Land dataset could also be the first hourly dataset for describing the water fluxes and energy balances at a spatial resolution of 0.1° on global land surface for potentially >70 years. In the present study, the variable of total precipitation in ERA5-Land is used, which represents the accumulation of liquid and solid precipitation that falls to the land surface. The ERA5-Land reanalysis total precipitation dataset was obtained through the website https://doi.org/10.24381/cds.e2161bac.
IMERG-Final.
Integrated Multisatellite Retrievals for GPM (IMERG) is the Level 3 multisatellite precipitation algorithm in GPM era, which focuses on intercalibrating, merging, and morphing “all” satellite MW-based precipitation estimates, together with MW-calibrated IR-based precipitation estimates, precipitation gauge analyses, and potentially other precipitation estimates at fine spatiotemporal scales over the entire globe (Huffman et al. 2019a). IMERG is now at its version 06 stage (Huffman et al. 2019b), based on which the corresponding IMERG precipitation products have been retrospect to cover the TRMM era at the end of September 2019. The IMERG Final run product (IMERG-Final) is the microwave-infrared estimates and calibrated based on the Global Precipitation Climatology Centre (GPCC) monthly gauge analysis. Initially, IMERG-Final was only available after June 2014. Currently, it is now available back to June 2000 (0.1°, 0.5 hourly) (Ma et al. 2020f; https://pmm.nasa.gov/data-access/downloads/gpm).
APHRODITE.
As previously mentioned, APHRODITE is one of the state-of-the-art gauge-based analysis datasets with the finest spatiotemporal resolutions (0.25°, daily, 1951–2015) and highest-quality data for Asia, integrating the largest number of ground observations from Asian countries (Yatagai et al. 2012, 2019). The APHRODITE dataset was first released in 2011 (V1101, 1951–2007) and was updated with an extensive period in September 2018 (V1101EX_R1, 2007–15). Since the release of the APRHRODITE dataset, it has received widespread attention in water-cycles-related investigations (Ji et al. 2020) and has been considered the “ground truth” observations or benchmark (Duncan and Biggs 2012; Tan et al. 2020) for calibrating satellite-based precipitation retrievals, e.g., IMERG-Final (Ma et al. 2020f). The APHRODITE products are available through the website http://aphrodite.st.hirosaki-u.ac.jp/download/.
Automatic gauge data from CMA.
The hourly point-based precipitation datasets from ∼2,400 rain gauge stations in Mainland China, from 2010 to 2014, were collected from the National Meteorological Information Center (NMIC) of the China Meteorological Administration (CMA) and were adopted as ground truth to evaluate the gridded precipitation estimates at the hourly scale. The spatial distribution of the rain gauge network in Mainland China is shown in Fig. 5a. The rain gauge precipitation datasets have undergone three levels of quality control, including an extreme values’ check, an internal consistency check, and a spatial consistency check (Shen et al. 2010, 2014). The hourly rain gauge data are available from http://data.cma.cn.
Methodology
Combination procedure of the daily total volume controlled merging and disaggregation algorithm.
Our preliminary findings indicate that the overall quality of ERA5-Land approximates that of IMERG-Final, whereas it is more prone to overdetecting the precipitation events and generally substantially overestimating the precipitation volumes, especially over the tropical and subtropical regions, e.g., the Indochina Peninsula and the South Sea Islands. Although APHRODITE has been used as the benchmark for calibrating the IMERG-Final (Ma et al. 2020f), it still contains a certain number of no-data grids, which introduces challenges for generating precipitation data with space–time continuum in the long-term round. Therefore, the present study aims to organically combine the ERA5-Land dataset with high resolutions and spatiotemporal continuity and the APHRODITE dataset with high quality using the proposed daily total volume controlled merging and disaggregation algorithm (DTVCMDA). The main steps of the DTVCMDA are explained as follows:
- 1)Aggregating the ERA5-Land: The ERA5-Land (0.1°, 1-hourly) data are first accumulated to ERA5-Land (0.1°, 1 daily) within a day, and then ERA5-Land (0.1°, 1 daily) data are aggregated to ERA5-Land (0.25°, daily) by spatially averaging the values in the corresponding 3 × 3 window of ERA5-Land (0.1°, daily), where the 3 × 3 window is coordinated to the grids at 0.25° spatial resolution from the APHRODITE dataset.
- 2)Filling the no-data grids in APHRODITE by merging their neighbors and the corresponding values from ERA5-Land (i.e., generating the Filled-APHRODITE data): In the long-term period (1951–2015), APHRODITE includes a certain number of grids with no data; in some cases, there are spatially continued no-data grids and simply filling these no-data grids by averaging their valid neighbor grids is inappropriate. Therefore, two situations are considered. If all the nearest eight grids have valid values around the no-data grid, then no-data grid is filled with the value by averaging its eight neighbors. Otherwise, the no-data grid is filled by merging the ERA5-Land and APHRODITE data. The main sequences are (i) searching the nearest 50 neighbors of the no-data grid from the APHRODITE and ERA5-Land datasets at corresponding resolutions (0.25°, daily) and locations, (ii) calculating the ratios between the APHRODITE and ERA5-Land data at each grid of the 50 pairs, (iii) deciding the median (not mean) ratio value between the ERA5-Land and APHRODITE data by carefully considering the negative effects of the anomalies from the APHRODITE and/or ERA5-Land data, and (iv) filling the no-data grid with the value by multiplying the median ratio value and the value from ERA5-Land at the corresponding location of the no-data grid with the resolutions of 0.25° and daily.
- 3)Calculating the spatial variation weights based on the ERA5-Land [e.g., generating SVW (0.1°, daily)]: The daily spatial variation weight at each grid corresponding to that of ERA5-Land (0.1°, daily) is obtained by calculating the ratio between the current grid value and the average value of the grids in the corresponding 3 × 3 window from the ERA5-Land (0.1°, daily). The moving window size is decided by considering the resolutions of the APHRODITE (0.25°) and ERA5-Land (0.1°) datasets, which utilizes the advantage of the ERA5-Land with respect to high spatial resolution, in capturing the detailed spatial patterns of the precipitation (Ma et al. 2020a).
- 4)Calculating the temporal variation weights based on the ERA5-Land [i.e., generating TVW (0.1°, 1-hourly)]: The hourly temporal variation weight is obtained by calculating the ratio between the current hourly value of ERA5-Land (0.1°, 1-hourly) and the corresponding daily value of ERA5-Land (0.1°, 1 daily) at the same location, which utilizes the advantage of the ERA5-Land dataset with respect to high temporal resolution, in capturing the evolution of the precipitation events (Ma et al. 2020a). In addition, if the daily value of ERA5-Land (0.1°, daily) is zero, the values of TVW (0.1°, hourly) within the corresponding day are all set to zero.
- 5)Deriving the daily spatial disaggregation estimates on the basis of the Filled-APHRODITE and spatial variation weights [i.e., generating AERA5-Asia (0.1°, daily)]: The AERA5-Asia (0.1°, daily) dataset is constructed by multiplying the Filled-APHRODITE and the SVM (0.1°, daily), under the total volume controlled by Filled-APHRODITE. In terms of the geographical matching strategy, three situations have been considered according to the relative spatial locations and coverage relationships between each grid of the SVM and that of the Filled-APHRODITE (for details, see Fig. 1).
- 6)Deriving the hourly temporal disaggregation estimates [i.e., generating the AERA5-Asia (0.1°, hourly) dataset]: The AERA5-Asia (0.1°, hourly) dataset is constructed by multiplying the TVW (0.1°, hourly) and the corresponding daily value of AERA5-Asia (0.1°, daily).
- 7)Trimming the AERA5-Asia (0.1°, hourly) dataset: In the present study, the Filled-APHRODITE data are considered as the benchmark for deriving the AERA5-Asia dataset. Therefore, one situation should be further processed: the situation where the Filled-APHRODITE data captures a precipitation event with a daily value larger than 0 mm, but the ERA5-Land data do not. In this case, the AERA5-Asia dataset is updated with the values within the day by equally disaggregating the daily observations from the Filled-APHRODITE data at the corresponding grid into 24 periods. Meanwhile, another situation, where the ERA5-Land captures rain events but the Filled-APHRODITE does not, has already been considered in step (6). In this situation, all the values of AERA5-Asia in the corresponding days and locations are set to zero by multiplying the TVW (0.1°, hourly) and the corresponding zero value of AERA5-Asia (0.1°, daily).
The flowchart of the daily total volume controlled merging and disaggregation algorithm (DTVCMDA) used to generate the AERA5-Asia dataset over Asia, 1951–2015.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
The final AERA5-Asia (0.1°, 1-hourly, 1951–2015, Asia) data are generated after all the aforementioned procedures, which takes advantage of both the ERA5-Land data with high resolutions and spatiotemporal continuity and the APHRODITE data with high quality over the Asia. A detailed flowchart for DTVCMDA is shown in Fig. 1.
Evaluation metrics.
To comprehensively evaluate the IMERG-Final, ERA5-Land, and AERA5-Asia precipitation products against ground gauge-based observations, seven classical statistical metrics are used: correlation coefficient (CC), relative bias (Bias), root-mean-square error (RMSE), mean absolute error (MAE), probability of detection (POD), false-alarm ratio (FAR), and the critical success index (CSI). In general, the CC, RMSE, MAE, and Bias are considered as continuous verification metrics: CC ranges in [−1, 1] with larger values indicating better agreements between satellite precipitation products and ground observations; Bias ranges in (−∞, +∞) with absolute values more close to 0 demonstrating smaller biases of satellite estimates compared with ground observations; similarly, RMSE and MAE both range in [0, +∞) with smaller values indicating smaller errors of satellite estimates compared with ground observations. The POD, FAR, and CSI represent the precipitation estimates’ capabilities in accurately capturing the precipitation events and are usually regarded as diagnostic verification metrics (Ebert et al. 2007; Wilks 2011; Ma et al. 2020f). For instance, the range of POD is from 0 to 1; the higher the value, the better the probability of product to detect precipitation events. The range of FAR is also from 0 to 1; the higher the value, the larger the probability of product to wrongly identify the nonprecipitation events as the precipitation events. As a more balanced score integrating the characteristics of POD and FAR, the range of CSI is also from 0 to 1; the higher the value, the more robust of product to accurately detect precipitation events. In addition, the threshold of 0.1 mm h−1 for hourly precipitation events is used to discriminate precipitation and no-precipitation events. The formulas and perfect values for these metrics are shown in Table 2.
Formulas and perfect values for the evaluation metrics used in the present study. Notation: n is the number of samples; Sn is satellite precipitation estimate; Gn is gauge-based precipitation; σG is the standard deviations of gauge-based precipitation; σS is the standard deviation of the satellite-based precipitation estimate; H is the precipitation event detected by both gauge and satellite simultaneously; F is the precipitation event detected by the satellite but not detected by the gauge; M is contrary to F.
For calculating the aforementioned metrics, 0.1° grid boxes with at least one rain gauge are first determined from 2010 to 2014 at the hourly scale. In addition, for those grid boxes containing more than one rain gauge, the average of the rain gauge observations in the same box is considered as the ground-truth value to evaluate the corresponding gridded precipitation estimate.
Results
AERA5-Asia product.
Unlike the process used to generate the AIMERG (0.1°, half-hourly, 2000–15) (Ma et al. 2020f), in the present study, the APHRODITE data, including a certain number of no-data grids in the long-term period from 1951 to 2015, have been first updated, as further demonstrated in the Discussion section (Figs. 12, 13). An overview of the spatial patterns of the mean annual gridded precipitation products of ERA5-Land (0.1°), APHRODITE (0.25°), Filled-APHRODITE (0.25°), and AERA5-Asia (0.1°) in the period from 1 January 1981 to 31 December 2015 shows that they share similar spatial distributions but exhibit substantial differences in terms of the volumes of precipitation (Figs. 2a–d). A comparison of the results in Figs. 2b and 2c reveals that the Filled-APHRODITE is substantially improved, with much more continuous spatial precipitation distributions, especially in eastern Indonesia and the entire country of Papua New Guinea, where the volumes of APHRODITE are abruptly small because of a large number of no-data grids. As noted by Hersbach et al. (2020), although the ERA5-Land (Fig. 2a) has strong abilities to capture the general spatial precipitation trends with decreasing precipitation from southeast to northwest Asia, which are mainly affected by the East Asian monsoon, it tends to greatly overestimate precipitation overestimations in the tropical and subtropical regions (e.g., southern China, Arunachal Pradesh, the Indochina Peninsula, and the South Sea Islands) compared with the Filled-APHRODITE (Fig. 2c). Therefore, the present study uses the Filled-APHRODITE (Fig. 2c) as a benchmark for calibrating the ERA5-Land (Fig. 2a). After being calibrated against the Filled-APHRODITE, the AERA5-Asia (Fig. 2d) data are mostly similar to the Filled-APHRODITE data (Fig. 2c) in terms of both spatial patterns and magnitudes.
Spatial patterns of the Asian mean annual gridded precipitation products of (a) ERA5-Land, 0.1°, (b) APHRODITE, 0.25°, (c) Filled-APHRODITE, 0.25°, and (d) AERA5-Asia, 0.1°, during the period from 1 Jan 1981 to 31 Dec 2015 over monsoon Asia.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
A comparison of the temporal patterns of mean precipitation products of ERA5-Land, AERA5-Asia, APHRODITE, Filled-APHRODITE (1 January 1981–31 December 2015), and IMERG-Final (1 January 2000–31 December 2015) (Fig. 3) at the monthly scale led to several critical findings:
Temporal patterns of the areal mean precipitation products of ERA5-Land, AERA5-Asia, APHRODITE, Filled-APHRODITE (1 Jan 1981–31 Dec 2015), and IMERG-Final (1 Jan 2000–31 Dec 2015) in monsoon Asia.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
- 1)The overall magnitudes of the systematic biases of ERA5-Land are large in both wet and dry seasons, even larger than those of the satellite precipitation product IMERG-Final, whose the biases have been substantially reduced by APHRODITE with the output of AIMERG (Ma et al. 2020f).
- 2)The temporal patterns of APHRODITE and Filled-APHRODITE are similar overall, but they have relatively obvious differences in the dry seasons; meanwhile, the temporal patterns of APHRODITE and Filled-APHRODITE are substantially smaller than those of ERA-Land and IMERG-Final.
- 3)The magnitudes of AERA5-Asia are smaller than those of both ERA5-Land and IMERG-Final, which are approximately the same as those of the Filled-APRHODITE (0.25°). In general, the AERA5-Asia integrates high spatiotemporal resolutions of ERA5-Land and the high quality of the Filled-APHRODITE, which substantially reduces the large systematic biases of ERA5-Land. For example, the magnitudes of ERA5-Land and AERA5-Asia are ∼150 and ∼110 mm in July, respectively, which exhibits the largest monthly precipitation volumes, and ∼50 and ∼30 mm in the dry seasons with the relatively smallest precipitation volumes (e.g., November, December, and January).
A zonal mean analysis of the annual precipitation from 1 January 1981 to 31 December 2015 based on AERA5-Asia, ERA5-Land, APHRODITE, and Filled-APHRODITE is presented in Fig. 4, which reveals several interesting reasons for the differences among these precipitation products. The ERA5-Land consistently overestimates the precipitation in both longitudinal and latitudinal directions compared with the APHRODITE, especially in the tropical and coastal regions (125°–150°E and 10°S–5°N), particular in the South Sea Islands, consistent with the results in Fig. 2. The APHRODITE and Filled-APHRODITE precipitation products are approximately the same in most cases, whereas the APHRODITE precipitation product is substantially smaller than the Filled-APHRODITE precipitation product in the region (135°–150°E and 10°S–5°N; e.g., eastern Indonesia and Papua New Guinea) because of a certain number of no-data grids. This result also indicates the reasonability of the proposed method, DTVCMDA, for updating the APHRODITE with the ERA5-Land. Under the total volume control by the Filled-APHRODITE, the AERA5-Asia is most like, and in some case the same as, the Filled-APHRODTE in the longitudinal and latitudinal directions, where the spatiotemporal resolutions of the AERA5-Asia (0.1°, 1-hourly) are substantially finer than those of Filled-APHRODITE (0.25°, 1 daily).
Zonal mean analysis of the annual precipitation, from 1 Jan 1981 to 31 Dec 2015, based on the ERA5-Land, AERA5-Land, APHRODITE, and the Filled-APHRODITE at 0.25° scale along the (a) longitudinal and (b) latitudinal directions in monsoon Asia.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
Overall performance.
Figure 5 shows a comparison of the mean annual precipitation obtained from AERA5-Asia with those obtained from the ERA5-Land, the IMERG-Final, and rain gauges in Mainland China at the annual scale for the period from 1 January 2010 to 31 December 2014. ERA5-Land substantially overestimates the precipitation, with a Bias of ∼20% (Fig. 5f), especially in southeast Mainland China. Similarly, the IMERG-Final also demonstrates large bias (>10%, Fig. 5e), whereas the AERA5-Asia is almost equal to the rain gauge observations, with absolute Bias values smaller than 5%. Overall, the AERA5-Asia outperforms the ERA5-Land and the IMERG-Final, with a CC of ∼0.98, Bias of ∼3%, RMSE of ∼128 mm yr−1, and MAE of ∼85 mm yr−1 (Fig. 5g). Therefore, compared with the ERA5-Land, the AERA-Asia dataset shows substantial improvement after calibration by the Filled-APHRODITE dataset, even by simple visual inspection.
Spatial distribution of the mean annual precipitation in Mainland China from (a) rain gauges, (b) the IMERG-Final, (c) the ERA5-Land, and (d) the AERA5-Asia in the period from 1 Jan 2010 to 31 Dec 2014 and scatter density plots of mean annual precipitation in the period from 1 Jan 2010 to 31 Dec 2014 for (e) the IMERG-Final, (f) the ERA5-Land, and (g) the AERA5-Land plotted against gauge-based observations.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
The AERA-Asia dataset was further compared with the ERA5-Land and IEMRG-Final datasets at the hourly scale to capture rain events. The distributions of the three classical continuous verification metrics (CC, RMSE, and MAE) using ∼2,400 rain gauges as ground truth in the period from 1 January 2010 to 31 December 2014 are shown in Fig. 6. In terms of CC, the AERA5-Asia notably outperforms the ERA5-Land, with CC values greater than 0.3 at most gauges, especially in central Mainland China, and the IMERG-Final performs worst, with CC values smaller than 0.3. In the case of RMSE and MAE, although the spatial patterns of the three products are similar, the volumes of the AERA5-Asia are markedly smaller than those of the ERA5-Land and the IMERG-Final, which means that the IMERG-Final exhibits the largest errors, followed by the ERA5-Land. Therefore, the AERA5-Asia dataset is clearly better than the ERA5-Land and the IMERG-Final datasets in terms of the three classical continuous verification metrics.
Spatial distributions of (top) CC, (middle) RMSE, and (bottom) MAE for the (left) IMERG-Final, (center) ERA5-Land, and (right) AERA5-Asia in Mainland China at the hourly scale in the period from 1 Jan 2010 to 31 Dec 2014.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
The other three diagnostic verification metrics (POD, FAR, and CSI) were introduced to compare the abilities of the AERA-Asia, ERA5-Land, and the IEMRG-Final at the hourly scale to capture rain events, where ∼2,400 rain gauges were used as ground truth, in the period from 1 January 2010 to 31 December 2014 (Fig. 7). In terms of the POD, the ERA5-Land appears to exhibit the best performance, with POD values larger than 0.6 at most gauges, potentially because it prefers nonzero estimates (Hersbach et al. 2020). This POD is slightly better than that of the AERA5-Asia, whereas the IMERG-Final demonstrates the worst performance, with POD values smaller than 0.5. As for the FAR, the AERA5-Asia performed best, especially in central and southeast Mainland China, with FAR values smaller than 0.5; the FAR values of the ERA5-Land and the IMERG-Final were generally larger than 0.6. As a comprehensive index of POD and FAR, CSI is usually used to provide an overall description of precipitation products for accurately capturing rain events. In general, the CSI values of the three products decrease from southeast to northwest in Mainland China, which means the three products perform better in the areas that often receive intense precipitation events. The AERA5-Asia even notably outperforms the ERA5-Land, with a CSI larger than 0.35 at most gauges, and the IMERG-Final still exhibits the worst performance, with a CSI smaller than 0.3. After calibration, the AERA5-Asia shows a clear improvement in accurately capturing the rain events when compared with the ERA5-Land. In addition, the IMERG-Final performs markedly worse than the AERA5-Aisa and the ERA5-Land. Overall, the AERA5-Asia also exhibits better overall performance than the ERA5-Land and the IMERG-Final in terms of the three diagnostic verification metrics.
Spatial distributions of the (top) POD, (middle) FAR, and (bottom) CSI for the (left) IMERG-Final, (center) ERA5-Land, and (right) AERA5-Asia at the hourly scale in the period from 1 Jan 2010 to 31 Dec 2014 in Mainland China.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
Boxplots of the six metrics (i.e., CC, RMSE, MAE, POD, FAR, and CSI) for the IMERG-Final, ERA5-Land, and the AERA5-Asia at the hourly scale in the period from 1 January 2010 to 31 December 2014 in Mainland China are shown in Fig. 8. Overall, the AERA5-Asia outperforms the ERA5-Land and the IMERG-Final, especially in terms of CC and CSI. Meanwhile, the ERA5-Land also exhibits better performance than the IMERG-Final in terms of all the six metrics, demonstrating substantial improvements of the reanalysis precipitation products in comparison with the state-of-the-art satellite-based estimates. The calibration using Filled-APHRODITE on the ERA5-Land preserves the large POD values of the AERA5-Asia inherited from the characteristics of the ERA5-Land but results in substantially reduced FAR values of the AERA5-Asia compared with those of the ERA5-Land. Therefore, even though the reanalysis precipitation products show substantial improvements, much room exists for further enhancements, especially when generating long-term precipitation products for scientific research.
Boxplots of six metrics for the IMERG-Final, ERA5-Land, and the AERA5-Asia at the hourly scale in the period from 1 Jan 2010 to 31 Dec 2014 in Mainland China. The bottom and top edges of the boxes indicate the 25th and 75th percentiles, respectively. The central black line indicates the median. The dashed lines extend from the interquartile with a length of 1.5 times the box width.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
Case studies.
To further investigate the performance of the AERA5-Asia, ERA5-Land, and the IMERG-Final against rain gauge observations, we focused on the regional extreme rainfall systems and selected two typhoon events, Trami and Usagi, which occurred in southeast Mainland China; these events spanned 47 and 15 h, respectively. Detailed descriptions follow.
Typhoon Trami.
The spatial precipitation distributions of the rain gauges, IMERG-Final, ERA5-Land, and AERA5-Asia (Figs. 9a–d) show that the ERA5-Land overestimates the precipitation over large areas with precipitation volumes > 30 mm (Fig. 9c) and has the largest Bias (>10%, Fig. 9f) among the investigated datasets. After calibration using the Filled-APHRODITE data, the absolute Bias of the AERA5-Asia data were substantially reduced (∼3%, Fig. 9g) and the correlation coefficient of the AERA5-Asia was substantially improved (∼0.85, Fig. 9g), resulting in the spatial patterns of AERA5-Asia being more similar to those of ground observations and the IMERG-Final. During this extreme event, the spatial patterns captured by the IMERG-Final were apparently better than those captures by the ERA5-Land when compared against ground observations. Meanwhile, the Bias of the IMERG-Final (∼5%) was also substantially better than that of the ERA5-Land (∼12%). These results show that the reanalysis precipitation product and the satellite-based products have their respective strengths and shortcomings. The AERA5-Asia notably outperforms the ERA5-Land and the IMERG-Final, demonstrating the excellent beneficial effect of calibration on the reanalysis precipitation estimates.
The spatial patterns of precipitation measured by (a) rain gauges, (b) the IMERG-Final, (c) the ERA5-Land, and (d) the AERA5-Asia. Scatterplots of (e) the IMERG-Final, (f) the ERA5-Land, and (g) the AERA5-Asia against rain gauge observations during Typhoon Trami, which occurred from 0900 UTC 21 Aug 2013 to 0800 UTC 23 Aug 2013.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
Typhoon Usagi.
In this subdaily extreme event, the IMERG-Final and the ERA5-Land overestimated precipitation in both spatial distribution and volume. For instance, compared with the spatial precipitation distributions of ground observations, both the ERA5-Land and the IMERG-Final indicated more areas with precipitation volumes greater than 80 mm (Figs. 10a–c). The IMERG-Final exhibited the worst performance, with the largest Bias of ∼40%; the ERA5-Land also substantially overestimated precipitation (Bias of ∼20%). After calibration, the AERA5-Asia showed a notable improvement, with a Bias smaller than 10% and a CC of ∼0.8. Therefore, both of the extreme cases of typhoons demonstrated the great benefit of calibrating the reanalysis precipitation estimates (ERA5-Land).
The spatial patterns of precipitation measured by (a) rain gauge, (b) the IMERG-Final, (c) the ERA5-Land, and (d) the AERA5-Asia. Scatterplots of (e) the IMERG-Final, (f) the ERA5-Land, and (g) the AERA5-Asia plotted against rain gauge observations during Typhoon Usagi, which occurred in the typical period 0700–2200 UTC 22 Sep 2013.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
Based on the abovementioned analyses (Figs. 2–10), there are notable improvements of AERA5-Asia against ground gauge-based observations, compared with ERA5-Land and IMERG-Final, especially in systematic biases. Whether AERA5-Asia could outperform EAR5-Land and IMERG-Final in capturing the extreme values still needs additional investigations. The spatial patterns of the 95th percentile of all hourly precipitation events measured by rain gauges, IMERG-Final, ERA5-Land, and AERA5-Asia, in the period from August 2013 to September 2013, over southeast Mainland China, are shown in Fig. 11, based on the synoptic standards that precipitation is classified as “slight,” “moderate,” or “heavy,” for precipitation rates less than 0.5 mm h−1, 0.5–4 mm h−1, and greater than 4 mm h−1, respectively (Met Office 2007). In general, both the magnitudes and spatial patterns of 95th percentile of AERA5-Asia are most similar to those of rain gauges (CC ∼0.80 and Bias ∼−4.0%), compared with those of IMERG-Final (CC ∼0.75 and Bias ∼21.5%), and ERA5-Land (CC ∼0.55 and Bias ∼10.0%), which indicates the notable improvements of the calibration using Filled-APHRODITE on ERA5-Land. Meanwhile, though the IMERG-Final could capture the spatial patterns of the extreme values, it still greatly overestimates the volumes against ground observations with the Bias value ∼20%. Therefore, AERA5-Asia still has the robust abilities to capture the extreme precipitation events both in magnitudes and spatial patterns.
The spatial patterns of the 95th percentile of all hourly precipitation events measured by (a) rain gauges, (b) the IMERG-Final, (c) the ERA5-Land, and (d) the AERA5-Asia, in the period from Aug 2013 to Sep 2013, over the southeast Mainland China, covering both the spanning periods of the two typhoon events and the regions mentioned in Figs. 9 and 10.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
Discussion
Reasons to fill the APHRODITE dataset with no-data grids and its rationality.
Though gauge-based APHRODITE data can be used as a benchmark for calibrating the satellite precipitation product to greatly reduce systematic and seasonal errors over Asia (Yatagai, et al. 2012; Ma et al. 2020f), the number of no-data grids in APHRODITE should first be investigated and clarified, especially in the long-term round. As summarized in Fig. 12, the pixels with no-data grids are mainly located in northwestern inner land regions and the South Sea Islands, from 1 January 1981 to 31 December 2015, in particular, Indonesia suffers the greatest number of the pixels with no data, varying from 1,000 to 3,000 days, in APHRODITE (Fig. 12a). Temporally, the number of no-data pixels greatly increases after the year 2010, especially in the period 2014–15, with ∼21,000 total no-data pixels per day (Fig. 12b). Therefore, it is necessary to first update the APHRODITE data by filling the no-data grids with reasonable estimates.
The characteristics of (a) spatial distributions and (b) temporal variations of the number of no-data grids of APHRODITE in monsoon Asia for the period from1 Jan 1981 to 31 Dec 2015, and (c) temporal variations of the mean monthly areal precipitation estimates of IMERG-Final, Filled-APHRODITE using mean ratios [Filled-APHRODITE (mean)], and Filled-APHRODITE using median ratios [Filled-APHRODITE (median)] in the South Sea Islands (marked by red rectangle in Fig. 12a) in the period from 2009 to 2015.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
The strategy and algorithm for filling the no-data grids are two critical issues. For the strategy to fill the no-data grids, using estimations from the ERA5-Land with higher spatiotemporal resolutions and continuity and reasonable quality demonstrated in Figs. 2–11, which show physically model-based estimates, is valuable (Muñoz-Sabater 2019a; Hersbach et al. 2018, 2020). In terms of the algorithm used to estimate values over the no-data grids, the second step of the proposed DTVCMDA method provides a detailed procedure to combine the ERA5-Land and the APHRODITE at the scale of 0.25° and daily. Notably, we decided to use the median (not the mean) value of the ratios between the APHRODTE around the no-data pixels and the ERA5-Land at the corresponding locations because the mean value of the ratios is easily affected by anomalies from the APHRODITE and/or ERA5-Land. To further investigate the rationality using median value of the ratios, the temporal variations of mean monthly areal precipitation volumes of IMERG-Final, Filled-APHRODITE using Mean ratios, and Filled-APHRODITE using median ratios in the South Sea Islands (marked by a red rectangle in Fig. 12a) in the period from 2009 to 2015 are displayed in Fig. 12c, which clearly indicates the monthly average volumes of Filled-APHRODITE using mean ratios fluctuate notably in the period from 2011 to 2015 when large numbers of no-data grids exist in this region, and even abruptly increase to ∼1,200 mm month−1 in January 2013. However, the monthly average volumes of Filled-APHRODITE using median ratios demonstrate much better stability and consistence with those of IMERG-Final, compared with those of Filled-APHRODITE using mean ratios. Therefore, using the median value of the ratios to fill the APHRODITE with no-data grids is much more reliable than using the mean value of the ratios. In addition, the number of no-data pixels’ neighbors is 50 in the present study, which we decided on the basis of our experience; the appropriate number of neighbors remains an open-ended issue for further exploration.
Taking the day on 8 October 2015 as an example in filling the pixels of APHRODITE with no data in combination with the ERA5-Land (Fig. 13), spatially continued no-data pixels of APHRODITE are clearly observed, such as in northwestern monsoon Asia, the Republic of Kazakhstan, and in almost the entire country of Indonesia (Fig. 13a). In general, the ERA5-Land demonstrates the characteristic of spatiotemporal continuity and demonstrates similar spatial patterns and volumes as the APHRODITE, which explains its suitability to be used to compensate for the APHRODITE no-data grids (Figs. 13a,b). However, the Filled-APHRODITE (Fig. 13c) obtained using the mean value of the ratios could potentially be affected by the anomalies from the APHRODITE and/or the ERA5-Land, such as over northeastern Indonesia, where the estimates are even larger than those of the ERA5-Land (Fig. 13b). On the basis of the proposed method (DTVCMDA) the Filled-APHRODITE developed using the median value of the ratios (Fig. 13d) retains the high quality of the APHRODITE dataset to the extent possible, whereas in the no-data grids, it considers the relative volume of the neighbors around the no-data grids from the APHRODITE dataset (Fig. 13a) and the spatiotemporal continuity of the ERA5-Land dataset (Fig. 13b), which is much more reasonable than the Filled-APHRODITE using the mean value of the ratios (Fig. 13c).
Daily precipitation of (a) the APHRODITE, (b) ERA5-Land, (c) Filled-APHRODITE developed using the mean ratios, and (d) Filled-APHRODITE developed using the median ratios, at a spatial resolution of 0.25°, over monsoon Asia on 8 Oct 2015.
Citation: Bulletin of the American Meteorological Society 103, 4; 10.1175/BAMS-D-20-0328.1
Potential reasons for the quality of ERA5-Land approximating that of IMERG-Final.
The high quality of the ERA5 dataset is the result of not only a decade of important developments in model physics, core dynamics, and data assimilation but also numerous improvements in the characterization, intercalibration, and processing of conventional and satellite measurements, which have enabled the quality of the historical observations (e.g., coverage and accuracy). ECMWF has assimilated observations collected at a rate from ∼0.75 million day−1 on average in 1979 to ∼24 million day−1 by January 2019 from more than 200 satellite instruments and conventional measurements, to produce the ERA5 as well as the ERA5-Land (Hersbach et al. 2020). In particular, the ERA5 has also assimilated ground-radar-based NEXRAD compositing gauge measurements (i.e., the NCEP Stage IV quantitative precipitation product over the United States from 2010 to present). Collectively, ∼94.6 billion observations have been actively assimilated in generating the ERA5 in the period from 1979 to 2019 inclusive. These substantial developments have greatly contributed to improvements of the quality of the ERA5 dataset, as well as the quality of the ERA5-Land dataset, and are potential reasons for the quality of the ERA5-Land approximating, or even exceeding, that of the IMERG-Final (Figs. 6–8). Nevertheless, the comparisons of ERA5-Land and IMERG-Final at regional extreme rainfall systems demonstrate that the reanalysis precipitation product and satellite-based products have respective strengths and shortcomings. For instance, in the case of Typhoon Trami (Fig. 9), the spatial patterns captured by the IMERG-Final appear to be better than those captured by the ERA5-Land when compared with rain gauge observations; the Bias of the IMERG-Final (∼5%) is also markedly smaller than that of the ERA5-Land (∼12%).
In the present study, we developed a new long-term precipitation dataset, AERA5-Asia (0.1°, 1-hourly, 1951–2015, Asia), by taking advantage of both the ERA5-Land dataset with high spatiotemporal resolutions and continuity and the APHRODITE dataset with the high quality. The new dataset calls for wide future evaluations and assessments on its qualities and usability for Asian precipitation scientific research and societal applications.
Conclusions
Benefiting from a decade of substantial advances in model physics, core dynamics, and data assimilation and an enormous number of observations from more than 200 satellite instruments and conventional manners, ECMWF has already released its fifth generation of global atmospheric reanalysis, ERA5 (0.25°, 1-hourly, 1950–present), as well as the ERA5-Land (0.1°, 1-hourly, 1950–present). The emergence of ERA5-Land (0.1°, hourly, 1950–present) provides great opportunities for exploring the finer details in the long-term evolution of precipitation events. However, whether it can be used for generating a long-term, high-quality precipitation dataset over Asia to extend the limited span of the AIMERG (2000–15) is an important question to be explored.
On the basis of the characteristics of the high spatiotemporal resolutions and continuity of the ERA5-Land dataset with substantial systematical and seasonal biases and the fine quality of the APHRODITE dataset with a certain number of no-data grids, we have proposed a new approach (Daily Total Volume Controlled Merging and Disaggregation Algorithm, DTVCMDA) for generating a new long-term precipitation dataset, AERA5-Asia (0.1°, 1-hourly, 1951–2015, Asia), for Asian precipitation-related scientific research and societal applications.
Our main conclusions include, but are not limited to, the following:
- 1)The quality of ERA5-Land is approximately equal to, or even better than, that of IMERG-Final, although they have respective strengths and shortcomings with respect to regional extreme rainfall systems.
- 2)A >70-yr period, 1-hourly, 0.1°, Asian precipitation product named AERA5-Aisa was developed and introduced.
- 3)The AERA5-Asia provides time series of precipitation of sufficient resolutions, length, consistency, continuity, and quality for investigating the trends and changes in spatiotemporal precipitation patterns due to climate change and natural variability.
- 4)The AERA5-Asia substantially outperforms the ERA5-Land and the IMERG-Final in terms of magnitudes and occurrences in Mainland China, especially with respect to systematic biases; for instance, the Bias values of the ERA5-Land, IMERG-Final, and the AERA5-Asia are ∼20%, ∼11%, and ∼5%, respectively.
- 5)The AERA5-Asia performs notably better than the ERA5-Land and IMERG-Final against ground-based observations in regional extreme rainfall systems (e.g., two typhoon events, Trami and Usagi.
- 6)AERA5-Asia should prove to be a useful precipitation dataset for addressing various key climatological and hydrological research questions that require precipitation data with longer spans and finer resolutions (0.1°, 1-hourly).
In addition, results of this study suggest that it would provide a valuable reference in calibrating schemes using gauge analysis for generating future GPM Level 4 global precipitation products, enabling targeting of its potential foreseeable drawbacks.
Acknowledgments.
This study was financially supported by the National Natural Science Foundation of China (Grant 41901343); the Second Tibetan Plateau Scientific Expedition and Research (STEP) program (Grant 2019QZKK0105 and 2019QZKK0103); the Key R&D Program of Ministry of Science and Technology, China (Grant 2018YFC1506500); the National Natural Science Foundation of China (Grant 91437214, 41830650, and 91837208); the Open Grants of the State Key Laboratory of Severe Weather (2021LASW-B12); the Open Fund of the State Key Laboratory of Remote Sensing Science, China (Grant OFSLRSS201909); the State Key Laboratory of Resources and Environmental Information System, China; and also the High-performance Computing Platform of Peking University.
Data availability statement.
All the datasets used in this study are accessible. The AERA5-Asia (0.1°, 1-hourly, Asia) data are freely accessible at doi.org/10.5281/zenodo.6367463 (1951–66) (Ma et al. 2020b), doi.org/10.5281/zenodo.6369796 (1967–81) (Ma et al. 2020c), doi.org/10.5281/zenodo.4266081 (1982–98) (Ma et al. 2020d), and doi.org/10.5281/zenodo.4264451 (1999–2015) (Ma et al. 2020e).
Appendix: Abbreviations with definitions used in this study
|
References
Adler, R. F., and Coauthors, 2018: The Global Precipitation Climatology Project (GPCP) monthly analysis (new version 2.3) and a review of 2017 global precipitation. Atmosphere, 9, 138, https://doi.org/10.3390/atmos9040138.
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.
Beck, H. E., and Coauthors, 2017: Global-scale evaluation of 22 precipitation datasets using gauge observations and hydrological modeling. Hydrol. Earth Syst. Sci., 21, 6201–6217, https://doi.org/10.5194/hess-21-6201-2017.
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.
Burroughs, W., 2003: Climate: Into the 21st Century. Cambridge University Press, 240 pp.
Chen, M., W. Shi, P. Xie, V. B. S. Silva, V. E. Kousky, R. Wayne Higgins, and J. E. Janowiak, 2008: Assessing objective techniques for gauge-based analyses of global daily precipitation. J. Geophys. Res., 113, D04110, https://doi.org/10.1029/2007JD009132.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Duncan, J. M. A., and E. M. Biggs, 2012: Assessing the accuracy and applied use of satellite-derived precipitation estimates over Nepal. Appl. Geogr., 34, 626–638, https://doi.org/10.1016/j.apgeog.2012.04.001.
Ebert, E. E., J. E. Janowiak, and C. Kidd, 2007: Comparison of near-real-time precipitation estimates from satellite observations and numerical models. Bull. Amer. Meteor. Soc., 88, 47–64, https://doi.org/10.1175/BAMS-88-1-47.
Funk, C., A. Verdin, J. Michaelsen, P. Peterson, D. Pedreros, and G. Husak, 2015: A global satellite assisted precipitation climatology. Earth Syst. Sci. Data, 7, 275–287, https://doi.org/10.5194/essd-7-275-2015.
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.
Grill, G., and Coauthors, 2019: Mapping the world’s free-flowing rivers. Nature, 569, 215–221, https://doi.org/10.1038/s41586-019-1111-9.
Harris, I., T. J. Osborn, P. Jones, and D. Lister, 2020: Version 4 of the CRU TS monthly high-resolution gridded multivariate climate dataset. Sci. Data, 7, 109, https://doi.org/10.1038/s41597-020-0453-3.
Hersbach, H., and Coauthors, 2018: ERA5 hourly data on single levels from 1979 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS), accessed 22 January 2021, https://doi.org/10.24381/cds.adbb2d47.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hong, Y., K.-L. Hsu, S. Sorooshian, and X. Gao, 2004: Precipitation Estimation from Remotely Sensed Imagery Using an Artificial Neural Network Cloud Classification System. J. Appl. Meteor., 43, 1834–1853, https://doi.org/10.1175/JAM2173.1.
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.
Hsu, K., X. Gao, S. Sorooshian, and H. V. Gupta, 1997: Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks. J. Appl. Meteor., 36, 1176–1190, https://doi.org/10.1175/1520-0450(1997)036<1176:PEFRSI>2.0.CO;2.
Huffman, G. J., and Coauthors, 2007: The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeor., 8, 38–55, https://doi.org/10.1175/JHM560.1.
Huffman, G. J., E. F. Stocker, D. T. Bolvin, E. J. Nelkin, and J. Tan, 2019a: 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 17 January 2020, https://doi.org/10.5067/GPM/IMERG/3B-HH/06.
Huffman, G. J., and Coauthors, 2019b: NASA Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG). Algorithm Theoretical Basis Doc., version 6, 34 pp., https://gpm.nasa.gov/sites/default/files/document_files/IMERG_ATBD_V06.pdf.
Ji, X., and Coauthors, 2020: Evaluation of bias correction methods for APHRODITE data to improve hydrologic simulation in a large Himalayan basin. Atmos. Res., 242, 104964, https://doi.org/10.1016/j.atmosres.2020.104964.
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.
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.
Kummerow, C., W. Barnes, T. Kozu, J. Shiue, and J. Simpson, 1998: The Tropical Rainfall Measuring Mission (TRMM) sensor package. J. Atmos. Oceanic Technol., 15, 809–817, https://doi.org/10.1175/1520-0426(1998)015<0809:TTRMMT>2.0.CO;2.
Lu, H., L. Ding, Z. Ma, H. Li, T. Lu, M. Su, and J. Xu, 2020: Spatiotemporal assessments on the satellite‐based precipitation products from Fengyun and GPM over the Yunnan-Kweichow Plateau, China. Earth Space Sci., 7, e2019EA000857, https://doi.org/10.1029/2019EA000857.
Ma, Z., Z. Shi, Y. Zhou, J. Xu, W. Yu, and Y. Yang, 2017: A spatial data mining algorithm for downscaling TMPA 3B43 V7 data over the Qinghai-Tibet Plateau with the effect of systematic anomalies removed. Remote Sens. Environ., 200, 378–395, https://doi.org/10.1016/j.rse.2017.08.023.
Ma, Z., J. Xu, K. He., X. Han, Q. Ji, T. Wang, W. Xiong, and Y. Hong, 2020a: An updated moving window algorithm for hourly-scale satellite precipitation downscaling: a case study in the Tibetan Plateau. J. Hydrol., 581, 124378, https://doi.org/10.1016/j.jhydrol.2019.124378.
Ma, Z., J. Xu, Y. Ma, S. Zhu, K. He, S. Zhang, W. Ma, and X. Xu, 2020b: AERA5-Asia: A long-term Asian precipitation dataset (0.1°, 1 hourly, 1951–2015, Asia) anchoring the ERA5-Land under the total volume control by APHRODITE (1951–1966), Zenodo, accessed 21 March 2022, https://doi.org/10.5281/zenodo.6367463.
Ma, Z., J. Xu, Y. Ma, S. Zhu, K. He, S. Zhang, W. Ma, and X. Xu, 2020c: AERA5-Asia: A long-term Asian precipitation dataset (0.1°, 1 hourly, 1951–2015, Asia) anchoring the ERA5-Land under the total volume control by APHRODITE (1967–1981). Zenodo, accessed 21 March 2022, https://doi.org/10.5281/zenodo.6369796.
Ma, Z., J. Xu, Y. Ma, S. Zhu, K. He, S. Zhang, W. Ma, and X. Xu, 2020d: AERA5-Asia: A long-term Asian precipitation dataset (0.1°, 1 hourly, 1951–2015, Asia) anchoring the ERA5-Land under the total volume control by APHRODITE (1982–1998). Zenodo, accessed 21 March 2022, https://doi.org/10.5281/zenodo.4266081.
Ma, Z., J. Xu, Y. Ma, S. Zhu, K. He, S. Zhang, W. Ma, and X. Xu, 2020e: AERA5-Asia: A long-term Asian precipitation dataset (0.1°, 1 hourly, 1951–2015, Asia) anchoring the ERA5-Land under the total volume control by APHRODITE (1999–2015). Zenodo, accessed 21 March 2022, https://doi.org/10.5281/zenodo.4264451.
Ma, Z., J. Xu, S. Zhu, J. Yang, G. Tang, Y. Yang, Z. Shi, and Y. Hong, 2020f: AIMERG: A new Asian precipitation dataset (0.1°/half-hourly, 2000–2015) by calibrating the GPM-era IMERG at a daily scale using APHRODITE. Earth Syst. Sci. Data, 12, 1525–1544, https://doi.org/10.5194/essd-12-1525-2020.
Markus, Z., R. Armin, B. Andreas, F. Peter, M. Anja, and S. Udo, 2018: GPCC Full Data Daily Version.2018 at 1.0°: Daily land-surface precipitation from rain-gauges built on GTS-based and historic data. Deutscher Wetterdienst, accessed 21 March 2022, https://doi.org/10.5676/DWD_GPCC/FD_D_V2018_100.
Met Office, 2007: Fact sheet No. 3 – Water in the atmosphere. National Meteorological Library and Archive, 15 pp., https://web.archive.org/web/20120114162401/http://www.metoffice.gov.uk/media/pdf/4/1/No._03_-_Water_in_the_Atmosphere.pdf.
Muñoz Sabater, J., 2019a: ERA5-Land hourly data from 1981 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS), accessed 22 January 2021, https://doi.org/10.24381/cds.e2161bac.
Muñoz Sabater, J., 2019b: First ERA5-Land dataset to be released this spring. ECMWF Newsletter, No. 159, ECMWF, Reading, United Kingdom, 8–9, www.ecmwf.int/en/newsletter/159/news/first-era5-land-dataset-be-released-spring?FORM=UCIAST&.
Peng, J., L. Duan, W. Xu, and Q. Li, 2021: Development and assessment of the monthly grid precipitation datasets in China. Front. Environ. Sci., 9, 656794, https://doi.org/10.3389/fenvs.2021.656794.
Saha, S., and Coauthors, 2014: The NCEP Climate Forecast System version 2. J. Climate, 27, 2185–2208, https://doi.org/10.1175/JCLI-D-12-00823.1.
Shen, Y., M. N. Feng, H. Z. Zhang, and X. Gao, 2010: Interpolation methods of China daily precipitation data (in Chinese). Yingyong Qixiang Xuebao, 21, 279–286.
Shen, Y., P. Zhao, Y. Pan, and J. Yu, 2014: A high spatiotemporal gauge-satellite merged precipitation analysis over China. J. Geophys. Res. Atmos., 119, 3063–3075, https://doi.org/10.1002/2013JD020686.
Tan, X., Z. Ma, K. He, X. Han, Q. Ji, and Y. He, 2020: Evaluations on gridded precipitation products spanning more than half a century over the Tibetan Plateau and its surroundings. J. Hydrol., 582, 124455, https://doi.org/10.1016/j.jhydrol.2019.124455.
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.
Ushio, T., and Coauthors, 2009: A Kalman Filter Approach to the Global Satellite Mapping of Precipitation (GSMaP) from combined passive microwave and infrared radiometric data. J. Meteor. Soc. Japan, 87A, 137–151, https://doi.org/10.2151/jmsj.87A.137.
Wilks, D. S., 2011: Statistical Methods in the Atmospheric Sciences. 3rd ed. International Geophysics Series, Vol. 100, Academic Press, 704 pp.
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.
Xu, J., Z. Ma, G. Tang, Q. Ji, X. Min, W. Wan, and Z. Shi, 2019: Quantitative evaluations and error source analysis of Fengyun-2-based and GPM-based precipitation products over Mainland China in summer, 2018. Remote Sens., 11, 2992, https://doi.org/10.3390/rs11242992.
Yatagai, A., K. Kamiguchi, O. Arakawa, A. Hamada, N. Yasutomi, and A. Kitoh, 2012: APHRODITE: Constructing a long-term daily gridded precipitation dataset for Asia based on a dense network of rain gauges. Bull. Amer. Meteor. Soc., 93, 1401–1415, https://doi.org/10.1175/BAMS-D-11-00122.1.
Yatagai, A., M. Maeda, M. Masuda, N. Suetou, N. Yasutomi, and S. Khadgarai, 2019: Asian Precipitation—Highly-Resolved Observational Data Integration Towards Evaluation of Extreme Events (APHRODITE-2). IPSJ Tohoku Branch SIG Tech. Rep. 2018-9-A2-2, 5 pp., www.topic.ad.jp/ipsj-tohoku/lib/exe/fetch.php?media=report:20180221-a2-2.pdf.
Zhang, W., G. Villarini, G. A. Vecchi, and J. A. Smith, 2018: Urbanization exacerbated the rainfall and flooding caused by Hurricane Harvey in Houston. Nature, 563, 384–388, https://doi.org/10.1038/s41586-018-0676-z.
Zhu, S., and Coauthors, 2021: A Morphology-based Adaptively Spatio-Temporal Merging Algorithm (MASTMA) for optimally combining multi-source gridded precipitation products with various resolutions. IEEE Trans. Geosci. Remote Sens., 60, 1–21, https://doi.org/10.1109/TGRS.2021.3097336.
