The evaluation of IMERG and ERA5-Land daily precipitation over China with considering the influence of gauge data bias

Abstract

Evaluating the accuracy of the satellite and reanalysis precipitation products is very important for understanding their uncertainties and potential applications. However, because of underestimation existing in commonly used evaluation benchmark, gauge precipitation data, it is necessary to investigate the influence of systematic errors in gauge data on the performance evaluation of satellite and reanalysis precipitation datasets. Daily satellite-based IMERG and model-based ERA5-Land, together with gauge precipitation data, were collected with the period from 2005 to 2016 over China in this study. Daily corrections for precipitation biases from wind-induced undercatch, wetting loss, and trace error were made for gauge measurements. A set of metrics, including relative bias, Kling-Gupta efficiency, frequency bias, and critical success index, were used to evaluate and intercompare the performances of IMERG and ERA5-Land against original and bias-corrected gauge data in different locations, years, seasons, climatic zones, classes of precipitation events, and precipitation phases. The results have shown that: After removing the bias in gauge data, the relative biases of IMERG and ERA5-Land both significantly decline. The noticeable changes of their accuracy occur and vary with different locations, years, seasons, climatic zones, and precipitation phases. Furthermore, the frequency biases of IMERG and ERA5-Land rise in no precipitation events and decline in light, moderate, heavy, and extreme precipitation events. The detection capability of IMERG and ERA5-Land in no and light precipitation events is also obviously affected. Therefore, this study has demonstrated the significant influence of systematic gauge precipitation errors on the assessment of IMERG and ERA5-Land and reinforces the necessity to remove negative bias in gauge data before using it as the benchmark.

The role of teleconnection in the occurrence probability of extreme precipitation over China based on extreme value theory

“…The variance distribution is consistent with the mean distribution (Figures 1B,C). Both the mean and variance of precipitation decrease from south to north in China (also see Xie et al, 2022). The large variance in southern China (Figure 1C) indicates that precipitation in these areas is highly dispersed.…”

Bias Correction for ERA5-Land Soil Moisture Product Using Variational Mode Decomposition in the Permafrost Region of Qinghai–Tibet Plateau

“…This may be because that the meteorological forcing data used by ERA5-Land, as well as the model parameters related to soil hydrothermal transport, are not accurate enough to represent the hydrothermal conditions and the soil characteristics in the QTP arid area [62], [63]. ERA5-Land tends to overestimate the precipitation and underestimate the soil temperature on the QTP [48], [63], which affects the infiltration and evaporation process, resulting in SM overestimation during the thawing season. Soil hydrological parameters in the HTESSEL model are determined by the Food and Agriculture Organization (FAO) dataset [42], which tends to overestimate the clay content on the QTP, likely resulting in higher SM estimates [6], [62].…”

“…We then averaged the hourly data of CAMP/Tibet and ERA5-Land data to the daily time step as CRS SM data. We adopted the pixel-to-point matching strategy used by many studies in data-scarce regions [48], through selecting the ERA5-Land grid nearest to the in-situ site with a distance ranging from 2.55 km to 6.35 km. To match the corresponding soil layer between ERA5-Land and multi-layer data, we first applied the linear interpolation to the multi-depth in-situ SM data to obtain the SM profile between the uppermost and deepest measurement depths, with vertical resolution of 1 cm.…”

Brief Communication: Effects of different saturation vapor pressure calculations on simulated surface-subsurface hydrothermal regimes at a permafrost field site

“…In most cases, variations of humidity (e.g., corrections, biases) receive less attention compared with air temperature, air pressure, and precipitation (snowfall and rainfall) when publishing meteorological datasets and/or collecting forcings or input data to land surface models (e.g., Li et al, 2017;Xie et al, 2022). These variations can come from observational errors and calculation (conversion) errors, such as the choices of expressions for the SVP over a surface of water or ice when humidity is reported as RH.…”

“…The value of the SVP over ice is smaller than that over water because the molecular forces bind much more tightly in an ice crystal than in the liquid water phase (Gill, 1982;Alduchov & Eskridge, 1996). However, the SVP is often calculated only over the water surface in many cold region studies (e.g., Xu et al, 2012;Valiantzas, 2013;Leppäranta, 2015;Li et al, 2017;Brauner, 2019;Nian et al, 2022;Xie et al, 2022), and the influences of differences in SVP over water and ice on the calculation of the humidity variable are not generally evaluated in these studies.…”