Evaluation of MSWX Bias-Corrected Meteorological Forcing Datasets in China

Near-surface meteorological forcing (NSMF) datasets, mixed observations, and model forecasts are widely used in global climate change and sustainable development studies. For practical purposes, it is important to evaluate NSMF datasets, especially those released latest, and determine their strengths and limitations. In this study, we evaluate the performance of Multi-Source Weather (MSWX) in China over the period of 1979-2016. For comparison, ECMWF Reanalysis version 5 (ERA5), China Meteorological Forcing Dataset (CMFD) and Princeton Global Forcing (PGF) dataset are also evaluated to determine the strengths and weaknesses of MSWX. The following variables are compared with observations over 2400 stations: 2 m air temperature (T2m), 2 m daily maximum air temperature (Tmax), 2 m daily minimum air temperature (Tmin), precipitation (P), and 10 m wind speed (V10). The evaluation is conducted in terms of climatology, inter-annual variations and seasonal cycles. Results show that MSWX reasonably reproduces the spatial pattern of T2m with root-mean-square errors (RMSEs) below 1.12 & DEG;C and spatial correlations above 0.97, but underestimates Tmax and overestimates Tmin, with biases ranging from -2.0 & DEG;C to 2.0 & DEG;C, especially over the North China and Northeast China. Compared with ERA5 and PGF, MSWX can better simulate the inter-annual variations of surface air temperature with high spatial correlations (>0.97) but shows higher RMSEs than PGF. For precipitation, MSWX accurately captures the primary features of precipitation, including significant characteristics or patterns of the precipitation climatology and inter-annual variation. Its inter-annual variation shows low RMSEs ranging from 0.55 mm/day to 0.8 mm/day, compared to ERA5 and PGF. However, regions with abundant precipitation exhibit higher biases. Because the biased Global Wind Atlas (GWA3.1) is used for the wind bias correction of MSWX, MSWX significantly overestimates the annual mean wind speed, but it is consistently well-correlated with observations, with RMSEs less than 1.5 m/s and spatial correlations greater than 0.6 over the period of 1979-2016. This study reveals both the advantages and disadvantages of MSWX, and indicates the need for research into climate change and sustainable development in East Asia.

Automated summary

This study evaluates the performance of the Multi-Source Weather (MSWX) dataset in China and compares it with other datasets. The results show that MSWX reasonably reproduces temperature and precipitation, but has biases in wind speed estimation. The study reveals both the advantages and disadvantages of MSWX and emphasizes the importance of research into climate change and sustainable development in East Asia.