A regional weighted mean temperature model that takes into account climate differences: taking Shaanxi, China as an example

The weighted mean temperature Tm is a key parameter of the global navigation satellite system (GNSS) inversion of precipitation. Taking the Shaanxi area in China as an example, this paper combines the reanalysis data of the European Weather Forecast Center (ECMWF) with the data of three sounding stations, and establishes a Tm regionalized regression model considering periodicity based on the principle of least squares. The data from three radiosonde stations in Shaanxi Province were used for verification. The results show that the Tm regional model established in this paper taking into account the cycle has an average improvement rate of 16.1% compared with the traditional Bevis model. In addition, in view of the differences in regions with different climate types, this paper establishes a sub-climatic zone Tm model with a piecewise linear form that changes with latitude, and solves the problem of adaptability of the regression model in different climate zones. Compared with sounding data, the Tm model that takes into account the climate difference has an external accuracy (RMS) range of 1.47~2.06 K. Compared with the Bevis model, the average accuracy improvement rate is 44.9%, and the improvement effect is significant; using ECMWF data to select 19 each grid point evaluates the accuracy of the model. The results show that the average RMS is 3.26 K and the maximum RMS is 3.67 K; the average STD is 2.69 K and the maximum STD is 3.19 K.