Local modeling of weighted mean temperature in Iran and its impact on GNSS meteorology

Abstract

Weighted mean temperature (T_{m}) is used to determine water vapor content, precipitable water vapor, and integrated water vapor (IWV) in GNSS. This parameter is highly correlated with climate conditions as well as the type of the region. The case study is performed in Iran which has diverse climate. ERA5 reanalysis datasets were used at a compact grid of 0.125 × 0.125 between 2007 and the end of 2019 to model the T_{m}. The data obtained from 12 radiosonde stations along with an IGS station located in Tehran were employed in this research. Five models were examined for T_{m}. Bevis model, linear grouping model (LGM), and linear nearest grid point model (LNGPM) were considered as T_{m} linear models, and harmonic model (HM) and GPT2w model were used as nonlinear models. In LGM method the study region was divided into smaller areas with different linear model coefficients using spatial grouping method. The local model in each radiosonde station was considered as a reference. According to the results, the accuracy of linear models (Bevis and LGM model) was between 3 and 8 K (radiosonde data as reference); also 7 out of 12 stations in the LGM had higher accuracy than the Bevis model (based on RMSE). The accuracy of the two GPT2w models and the harmonic model was higher than the previous two models, and it was between 2 and 4 K. The IWV values were obtained using zenith total delay observations of IGS station located in Tehran using 5 models and were compared with the IWV values of the radiosonde station. The accuracy of the values in three linear models, Bevis, LGM, and LNGPM, was, respectively, 0.2, 0.17, and 0.14 kg m−2, and in the two nonlinear models, GPT2w and HM, was 0.13 kg m−2.