A novel approach for assimilating GNSS tropospheric gradient information to improve numerical weather prediction

Abstract

Global Navigation Satellite System (GNSS) has emerged as a well-established atmospheric observing system, with Zenith Total Delay (ZTD) and integrated water vapor routinely assimilated by several global and regional Numerical Weather Prediction (NWP) centers. While co-derived tropospheric gradients effectively capture water vapor horizontal anisotropy, their assimilation has yet to be widely adopted. Here, we introduce a novel approach for assimilating GNSS tropospheric gradients into the Weather Research and Forecasting (WRF) model by constructing pseudo-ZTD observations from GNSS-derived ZTD and gradient data. Through two comparative experiments, we evaluate the potential influence of GNSS tropospheric gradients on WRF forecasts. The results indicate that assimilating these gradients improves humidity and wind field predictions in the lower-to-middle troposphere (850–500 hPa), with a neutral impact on surface fields. Verification against radar estimates further demonstrates enhanced precipitation forecast skills, particularly for heavy precipitation events, by better resolving the spatial distribution and intensity of precipitation systems. A diagnosis of a precipitation event suggests that the assimilation of GNSS tropospheric gradients is able to adjust the forecast mid-level moisture distribution and modulate the forecast upward motion, thereby influencing the formation of spurious precipitation.