The Issue of Stochastic Constraints Used for Zwd Estimation from GPS Observations: The Case of Tropical Regions

Abstract:

A better understanding of tropical weather processes is necessary to improve numerical weather models, which are not fully satisfactory in tropical regions. Water vapor plays a key role in humid atmospheric processes and precipitable water vapor (PWV) is a widely employed quantity to study these processes and compute water budgets. PWV can be retrieved with an accuracy of about 1-2 kg.m^-2from the zenithal wet delays (ZWD) estimated during GPS data processing. In that perspective, six permanent GPS stations were deployed in West Africa within the framework of the African Monsoon Multidisciplinary Analysis (AMMA) project.

The current quality level of PWV could be reached thanks to successive improvements of the mapping functions used (NMF → GMF → VMF1) and of the a priori ZHD retrieved from meteorological models. However, standard GPS data processing does still not optimally take into account the climatic variability across stations and time: the stochastic constraints used for ZWD estimation are generally the same regardless of the station and of its local weather. This non-optimal modelling of GPS data affects the accuracy of GPS-derived PWV and is therefore one limiting factor for weather and climate studies. The purpose of this study is to quantify the influence of the stochastic constraints in the tropospheric model on ZWD and PWV estimates for tropical GPS stations. Possible improvements are suggested.