Estimation of Atmospheric Water Vapour Using the Global Positioning System: a New Observing System for New Zealand

Abstract

Atmospheric 'zenith delay' parameters are routinely estimated in high precision GPS data analysis to account for phase delays induced by tropospheric refraction. It has been recently shown that the zenith delay may further be used to obtain estimates of total precipitable water (PW) at high temporal resolution, providing a potentially useful meteorological output of continuous GPS networks that exist (usually to monitor ground deformation) in many parts of the world. In this thesis, we describe the development, validation, implementation and application of a GPS PW observing system for New Zealand. The accuracy of GPS PW is validated using data from New Zealand's continuous GPS network. Hourly GPS PW estimates are compared with collocated PW estimates derived from 38 months of radiosonde observations and 6 months of moisture fields from the NCEP AVN global analysis and forecast system. It is found that our processing methods estimate PW with high accuracy (±1.2 mm rms), sufficient to add value to the water vapour analyses currently produced by international numerical weather prediction agencies. A modified processing strategy, designed to provide accurate estimates of PW in near real-time (NRT), is developed and implemented. The strategy incorporates orbit adjustment procedures to correct the error-prone predicted satellite orbit data that must be used in NRT. It is shown that the modified processing strategy yields NRT PW estimates whose accuracy is only slightly degraded compared to post-processed PW. Finally, a mesoscale atmospheric model is used to examine the potential of GPS PW estimates to improve numerical simulations of orographic rainfall. Two cases of heavy orographic precipitation are examined, using GPS measurements collected during field experiments in the Southern Alps (SALPEX'96) and Tararua ranges (TARPEX'99). PW estimates upstream of the mountains are assimilated into the mesoscale model using the analysis nudging method, and the impact upon simulated orographic precipitation is examined by comparison with rain gauge data. In the SALPEX'96 case the assimilation of PW improved model performance. The temporal variation of orographic rainfall was better simulated, and wind and pressure biases upwind of the mountain ranges were reduced. However, in the TARPEX'99 case, the rainfall simulation was poorer after the assimilation of PW. This behaviour is attributed to insufficient knowledge of the error characteristics of the background moisture field in our assimilation scheme. In both cases, the most positive results were obtained when GPS PW data were assimilated along with other meteorological observations.