| Abstrakt: |
Abstract: Troposphere delay values may be applied either in positioning or meteorology. Several troposphere delay empirical models are available as functions of meteorological parameters (temperature, air pressure and relative humidity); the zenith total delay (ZTD) values are also available as NRT (near real - time) product of GNSS (Global Navigation Satellite System) processing. To provide fully operational service for real - time PPP (Precise Point Positioning) it is essential to provide real - time ZTD estimates or short - term forecasts from near real - time estimation. This paper presents statistical approach to predict short - term ZTD from long time series. Several time series models have been used, such as autoregressive model (AR) or autoregressive moving average model (ARMA). Depending on purpose of forecasts, different time series lengths and various prediction horizons have been considered (form 1 to 24 hours). Predictions were included in both global and local model. The global model term means that one statistical model is used for all stations and the local one that each station has its own statistical model. Methods of ZTD prediction have been verified by two independent validators: deterministic Global Pressure and Temperature (GPT2) model and the Numerical Weather Prediction model COAMPS (Coupled Ocean/Atmosphere Mesoscale Prediction System). The ZTDs were calculated from meteorological parameters and compared with statistical predictions. |
