Correction of Tropospheric Delay Effects in Radar Interferometry Using a Weather Research and Forecast Model (WRF)

Radar interferometry is a common technique for measuring surface deformations of the Earth. Tropospheric and Ionospheric delays are Tow challenges in this respect. The ionospheric effect on the interferogram phase is negligible. However, the tropospheric effect on the radar signal phase can produce a maximum error of about 10 to 14 centimeters in terms of displacement. The error generated by the troposphere is a seasonal systematic error. One of the methods for atmospheric correction in radar interferometry is based on using numerical weather forecasting models. These models provide meteorological parameters in the form of a 3D grid. In this study, the Weather Research and Forecasting (WRF) model was used to reduce the effect of tropospheric error. The WRF model is capable of processing and generating three-dimensional networks of meteorological parameters with arbitrary spatial resolution. It is also very flexible in terms of time and it can be output at precisely seconds. This advantage is very suitable for interferogram tropospheric correction, and themporal interpolation of the weather parameters, which itself is a source of error, is left out of the correction steps. But the disadvantages of this weather model is the time consuming process to produce a high spatial resolution outcome. The northwest of Iran was considered as the study area and the GP ray tracing method was used for calculating tropospheric delays. A new method which is independent of the estimated rate of deformation is proposed for analyzing the reliability of the applied corrections. In the method, the reliability of the applied corrections is checked through the analysis of RMSE errors of phase measurements. Interferograms whole displacement phase is statistically zero or is not comparable to the phase of tropospheric delay are used. To this end, the radar interferometry measurements are firstly corrected for the phase of tropospheric delays. In other words, the more the RMSE is closer to zero, the corrections have been done more successfully. The reliability of the applied corrections is then examined through the analysis of the repeatability of phase measurements wherever the rate of deformation is negligible. Obtained results show the positive performance of the WRF model (used with different spatial resolutions). According to the obtained results in the Tabriz urban area, application of the computed corrections improves the RMSE error of the tropospheric phase by 43%, 17%, 14% and 47% when the spatial resolutions of the WRF model is 1km, 3km, 9km and 27km, respectively. This reduces to 17%, 14%, 15% and 21% near the Urmia Lake. It is worth noting that the improvement of RMSE in the WRF model with spatial resolution of 27 km is more than other WRF models with Higher spatial resolutions. This is due to the spatial shift in the WRF model. Because usually the spatial shift in the WRF model with higher resolution is more than the lower resolution, it is accompanied by a greater error in tropospheric correction on phase data.