slide 69

The variation of water vapor, temperature, and pressure in the troposphere can cause errors in INSAR measurements. Different correction methods are used, such as using tropospheric models like ERA-5 and GACOs. These models can capture the dynamic conditions of the troposphere. Linear and power law height phase estimation methods are used to model the relation between INSAR measurements and topography. The tropospheric error can also be extracted using a spatial temporal filter that separates high and low frequency variations in time and space. The variation of water vapor, temperature, and pressure in the troposphere can delay the SAR signal when it propagates the troposphere. It is usually the biggest error in the INSAR measurements because tropospheric conditions vary rapidly in time and space. Due to its varying properties, now no approach can perfectly reduce it. There are several correction methods. Using tropospheric models to correct an error is popular. ERA-5 and GACOs are the frequently used models in the INSAR literature. For example, for ERA-5, the spatial resolution is 31 km and the temporal resolution is 1 hour, which can capture the dynamic of the tropospheric condition. In terms of the linear height phase estimation, the assumption is that the tropospheric phase delay has a linear relation with the topography. The assumption makes sense because the water vapor, temperature, and pressure generally vary with height. So this method builds a Gauss-Marx model to model the relation between INSAR measurements and the topography. The phase having the linear relation with height is considered the tropospheric error. For the power law height phase estimation, the principle is similar to the linear height phase estimation. Instead of assuming the linear relation between INSAR measurements and height, this method uses the power law to approximate the tropospheric variation with height because the tropospheric variation with height is more similar to power law and relation. For the spatial temporal filter, the tropospheric conditions are assumed to be very rapidly in time and smoothly in space. So we can say the tropospheric conditions are high frequency in time and low frequency in space. As a result, the tropospheric error can be extracted by the high pass filter in time followed by the low pass filter in space.