When you say model, you usually mean the numerical model of remote sensing, especially the remote sensing inversion model. Both input and output variables should be remote sensing parameters with physical significance. You should first analyze the available remote sensing parameters in satellite images from GIS software, and then establish the relationship according to the measured data values, so that the model basically comes out. This numerical model can be a simple function or a complex set of equations, which can be established by analytical physical methods or statistical regression methods. For example, if you can get land temperature data from forest remote sensing images, you can use the relationship between land temperature and wind speed to build a numerical model to predict the weather.