I am new to SAR and have some questions.
I want to incorporate optical and radar to do a very dense time series of land use change since the early 1990s. I have to incorporate SAR because the study area is a tropical region with lots of cloud cover.
My goal is to come up with a value from 0-1 that represents level of vegetation using SAR, so I can use it in a time series alongside optical data (NDMI).
The land use change I am most interested are fields opened up (less than 17 hectares) within intact rainforest. I am thinking of using HV polarization because it is more sensitive to volume scattering and has high contrast between low vegetation and high vegetation, which may make it nice for field identification.
The availability of images is your first challenge. There is no SAR sensor which operates since the 90s. You can check if you find a time series of ERS, ENVISAT, and Sentinel-1 which are all c band satellites, but they differ regarding spatial resolution and polarization (ERS is VV only), even then there is a gap between 2010 amd 2015.
So I am only using optical in the 1990s. When there is too much cloud cover in the 2000s, I will try and fill it those gaps with radar.
My main question here is finding an NDMI equivalent for radar data. I am looking for some proportion that represents the amount of vegetation: 0 being bare soil and 1 being dense tree canopy. I assume this should involve a proportion of volume scattering (more volume scattering = denser tree canopy). My trouble is finding a proportion of vegetation cover which would make it comparable to NDMI.
sorry, I was not aware of that. Would have been good to put that in the first post
You are quite limited with dual-pol data, but you could test the fraction of cross-polarization to co-polarization: VH/VV (or VH_db - VV_db) to highlight those areas where cross-polarization is comparably high.
