I have 50-meter DTM and 20-meter DSM data available to choose from to use for calculating a compound topographic index (CTI). While the 20-meter resolution is more attractive, logic tells me that when calculating the compound topographic index (CTI), one should be using a bare-earth elevation model (DTM), not a surface elevation model (DSM). If flow direction and accumulation are being calculated, using a surface model would not provide a good representation of earth surface hydrology/routing; however, I see many examples where SRTM or other DSMs are used as the input elevation surface. Is this a valid approach? Your comments and any references are appreciated.
A DSM may cause some problems when determining flow direction and subsequent flow accumulation and CTI/TWI. As you mentioned a DSM is a surface model, which contains vegetation and building heights. A good discussion at this link What is the difference between DEM, DSM and DTM?