A direct method that is still in use and produces fairly good results for shallow waters is electrofishing by defining an area, isolating the sampling area with nets in order to avoid escapement or arrivals from outside and using the device for stunning fish to catch them and keep alive in a container after the end of the sampling process. Three or more operations performed with an exact protocol (operator, time, fishing strategy) are in general necessary for computing standing biomass and catchability coefficient estimates based on Leslie-DeLury approaches analysing the cummulative catches. Electrofishing is a common method used to sample fish populations to determine absolute or relative abundance.
Set nets allows estimation of relative abundance for those species considered vulnerable, but is difficult with set nets to estimate absolute abundance because catchability and area of influence of the gears are most of the times not well known.
Whenever a commercial activity occurs and a good time series of data on catches and age composition are available, it is possible to use some approach based on catches by age as some VPA SCAA or other variants that allows estimates of biomass, mortality, etc.
My research group recently published a paper comparing electrofishing and visual surveys for fish quantification in small to medium streams. It may help you design you field survey. Don't hesitate if you have any question regarding the paper ;) http://adn.biol.umontreal.ca/~numericalecology/Reprints/Macnaughton_et_al_Riv.Res.Appl_2014.pdf