Interesting Project.

Please see http://gfcc.bih.nic.in/Docs/Guidelines-on-Flood-Mgmt.pdf

I have never worked in your area of interest. A quick review of the satellite coverage suggests that much of the Ganga is braided, which is associated with high sediment loading. Braided streams are unstable stream types (Rosgen classification), due to sediment loading and the aggradation of channels, loss of channel capacity, frequent flooding, movement and expansion of channel network onto the floodplain. The USGS has developed equipment through time to sample sediment loading, and would typically select bridge(s) to collect samples across the cross section. David Rosgen has done various studies with sediment and banks which may interest you (www.wildlandhydrology.com). If in fact the maximum sediment loading is at bankfull stage, collecting data at this stage would probably yield the maximum value, and his approach normalizes sediment loading and stream discharge to the bankfull stage, which can help estimate sediment loading at lesser stages.

Aerial reconnaissance with LiDAR at low flow might help estimate bankfull heights if that is desired.

https://www.bae.ncsu.edu/workshops-conferences/wp-content/uploads/sites/3/2018/08/DeMaury-Collection-of-BEHI-and-NBS-Data.pdf

Article Validation of BANCS model for assessing stream bank erosion ...

Here are a couple links about the Rosgen BANCS Model, second one was an example used in India.

Article Magnitude-Frequency Analysis for Suspended Sediment Transpor...

Sorry, Our dog decided to get sick. Anyway, Rosgen approach suggests using the Rivermorph software to help collect and analyze data if using the BANCS Model, and it is a place to put cross section data, information on bank changes, data on erosion associated with bank pins, integrate flow data if available, BEHI data, shear stress, etc. Collecting field data is a strong point of the Rosgen approach, but with LiDAR and perhaps some measurements from boat, one might be able to collect enough basic information.