As you have answered yourself, 1D will certainly miss the flow around the island/sandbars and may not mimic the actual conditions. But if you can consider both (i) the gradual depth changes (reduction from u/s to mid section of island and increase d/s) and (ii) assumed cross sectional changes in the model, you may get some guestimates of flow field.

Thanks Prof Balaji.

What you seem to be describing is aggrading or braided streams, where the aggrading is evidenced by mid channel bars, or in some cases the multi-thread braiding, typically from excessive sediment loading, or lower gradient sections that are unable to maintain stream power to flush sediments, necessitating channel infilling, expanded flooding frequency and channel adjustments. Each channel may need to be treated separately, and 1D may not be the best. I am not sure what modeling approach would be best, but you might try checking with Rivermorph software company to see if they can handle braided rivers, assuming you have some cross section data and other channel metrics, etc.

If channels are relatively stable in location by strong vegetation support (Anastomosed), then probably you run cross sections of each channel to determine bankfull flow, and then check at the flood flow elevations and rates when channels combine and the floodplain or valley surface floods. I remember with a Rosgen training team running sediment loading in Rivermorph, and the stream had sufficient power to move the sediment load up to the bankfull flow rate, but at the stages when water was put onto the floodplain, the velocity dropped enough that the sediment load could not be passed.

The largest stream we worked on was the Broad River and the bankfull width was about 500 feet, it was not braided, but aggraded somewhat. We installed some stream barbs using some of the ideas of David Rosgen (www.wildlandhydrology.com) and designs by David Derrick (Corps of Engineers researcher). On very wide aggrading systems, I dont know what might work. If you have huge excessive sand loading, dredging portions up to the annual loading rate might help, and sometimes clean high quality sand can be very valuable. Instream mining of channel materials can also produce upstream and downstream impacts to consider, and there is some research on this subject. The most stable channels are generally single thread (one channel) unless Anastomosed.

By running the river and floodplain cross section, you will pick up all the channel threads, elevations, bankfull indicators, etc. So at any stage, you can estimate the functioning cross section of each channel. But under this situation, each channel may not be functioning at the same capacity, so would be treated individually, then added. When all channels are filled to bankfull and beyond, the mean velocity drops due to the greatly expanded width, increased roughness, etc. Its complex.