I think it depends on budgets, available data, and manpower. For preliminary decision-making, I will always go for data-driven models. Reasonable accuracy but low effort.

Depending on the availability of resources and skills, one can either go for empirical or machine/ deep learning models OR process-based (physics-based) methods.

Data is preferred to me. But in developing countries, data may be scarce or insufficient. David Rosgen found in some of his studies, that the bankfull indicators (such as top of point bar) approximate the mean annual flood with return period typically ranging from 1-2 years. If one takes the bankfull depth at thalweg and multiply by 2, that depth approximates the 50 year flood depth. It would be best to read his papers, and some of which is in the USA NRCS Engineering Handbook part 505 as I remember. Beyond that, soil scientists are generally good at separating alluvial and colluvial soils, and if the flood is recent, indicators as debris, sediment lines on trees or buildings, debris deposited on trees, etc. Residents may also good sources of flood history information. LiDAR coverage can be helpful due to its landform and elevation accuracy helpful for soil type separations with flooding interpretation. If one can validate hydrological model with available data, that can be helpful. Evaluating stream type (Rosgen) also helpful, as heavily entrenched streams such as gullies don’t flood, and braided or heavily meandering (sinuous) streams flood frequently. Land use, geology, climactic, Watershed size and form, and other factors can affect flooding.

Agree with WFH. However, I believe forecasting is becoming more problematic as unprecedented climate change is changing the properties of the soil, so their capacity to store water and slow down flood is affected. Also, rainfall events appear to become intense, limiting the landscape's capacity to absorb the extra water and direct it away. Changing seasonal temperatures makes flooding due to icemelt occur earlier. Another problem is that the changing climate appears to be resulting in areas that in the past did not experience rainfall events finding themselves in this new situation.