A LiDAR based DEM has substantially more detail to help with an effort like this. Areas with soil mapping to series also suggest flooding potential. Some vegetation types withstand flooding, others do not, so that can help. If you understand the basic Rosgen stream classifications when reviewing the detailed DEM, you may be able to separate out entrenched channels as G (gullies) and F channels, and steep gradient A channels have no regular access to a significant floodplain feature. At the other flooding extreme, braided systems have high sediment loading, channel aggradation, instability that increases flooding frequency. Other highly sinuous E streams often associated with low gradient wet meadow valleys have extended floodplains. Cowardin et al, classified wetland and deep water habitats from remote sensing with ground truthing. Learn to recognize the channel and land forms, vegetation shifts, etc., and these can help. In addition, floods often leave signs that aid in field checking, as well as community observations sometimes helpful. Soil scientists commonly recognize flood prone soils in their mapping efforts. Forest and vegetation coverage maps also helpful. Consider all the Information available and do some field checking. Floodplain mapping can have considerable benefits for identifying hazards, but can also affect topics such as land values, so it is important topic and should also include limits of effort, and also good to note that channel conditions can change through time and with hydrologic modifications, significant land use changes, climate changes or localized extremes that would not be represented in most rainfall records such as an atmospheric river event. We had one instance on the Sumter National Forest where a severe storm and bank failure above the Mulberry Creek culvert, temporarily flooded a local gully terrace and caused culvert filling and failure downstream. From my observations, it was very unusual to see any signs of gully terraces flooding. As the sediment of the bankfailure moved downstream, the gully channel reestablished as a gully, so flooding to the terrace was probably a one time event. If you can correlate rainfall record to stream gauging, antecedent soil moisture and flooding records, this would also help. FEMA (Flood Emergency Management Agency) also has estimated, classified and mapped most of the floodplains in the USA. Information on their approach is probably available on the internet and may be helpful, even if your access to pertinent data is far more limited.
You cannot really just use a digital elevation model and rainfall data to map the flood extent accurately.
How you map the inundation, partly depends on what type of flooding you're interested in.
For example, if you're investigating surface water (or pluvial) flooding in an urban area that you could delineate the flood extent as follows:
1. Set up a two dimensional (2D) hydraulic model of the urban area based on your DEM.
2. Use your rainfall data to produce a design rainfall hyetograph for an annual exceedance probability (i.e. return period) of interest.
3. Use the design rainfall hyetograph as an input to your 2D hydraulic model to obtain surface water flood extents.
If you're interested in river (fluvial) flooding you'll need to carry out the following steps:
1. Setup a 1D, 2D, 1D-2D hydraulic of the river. You'll need channel cross-sections of the river and the DEM to represent the floodplain.
2. Set up a rainfall - runoff model to produce a design flood flow hydrograph for the annual exceedance probability (i.e. return period) of the flood in which you're interested.
3. Use this hydrograph as an input to your hydraulic model to obtain fluvial flood extents.
You should note that the above steps are only a very brief and basic summary of what needs to be undertaken! There are many text books and papers on this subject.
If you need free software to undertake this type of modelling a good place to look is the US Army Corps of Engineers Hydrologic Engineering Center see: https://www.hec.usace.army.mil/