There are many papers on collecting hydrological data, and making predictions of floods based on long term records. You might start with reading some of the experimental catchments and watersheds that the US Forest Service has studied. They have experiment stations in mountains to coastal plain, rain forests, dry or desert climates, etc. I am not sure what you might find for your area. If you don’t have a hydrology course at your college, you might contact the civil engineering department to see if they might allow special studies. Many of the techniques are presented in hydrology texts, such as Handbook of Hydrology, by Chow, was one of the texts I learned from. Hydrologists use rainfall and streamflow data, but some learn to assemble field indicators, based on channel type and morphology. A limited number of geologists, hydrologists and geomorphologists seek for historic to prehistoric indicators (paleofloods). Many countries besides USA also have scientific efforts and scientists and professionals working toward improving or applying flood estimates, prediction, design, and mitigation.

Some stream types flood frequently such as braided streams. Some infrequently or not at all, such as entrenched channels as gullies. For non-entrenched channels, the bankfull channel is full about every 1.5 years approximately, before spreading to floodplain. Soil scientists map soil types, and they can typically identify if soils are deposited by stream or river. In some cases, LiDAR mapping can improve valley surface elevation details and channel georeferencing, potentially helpful in flood estimates. Since flood estimates and design can effect life and property, it is best to have credentials or review and certification by licensed professional to help cover liability. It is certainly an interesting subject, and it is probably not unusual, even with great data at landscape level, to have flood confidence limits of plus or minus 40%.

Dear,

To work on flood forecasting, always have a history of floods (10, 50 and 100 years). It's the starting point. In developing countries, urbanization everywhere has meant that the banks of waterways have been squatted by buildings, which considerably reduces the section, for this it is necessary to have the state of occupation of the soil at different dates.

Regards

Dear Katie,

I assume you are interested in flood forecasting for the short term (hours-days). A useful review paper on (ensemble) flood forecasting is the one by Cloke and Pappenberger (Article Ensemble Flood Forecasting: A Review

In general, you need a relation between the meteorological forcing (most important precipitation) and the hydrological variable to be forecasted (probably discharge or water level). Whether you use a model or data-based approach, you can use observed precipitation data for very short term forecasting (hours-day depending on size and response time of catchment) and forecasted precipitation data (by weather models) for lead times of 1-10 days (maybe a bit more for larger catchments). Obviously, the uncertainty in the forecasts will increase with increasing lead times. It would be most logical to directly forecast water levels (instead of discharge), but most hydrological models simulate water fluxes and water balances of important stores in the hydrological system and hence have as most important output the total discharge (a flux) instead of the water level. Rating curves, which are relations between the discharge and water level, are needed to convert discharges to water levels. Data-based methods might be used to directly translate observed and forecasted precipitation to water levels, but obviously the results will be very much dependent on the calibration conditions.

As a first step it might be interesting to compare the precipitation observations obtained with sensors with gauged precipitation by looking at simple statistics (mean, variance, temporal correlation, etc.) and the time series. A next step might be to set-up a relatively simple hydrological model for a catchment and compare simulations (not in a forecast mode yet) using gauged and IoT-based precipitation as input. We have good experiences using the HBV model for simulation and forecasting, see for instance Article Effect of different uncertainty sources on the skill of 10 d...

Preprint Performance of ensemble streamflow forecasts under varied hy...

Hope this helps,

Martijn

Dear Alls,

The most important is certainly the quality of the data, the latter depends on who does the measurements and how. The flow remains important because it answers simultaneously two questions which are the quantity and the quality of the waters passing by the stream. Moreover, to minimize the risk of error and to be able to make comparisons, it is necessary to have several stations that are not distant from each other. To refer to the model only without knowing the reality of the field is a risk not to take. Good luck.

Dear Katie,

Flood is mostly occurred due to surface runoff and for a given precipitation, the evapotranspiration, initial loss, infiltration and detention storage requirements will have to be first satisfied before the commencement of runoff. So, Flood predictions require several types of data: ... Knowledge about the characteristics of a catchment area such as soil-moisture conditions, ground temperature, snowpack, topography, vegetation cover, and impermeable land area and many other characteristics along with the rainfall data. There are many empirical formulas regarding the increase in discharge and peak-flood discharge for a given catchment area's characteristics and rainfall data. (You can find many in the book named Engineering Hydrology by Subramanya. Tata McGraw-Hill Education, 2013 - Hydraulic engineering ) These empirical formulas developed are based on the statistical correlation of the observed peak and important catchment properties. So, first, you need to know about the characteristics of the catchment area and the discharge can be calculated in the river by the data obtained by the sensors you mentioned. The discharge in the river can be used to know whether it reaches the flood level or not. The concept of hydrograph and flood routing(for proper knowledge see Engineering Hydrology by Subramanya ) can be essential for the subject you are intended to pursue.

Its written mess, but hope it helps.

Saurav Bhattarai