EO Data
To demonstrate the approach, an implementation of the data fusion framework has been implemented at Bregava Catchment.
Bregava belongs to the Neretva river basin with estimated catchment area of 720 km2.
Land Cover
Land cover (LC) represents one of the key descriptors of landscape and ecosystems and as such also plays a critical role
in hydrological modelling, since land cover type fundamentally affects water retention and runoff processes and regime.
For this purpose, a highly accurate LC was derived based on times series of Sentinel-2 data for the whole catchment area
as one of the main inputs for deep learning-based flood forecasting and modelling approach. Fine spatial and temporal
resolution, suitability for large scale application and certainty of continuity in future along with automatized
processing secure a good base for operational use.
Elevation Model
Digital terrain models are an important source for parameterization of hydrological models. In addition, a number of
other characteristics related to floods can be derived on their basis and are also suitable for floodplain management or
delimitation of flood risk zones. Although very high resolution models must be used for accurate modeling, freely
available elevation models such as EU-DEM represents cost effective option for large area applications.
Vegetation
Computer vision and machine learning techniques are capable to adopt additional spatiotemporal data to improve and
recognize complex patterns such as runoff. Several variables influence rainfall-runoff and drainage conditions and among
Landcover or meteorological variables, the vegetation structure is of key importance and has a strong response in
spectral reflectance. Besides standard vegetation index as NDVI, the Leaf Area Index plays a critical role and can be
derived from the time series of Copernicus Sentinel 2 data at high spatial and temporal resolution.
ERA 5 Land
ERA5-Land is a reanalysis dataset providing a consistent view of the evolution of land variables over several decades at
an enhanced resolution. ERA5-Land has been produced by replaying the land component of the ECMWF ERA5 climate reanalysis
and combines model data with observations from across the world into a globally complete and consistent dataset using
the laws of physics (Copernicus). Even though not with NRT availability, ERA5-Land provides hourly estimates of
variables such as temperature or precipitation inevitable for rainfall-runoff modelling and flood forecast.