Help Content > Basic Module > Flooding > Flooding Classification Refinement

Purpose

This tool generates a flood probability raster map, indicating the probability for each class for every pixel within the Area of Interest (AOI). The input data should be derived from the results obtained using the Flood Classification tool.

Technical Note

Here below the requirements regarding the input images:

•The images must be coregistered since the analysis is performed pixel by pixel

•The images must be geocoded.

Notes:

•The usage of a DEM and Slope files will increase the accuracy of the result.

•To increase the accuracy of the result it is suggested using multiple pre-event images.

Input Files

Post Event File

Input file name of the coregistered and geocoded post-event image generated by the Flooding Classification tool. This file is mandatory..

Pre Event File

Input file name of the coregistered and geocoded pre-event images generated by the Flooding Classification tool. At least one image is required.

flood_class_refinement

The Classified raster file generated by Flood Classification tool. This file is mandatory.

Ratio File

The geocoded ratio raster file generated by Flood Classification tool. This file has the suffix ''_ratio''. This file is mandatory.

Optional Files

DEM Files

Digital Elevation Model file name. This should be referred to the ellipsoid. In case a list of input files is entered, the DEM must cover the whole imaged area.

Slope File

Slope file name.

Principal Parameters

Water Threshold (dB)

This is the minimum dB value that will be used to detect the presence of water, all the pixels under this value will be considered.

In case of stable water area between the pre-event image and the post event image, the area will be classified as Persistent Water Area.

This parameter is band dependent and it is automatically set from the Flooding Menu inside Preferences Common.

Slope Threshold (deg)

This is the minimum deg value that will be used to remove the presence of Stable Water or Flood, all the pixels over this value will be considered.

Water Probability Threshold

This is the percentage threshold related to the probability of water presence to consider when determining the presence of water in the AOI.

Flood Low Probability Threshold

This is the percentage threshold related to the low probability of flooded areas being present in the AOI.

Flood Mid Probability Threshold

This is the percentage threshold related to the Mid probability of flooded areas being present in the AOI.

Flood High Probability Threshold

This is the percentage threshold related to the High probability of flooded areas being present in the AOI.

Parameters - Global

It brings to the general section of the Preferences parameters. Any modified value will be used and stored for further processing sessions.

Parameters - Other Parameters

Fuzzy Coefficient Persistent Water

This is the coefficient that controls the degree of fuzziness of the Persistent Water cluster. Higher values of this coefficient make the clusters fuzzier. When the value is equal to 1, the clustering is crisp, like in k-means, with each data point belonging to a single cluster. When the coefficient is > 1, the clusters become fuzzier, allowing data points to belong to multiple clusters with varying degrees of membership.

MRF Kernel Size Persistent Water

This is the dimension of the Markov Kernel window size used to account for the dependence of the persistent water pixel probability with respect to the nearest ones.

MRF Spatial Penalty Coefficient Persistent Water

This is the penalty coefficient of the Markov Kernel size used to weight the dependence of the persistent water pixel probability with respect to the nearest ones.

Iteration Threshold Persistent Water

This is the threshold of the intensity centroid variation that defines the convergence criterion.

Iteration Max Persistent Water

This is the maximum number of iteration of the convergence cycle.

Fuzzy Coefficient Flood

This is the coefficient that controls the degree of fuzziness of the Flood cluster.Higher values of this coefficient make the clusters fuzzier. When the value is equal to 1, the clustering is crisp, like in k-means, with each data point belonging to a single cluster. When the coefficient is > 1, the clusters become fuzzier, allowing data points to belong to multiple clusters with varying degrees of membership.

MRF Kernel Size Flood

This is the dimension of the Markov Kernel window size used to account for the dependence of the Flood pixel probability with respect to the nearest ones.

MRF Spatial Penalty Coefficient Flood

This is the penalty coefficient of the Markov Kernel size used to weight the dependence of the Flood pixel probability with respect to the nearest ones.

Iteration Threshold Flood

This is the penalty coefficient of the Markov Kernel size used to weight the dependence of the Flood pixel probability with respect to the nearest ones.

Iteration Max Flood

This is the maximum number of iteration of the convergence cycle.

Output Files

_refinement_class

Classified raster file that provides the probability of a certain pixel to be included in the ‘’Flooded Area’’ class..

_refinement

Raster file providing the flooding probability value, higher is the value higher is the probability of the pixel to be flooded.