The International Charter aims at providing a unified system of space data acquisition and delivery to those affected by natural or man-made disasters through Authorized Users. Each member agency has committed resources to support the provisions of the Charter and thus is helping to mitigate the effects of disasters on human life and property.
The “Charter” was activated last week by the French Civil Protection for the wind storm and floods in the South-West of France. CNES, was Project Manager for this activation and the value added products were made by SERTIT.
OTB has been used in order to process some of the images acquired. Some change detection images (using ALOS/PALSAR, ENVISAT/ASAR and TerraSAR-X) were generated using OTB’s change detection framework.
The most interesting application of OTB was a supervised classification (using the otbSupervisedClassificationApplication) of a SPOT4 image in order to produce a flood mask.
Due to bad weather conditions, the SPOT4 image was difficult to handle.
As one can see, a major part of the flooded area is covered by translucid clouds. Segmentation by thresholding is very difficult. Manual extraction of the water bodies could take too long.
The supervised SVM classification helped a lot here. Less than 10 regions of interest were manually entered (using the otbSupervisedClassificationApplication). A simple linear SVM with the default parameters was used. The learning took 30 seconds, for an overall accuracy (on the test set) of nearly 100%. The classification of the whole scene took less than a minute (3000×3000 pixels in 4 bands). The result is shown here:
Well. This is a very good result from the image processer point of view. However, this is not the kind of thing that one can give to end users dealing with the crisis management.
This binary layer was made available to the expert interpreters from SERTIT. Using their experience in image interpretation, they could improve this product (reduce the false detections due to clouds and other artifacts). But all in all, the product helped saving time.
The final, professional looking map, which was sent to the users is here:
As a conclusion, one can say that OTB helped in an operational context, although the production of good quality maps needs (and always will) human experts.