The image shown below is the result of a data fusion process using two images acquired by two different Synthetic Aperture Radars (SAR): the C-HH band SAR onboard the Canadian RADARSAT-1 satellite (13 Feb. 1996), and the C-VV band SAR onboard the European ERS-2 satellite (14 Feb. 1996). The surveyed area is the southern part of The Netherlands (64 km x 55 km), centered around the city of Amsterdam. Note that during most of the winter this area is often covered by clouds which hinder the acquisition of useful optical or near infra-red satellite imagery.
Data fusion has been realized using a processing based on the principles of control systems. This processing has been developed by PRIVATEERS N.V. Its basic principles are exposed in E. Nezry, F. Zagolski, A. Lopès and F. Yakam Simen: "Bayesian filtering of multi-channel SAR images for detection of thin details and data fusion", Proceedings of SPIE, Vol.2958, pp.130-139, Sept. 1996. Each step of this processing is designed as a control process, and is piloted by theoretical models used as a set of commands. A cascade of commands / controls processes allows the retrieval of a wide range of geophysical parameters characterizing the environmental situation, using the physical complementarity between the two sensors (here the complementarity provided by the different polarization configurations of the incident radar waves emitted by the two SAR's).
In the present representation:
These three channels are formed using control system based processes. The green channel are formed during the multi-channel filtering process which is itself a controlled process that implies data fusion aspects. The red and blue channels are produced using physical models describing the concerned geophysical parameters.
This representation enables, either by photo-interpretation (guided by radar physics knowledge), or by automatic computer-based process (E. Nezry, F. Zagolski, I. Supit and F. Yakam Simen: "Control systems principles applied to speckle filtering and to the retrieval of geophysical parameters through RADARSAT and ERS Bayesian data fusion", Proceedings of the GER'97 Symposium, Ottawa, Canada, May 1997):
Low resolution (100x100m) representation ©PRIVATEERS N.V. 1996.
Such a product can be produced at any resolution from 12.5x12.5 to 100x100 meters per pixel.
Some examples of interpretation, corresponding to the marked areas of the above image are given, to illustrate the variety of information regarding:
Note that the two acquisitions have been done in two days. This methodology has been developed in order to provide an answer to an issue which concerns, sometimes dramatically (natural hazards, etc.), the decision-makers at the regional or local scale. This issue is the very rapid (in real- or near-real-time) evaluation of a global situation, with regard to a very wide variety of aspects. The most important of these aspects are:
With its research progresses, PRIVATEERS N.V. is now ready to provide responsible decision-makers with a service or/and with the tools which would enable them to evaluate with the due rapidity the impact of critical situations such as floodings, natural hazards affecting agriculture, etc.
These results have been obtained in the framework of PRIVATEERS N.V. Pilot Projects:
PRIVATEERS Remote Sensing Page