Many algorithms and data analysis techniques have been proposed up to now which make use of satellite observations within atmospheric widows mainly for clouds and surface parameters studies and for environmental monitoring applications. Quite all these algorithms are difficult to extend to different geographical, seasonal conditions, having generally poor performances and uncertain reliability especially when applied in environmental risk prevision, monitoring and/or mitigation. In this paper a new, robust (in a statistical sense), approach (RAT) to the use of NOAA/AVHRR observations is proposed and its successful application to different environmental emergencies illustrated also in expectation of next generation of satellite sounders devoted or usable for environmental monitoring purposes.