Rapid remote recognition of habitat changes

An attempt is made to give a realistic appraisal of the possibilities and shortcomings of remote sensing techniques in the context of detecting habitat changes.

Consideration has to be given to the various spatial scales at which remote sensing data are exploited, and to the relationship between the size of the object of an investigation and the intrinsic limitations on the ground resolution (instantaneous field of view, IFOV) of the detecting system. Features larger than the IFOV are commonly studied and, in certain circumstances, features that are smaller than the IFOV can also be detected. However, the question of obtaining information about objects or features that are very small compared with the IFOV is much more difficult. Examples include sheep distribution, locust prediction and the development of toxic algal blooms at sea. The monitoring of blooms, once they have begun to appear, is now feasible. However, it is much more difficult to study the conditions leading to the development of blooms and to give advance warning of their growth. The study of this type of problem is still in its infancy.

The other question that has to be addressed concerns the timeliness of the extraction of information from remotely sensed data. With the exception of the meteorological community, the philosophy for the last decade or two has been simply to capture remotely sensed data from satellites, to archive it on magnetic tape and to hope that someone, somewhere, someday, will retrieve the data from the archive, analyse it and possibly make use of it. In many applications of remotely sensed data, such as oil pollution monitoring, agricultural yield prediction, and epidemic and plague prevention, the remotely sensed data must be analysed, interpreted and the results distributed to the end users in something approaching real time. It is pointless, other than as an academic exercise, to predict a plague or epidemic 5 years after it has happened!