The Global Monitoring for Food Security Project: Agricultural area estimation

摘要

The Global Monitoring for food Security project (GMFS) is an European Space Agency (ESA) funded project. It forms part of the Global Monitoring for Environment and Security (GMES) Services Element. It is part of the ESA contribution to GMES, a joint ESA and European Commission (EC) initiative. The project provides earth observation based services for monitoring Food Security and related environmental processes in Sub-Saharan Africa. A very important issue as defined by the international, regional and national stakeholders is the accurate estimation of the cropped area per country. High resolution (HR) earth observation images can partially provide the answer to this problem, but there are drawbacks. The first is the cost of this kind of data when trying to cover a complete country and the second is the revisiting time. On the other hand medium to low resolution images are available at no or relative low cost and their revisiting time is much better than for the HR images. Unfortunately their resolution is in most case to low to differentiate agricultural field from other land use practices. In order to overcome these problems a new technique for area estimation is developed within the GMFS project. It is based upon the integration of high resolution images (SAR and or optical) at specific sites per country and a nation wide coverage based on medium resolution images. High resolution images are used as a basis for sub pixel classification of the medium resolution EO data. The results in a set of Area Fraction Images (AFI), one for each of the N considered land use units (e.g. 2 AFI's, one for agriculture and one for none agriculture). These AFIs have the same spatial resolution as the underlying input image sets (medium resolution) and give for each pixel the area fraction occupied by the considered class (sum of this fractions per pixel equals 1 or 100%). In other words no exact location of the agricultural area is given (which is impossible at medium resolution) but from the fraction of agriculture per pixels a total area can be calculated. This technique was especially developed to cover large area as accurately as possible at reduced cost. The cost is an important issue, since one of the goals of the GMFS project, it is to transfer the technologies and the capacities to the end users so that they will be able to implement the techniques in the regional and national early warning systems.