Crops from Space
 
In order to ensure a good and consistent supply of food it is very important that countries and regions know how much harvest a particular area will produce. PhD student Mobushir Riaz Khan at the University of Twente’s ITC Faculty has developed a method for using satellite images to determine which crops are cultivated in which areas. This method can also estimate how much harvest a given area is likely to yield, with an accuracy of 95 per cent. He will defend the dissertation based on this research on Wednesday, 23 February at the ITC Faculty.
 
From the perspective of food safety, for the allocation of agricultural subsidies and to make the best possible use of available agricultural land it is vital that policymakers can estimate which crops grow in a certain area and how much harvest they will yield. For a large area it is usually difficult, time-consuming and therefore expensive to determine such matters. With this in mind, Mobushir Riaz Khan of the Faculty of ITC at the University of Twente developed a method for using satellite images to provide accurate estimates of which crops grow where, in what quantities and how much can be harvested. The method can benefit poorer countries as well as wealthier countries, as the satellite images used are freely available.
 
Surface area
The first stage of the study focused on identifying agricultural land on special satellite images (hypertemporal SPOT-Vegetation NDVI images) and determining the types of crop being grown. The results were then linked to field data, as a check and as a way of optimizing the method. It turns out that the method provides an accurate reflection of the situation on the ground. For example it showed that a much smaller surface area was being used for the cultivation of sunflowers in the Spanish region of Andalusia than had been registered with the European Union (which subsidizes the cultivation of sunflowers).
 
Production
In the second phase of his doctoral research, Mobushir Riaz Khan developed a method of accurately estimating how much harvest an area will produce, based on the type of crop growing in a particular area. His method was based solely on the use of satellite images with information on the local temperature and a crop growth model. This method made it possible to estimate the yield of an area measuring one square kilometre to an accuracy of 95 percent.
 
Further development
A survey has revealed that the vast majority of professionals who work with agricultural maps see added value in the new method and are interested in using it. Khan and other researchers at the ITC Faculty will develop the method to further increase its accuracy and make it suitable for a wider range of crop types.
 
Mobushir Riaz Khan
Mobushir Riaz Khan carried out his doctorate research at ITC, the Faculty of Geo-Information Sciences and Earth Observation at the University of Twente. He was supervised by Professor Eric Smaling and Dr Kees de Bie.
 
Internet: www.itc.nl