Crops from Space: Improved earth observation capacity to map crop areas and to quantify production
Summary of PhD thesis to fulfil the requirements for the degree of Doctor on the authority of the Rector Magnificus of the University of Twente Prof. dr. H. Brinksma, to be publicly defended on Wednesday 23 February 2011 at 14.45 in Waaier room 4, University of Twente, Enschede, Netherlands, by Mobushir Riaz Khan.
Sustainable use of land critically depends on continuous assessment and monitoring of the status of the land resources. Policy makers, responsible for food security and land use planning need accurate and timely information on crop production at regional levels. This thesis presents a systems approach to map and monitor agricultural land use using a combination of remote sensing, GIS, crop statistical data and crop modeling. The methods developed in the thesis describe firstly, what is grown where; secondly how much is grown there and lastly how much is harvested.
Hypertemporal SPOT-Vegetation NDVI images were used to map areas with different land cover types. The NDVI images were used to stratify the study areas (Nizamabad, India and Andalucia, Spain) into map units. In Nizamabad, the areas of the mapping units were related to an existing land cover map compiled from high resolution images, crop calendar information, and crop statistical data. In Andalucia, the areas of NDVI classes were related with crop statistical data to prepare the crop area maps. The NDVI data explained 77% - 98% of the variability in the crop areas. The estimated crop maps of Andalucia, showed good agreement (85-90%) with the primary field data. Soil and geomorphology data were also used along with the NDVI data to understand whether the NDVI data reflected the influences of varying types of soil. Use of NDVI data rendered the additional use of soil data unnecessary because addition of soil data only improved the amount of explained variability by 1 %. The results of the mapping exercise showed that the NDVI data were suitable for mapping crop areas and location specific crop monitoring. The NDVI data could also be incorporated in the area frame sampling method for compiling more reliable statistical data on crop areas per administrative units.
In the next part, we evaluated the output of a crop growth model (Cƒ-Water), driven by remotely sensed data, that estimates actual crop yields at 1-km2 resolution. The evaluation was done by comparing the output of Cƒ-Water and the output of an operational crop growth model, CGMS (Crop Growth Monitoring System) with published crop statistical data at regional level (province). The Cƒ-Water has less data requirements as compared to CGMS which requires also the soil and the historical yield data. Comparison of the estimated actual rainfed wheat production of Andalucia, Spain (2001) by Cƒ-Water and the estimated actual rainfed wheat production by CGMS with published agricultural statistics showed for Cƒ-Water an excellent agreement (R2 = 98%; RSME = 16 tons) and for CGMS a good agreement (R2 = 67 %; RSME= 41 tons). The accuracy assessment of Cƒ-Water estimates using the primary field data showed an excellent agreement (Adj. R2 = 98%).
Lastly, a qualitative validation of the generated outputs of the thesis was done by interviewing professionals working with agricultural land use maps and data bases. Feedback from the respondents indicated that 75% were interested in using the method developed for crop area mapping and 70% in the use method for crop yield estimation. Full text
Curriculum Vitae Mobushir Riaz Khan
Mobushir Riaz Khan was born in Faisalabad, Pakistan on 27 November 1974. In 1991, he completed his higher secondary education from his home town. Having always been interested in agriculture, he chose to extend his education in this direction and went to the University of Agriculture Faisalabad, Pakistan. At the end of 1991, he completed his M.Sc. (Hons.) in Agriculture. To get field knowledge in agriculture, he joined the Agro-service industry in the same year and managed the territory of mixed cropping systems in the province of Punjab. During his stay, there, he delivered “on and off farm” trainings to farmers and the company’s field staff for pest management and crop production. Later on, in November 2002 he joined the PMAS-University of Arid Agriculture as Lecturer in the Faculty of Crop and Food Sciences. Where he taught various courses in the department of Entomology and supervised Bachelor and Masters level students. In 2004, he was honored by the Vice Chancellor of the University and was appointed as his Technical Staff Officer. He assisted the Vice Chancellor in the meetings and seminars. He conducted many international seminars and workshops. In 2005, he got the training on Decision Support System for Agro-technology Transfer (DSSAT) at the Chiang Mai University, Thailand with the collaboration of University of Georgia, USA. In 2006, he joined the department of Natural Resources in the then International Institute for Geo-Information Science and Earth Observation (ITC), Enschede together with the Wageningen University and Research Centre, the Netherlands to pursue his doctoral studies. There, he worked in the theme of Food security and environmental sustainability (FSES) as a Ph.D. research student. In 2008, he was selected as Assistant Professor in PMAS-University of Arid Agriculture. He focused his research work on the integration of RS and GIS technologies to study crop production systems. This thesis is the outcome of this research/study work.