Drivers of landuse and landcover change in western Kenya, Lake Victoria Drainage Basin

In 1983 The Lake Basin Development Authority (LBDA) commissioned an Integrated Land Use Survey (ILUS) of the ≈38,000 km2 LBDA Region defined by that part of the Lake Victoria drainage basin falling within Kenya. The core data of the LBDA/ILUS consisted of some 12,000 high resolution (1:6000 image scale), sample, aerial colour photographs spaced systematically across the LBDA area. A wide range of landcover and landuse attributes were measured off these sample photographs at sub‐meter resolution, including the cover of crops and intercrops; the cover of both natural and "managed" vegetation in terms of natural forest, plantations, woodlots, hedgerows and other woody vegetation; and the extent of bush and herbaceous cover. Furthermore, these ≈12,000 point data had been integrated with a wide range of environmental and socio‐economic data sets within a formal geographical information system (GIS) environment. ICRAF has, over the last few years, developed a number of programmes in western. Kenya involved specifically with the application of agroforestry to enhance both food security and resilience to climate change among the smallholder farming sector. ICRAF has also established in western Kenya, within the Nyando and Yala River Basins, five Sentinel Sites of 10km * 10km (100km2) each in which they use a variety of remote sensing applications to monitor ecosystem health and soil conditions. ICRAF acquired modern (2010) Quickbird satellite imagery for assessing landcover and landuse, and for general environmental mapping, in each of their five Sentinel Sites in western Kenya. T his modern Quickbird satellite imagery has comparable resolution to the 1983 sample photography obtained for the LBDA/ILUS. Since LBDA had archived all the reports and photography from the 1983 ILUS along with critical georeferencing data for each of the ≈12,000 sample photographs, a rare opportunity was created to quantify landuse and landcover changes over an approximately 30 year period by comparing landuse and land cover on the 1983 sample photographs with that on the modern Quickbird satellite imagery and to relate these changes to environmental and socio‐economic drivers. The initial work carried out at ICRAF had already indicated that it was possible to reconcile the landcover and landuse classifications used on the 1983 ILUS with modern Quickbird imagery. Although more work needed to be done it was clear that a strictly comparable set of landuse and landcover attributes could be