Use of Geospatial Information Systems and Remote Sensing to Assess Landcover Changes and Water Resources of Gusii Counties

Andrew Nyakundi Mocha


This report presents findings of the study on land cover
changes and river discharge trends as from 1973 to 2006 within Kisii and Nyamira counties. The area is characterized by high rainfall and fertile, deep soils that support intensive agricultural activities. Land cover in this area is predominantly related to the human activities. Vegetation cover includes crops such as tea, coffee, maize, bananas and trees mainly eucalyptus.

The main objective to the study was to fill the information gap
with respect to the land cover of area for effective management of water and other available resources for the benefit of the residents of the counties and the nation at large. Specifically the study sought to assess landcover changes and the effectiveness of medium resolution satellite images like Landsat for such assessment in a highly populated rural environment. Additionally, the study tried to analyze the surface water discharge trends of the area.

The study utilized four sets of Landsat remote sensing images
(1973, 1986, 2000 and 2006) for land cover classification and
temporal change analysis. The land cover classification came up
with five classes namely woodlot/forest, tea/woodlot, cultivated land, buildup areas and grassland/valley bottom. Grassland/valley bottom reduced and eventually disappeared by the year 2000 in favour of woodlot and cultivated land. The other land cover types changed variously over the years.

Flow data for River Gucha is available for the period 1964-87
which does not adequately cover the study period. Thus analysis of water discharge trends was only done for the period between 1973 and 1986, during which there was little change in the river flow. The accurate assessment of land cover changes was hampered by two main factors. First, images from the same season for each year of interest could not be accessed given that the study area is covered by four Landsat scenes. Second, the study area has highly mixed land cover classes not suited for interpretation by medium resolution images such as Landsat.

The study found that there were land cover changes of varying
proportions during the period 1973-2006. It has also shown that
Landsat images may not be the most suitable for land cover studies in this kind of environment. However, it was difficult to draw any meaningful conclusions on water discharge. Finally, the use of high resolution satellite imagery is recommended, without overemphasizing the necessity of adequate and up to date data.