Disaggregating human population for improved land use management in Kenya

Abstract

Understanding the spatial distribution of population across a landscape is important in land use planning. In developing nations, where resources are limited, such information can facilitate more efficient decision-making for resource allocations. This article examines three methods for characterizing the distribution of human population within a natural boundary that overlaps but is substantially different from the geography of the corresponding census data. Using land cover information in conjunction with available census data, simple areal weighting, binary dasymetric mapping, and global regression methods are tested. Strengths and weaknesses of each technique are discussed. Our findings offer a new population estimate for the River Njoro watershed in Kenya and suggest that census data alone are inadequate for many resource management decisions. At the same time, however, care must be taken to account for the spatial characteristics of the environment when developing a model for improved estimates of population distribution.