Geostatistical analysis of land use/land cover changes and population growth trends in the Komadugu-Yobe River Basin in Nigeria
Metadata[+] Show full item record
The Komadugu-Yobe River located in northeastern Nigeria is an important tributary of Lake Chad. Since the 1970s, the Komadugu-Yobe basin has experienced significant changes in population density and land use/land cover. The present study consists of the application of geostatistical methods to examine the land use/land cover and population density dynamics in the river basin area bound between 11Â° and 13Â° N latitude and between 8Â° and 14Â° E longitude, covering approximately 151,060.95 Square Kilometers. The geostatistical methods applied include kriging interpolation, spatial autocorrelation, overlapping neighborhood statistics with Pearson's correlation coefficient, indicator variogram, rose diagram, and histogram analysis. Spatiotemporal changes for the two variables were analyzed for five years of observation: 1970, 1986, 2000, 2005, and 2009. The time-series summary and kriging interpolation of yearly precipitation identified the year 1980 as a time period of significance for the present study. The calculation of net changes in land use/land cover indicate significant variation in rainfed cropland, mosaic cropland, mosaic vegetation, grassland, sparse vegetation, and vegetation regularly flooded. Spatial autocorrelation analysis showed that the distribution of these classes is clustered. A new geostatistical tool was designed to calculate the overlapping neighborhood statistics with Pearson's correlation coefficient between the land use/land cover and population density datasets. Output maps show areas of direct correlation in red, inverse correlation in blue, and no relation in pale yellow. The 10x10 neighborhood cell unit showed a clearer correlation between the variables in certain zones of the study area. The ranges calculated from variogram plots of land use/land cover and population density were used to build rose diagrams. These diagrams depict the classes most closely related to population density: mosaic cropland, rainfed cropland, sparse vegetation, mosaic vegetation, and grassland. Population density histograms show a nearly normal distribution, which produced similar range and direction values for all the years of observation. This indicates that when variable changes are anisotropic and linear in time the ranges of the variogram functions are similar. The variogram function is a useful spatial comparison tool to inspect a single period of observation, but the analysis of trends over time may require a different methodology.
Table of Contents
Introduction -- Literature review -- Data sources and pre-processing -- Geostatistical analysis methodology -- Results and discussion -- Conclusion