Geoinformation Modelling of Peri-Urban Land Use and Land Cover Dynamics for Climate Variability and Climate Change in the Bosomtwe District, Ghana

Abstract

This study modelled the socioeconomic drivers and the actual land use and land cover (LULC) changes, with special reference to vegetation cover and local climate variability and change, in the peri-urban Bosomtwe District of the Ashanti region of Ghana. A triangulation of qualitative and quantitative design was used among 270 household respondents. Using a multistage sampling technique, partially pre-coded questionnaires were administered in 14 communities. Data were analyzed using non-parametric tools as Pearson’s Chi-square, Nagelkerke R2 and Cramer’s V, and step-wise binary logistic regression analyses embedded in the Statistical Package for Social Scientists (SPSS v.16). Remote Sensing and Geographic Information System algorithms were used in ENVI, ERDAS Imagine ESRI and ArcGIS environment, to classify LULC, using maximum likelihood classifier to analyze Landsat Thematic Mapper and Enhanced Thematic Mapper+ images, and Landsat 8, Operational Land Imager and Thermal Infrared Sensor (OLI/TIRS) images of 1986, 2002, 2007/2008, 2010 and 2014 respectively. The best Kappa hat statistic of classification accuracy is 83%. A Mann-Kendall trend analysis was also done on the climatic data, using MAKESEN Solver in EXCEL. At a confidence level of 99.9%, perceived drivers of peri-urban land use changes in the communities were significant at p < .000. The Cramer’s V test of a strong association between two nominal variables was 0.412. The logistic regression model reported confidence intervals (CIs) of 1.218 ≤ CI ≤ 4.234, for the three main predictors of the land use conversion outcome, indicating the odds of up to 4 times of conversion likelihood. The normalized difference vegetation indices (NDVIs) for forest cover, plantation and other cultivated lands ranged between 0.384 and 0.570; while recent fallows and grassland as well as bare lands had NDVI values of between 0.081 and 0.250. Over the period, dense forest decreased by 1380 ha in 2014, while low forest (which included some oil palm and citrus plantation farms), increased by 1766 ha in 2014. However, Built up, bare and concrete land uses increased by 3360 ha, while recent fallows and grasslands decreased by 3356 ha, over the same period. Inter-quarterly rainfall and temperature anomalies depicted generally increasing trends in the mean iii monthly temperature over the last two inter-decadal periods (1990-2000 and 2001 to 2011). Land surface temperature profile extracted from the satellite images cohered with the LULC characteristics, expressed in the novel concepts of Rural Cool Troughs (RuCT) and Peri-urban Heat Troughs (PuHT). The study indicates major changes in the deteriorating livelihoods, easy access to, as well as changing demand for land are affecting peri-urban landscape of the Bosomtwe district. The study recommends that a strict enforcement of district land use plans by the Bosomtwe District Assembly. For further research, the use of high resolution images is recommended to forestall the inherent minor misclassification. Relationship between DEM and LULC in a mountainous landscape as the Bosomtwe district should be explored along-side policy mechanisms.