Deforestation Susceptibility Modelling Using Logistic Regression: In Buikwe District, Uganda.
Abstract
Forests are some of the most important ecosystems because of the services they offer, however,
they continue to be lost due to both natural and man-made causes. Due to the forest loss scare,
many studies have been done about the reduction in forest cover and the rate of destruction.
These studies have left out predictability which is important for environmental managers.
Predictive data shows areas of potential change, so that the environmental managers can focus
their attention on the predicted areas. In Uganda, forests that require conservation priority need
to be identified so that the National Forest Plan of 2011/12 to 2021/22 and the United Nations
SDG 15 can be attained. Therefore, the main aim of this research was predicting areas in
Buikwe district that are susceptible to deforestation. The specific objectives of this research
were; to determine the causal factors of deforestation in Buikwe district, to determine the forest
cover changes; (2011-2016) and (2016-2021), and to determine deforestation susceptibility
zones in Buikwe district.
The literature both in this field and related fields was reviewed to identify factors that influence
deforestation. Decision on factors was done basing on the researcher’s findings, conclusions
and recommendations. Forest cover was obtained by classifying Landsat images of different
years. Logistic regression was used to analyse the relationship between the identified causal
factors and deforestation. Factors with stronger relationships were selected to predict areas that
are susceptible to deforestation. The accuracy of the model was assessed using a ROC curve.
Causal identified factors were; slope and elevation, distance from; built-up areas, roads, and
agricultural areas. Forests were covering 36%, 25% and 23% of Buikwe district in 2011, 2016
and 2021 respectively.
Use of more advanced methods in future studies in this field was and revision of the Mabira
forest management plan were recommended.