Influence of socioeconomic drivers on land use and cover changes using
Remote sensing and GIS in Dar es Salaam, Tanzania
Abstract
Understanding human interactions demands modelling human-environment
interactions. This study uses remote sensing and machine learning to
evaluate land use and land cover (LULC) changes over 27 years in Dar es
Salaam Metropolitan City, Tanzania, and the spatially varying
relationships between LULC changes and socioeconomic driving factors.
LULC change values and factors are retrieved from data points generated
by regular sampling methods. The geographically weighted regression
(GWR) model is then employed to analyse the relationships between LULC
changes and the identified factors. The analysis of LULC changes reveals
a dynamic transformation in land cover between 1995 and 2022,
characterised by a notable 14.9% increase in built-up areas and a
corresponding decline of 14.6% in bushland. 65.8% of the land cover
experiences gains and losses, while 34.2% remains relatively stable
over the 27 years. The GWR model surpasses the OLS model, achieving an
R2 value of 0.73, signifying a strong
association between LULC changes and the identified socioeconomic
factors, explaining 73% of the LULC variation. Additionally, the
influences of these factors, including the signs, significances, and
coefficient values, exhibit considerable variations across different
LULC change types. Notably, population density and proximity to the city
centre significantly contribute to LULC changes, whereas the impact of
gross domestic product and distance to roads is comparatively lesser.
Moreover, poverty does not significantly drive LULC changes. This
study’s findings suggest that urbanisation and urban sprawl, as
indicated by population density and distance from the city centre,
significantly influence land cover changes in the study area.