Analysing land use/land cover change and prediction in a cloudy urban area using SAR: the case of Douala, Cameroon

Abstract. Land use/land cover change analyses and prediction remains a fundamental tool in shaping urban decision making. This is because it keeps a spatial track record of the past, present and predict the future. The city of Douala is one of those coastal cities that is under permanent cloud cover which makes it difficult to use optical sensors, therefore existing works within this area remain inadequate. The Synthetic Aperture Radar images were processed using deep machine algorithms in Google Earth Engine Platform using multiple polarisations. Moreso, Random Forest Classifier was used to classify both images. Results show that the built-up area has increased from 26 to 34% from 2016 - 2024, vegetation has drastically reduced from 38 to 30% while the areas occupied by bare land and water show a slide increase 0.08 and 0.15% respectively. In addition, land use/land cover prediction reveals that the built-up area will occupy close to half of the total surface area in 2035 (49%) and vegetation will reduce drastically to less that halve of its present state. Meanwhile bare land and water remains more or less the same compare to their presence state. The overall accuracy ranges between 80-84% and kapa between 79-85%. Thus, this can be a strategic information to both public and private agencies involved in drafting, orienting and monitoring urban growth in a cloudy environment.