Application of Random Forest Classification Algorithm and Time Series Decomposition to Assess the Impacts of Climate Change on Forests in Chimanimani District Using Landsat Data.
- Author
- Matamande, Thania
- Title
- Application of Random Forest Classification Algorithm and Time Series Decomposition to Assess the Impacts of Climate Change on Forests in Chimanimani District Using Landsat Data.
- Abstract
- Climate change poses a significant threat to forest ecosystems worldwide with far reaching consequences for biodiversity, ecosystem services and human livelihoods. Despite the importance of forests in Chimanimani district the impacts of climate change on the ecosystem remain poorly understood. The study aims to assess the impacts of climate change on forests in Chimanimani district using time series decomposition and Random Forest classification algorithm. The study was focused on Chimanimani district located in the eastern parts of Zimbabwe as the area of study. Data was collected from Landsat satellite imagery (USGS) and climate data from the Zimbabwe Meteorological Service Department. Landsat data from 1994 to 2024 downloaded from USGS and applied in the thesis. A total of 30 Landsat images were downloaded and classified using Random Forest algorithm. Results on changes in forest cover were determine in % percentages from 1994-2024 and linear regression analysis was used giving a negative correlation (𝑅2 = 0,8964). Whilst climate data from 1994 to 2023 and Land cover results were analyzed to show relationship between climate change and change in forest Land cover. The results showed that forest cover in Chimanimani district decreased by 31.77% from 1994 to 2024. Also non forest area has increased with 8.29% in 1994 and 40.06% in 2024 which give a -31.77% increase. The study also show results of time series decomposition of time series decomposition using R software 4.3.3. The thesis presents time series, smoothed time series and decomposed time series plot. Therefore climate data was found to be significantly correlated with forest cover changes. The study also revealed that local communities perceived climate change as a major driver of forest degradation in the district. The results suggests that climate change poses a significant threat to forest ecosystems in Chimanimani district. In conclusion the study demonstrates the importance of using machine learning algorithms and time series decomposition to assess the impacts of climate change on forest ecosystems. Therefore it is recommended that policy makers and conservationists prioritize the use of these techniques in developing effective climate change mitigation and adaptation strategies for forest ecosystems in Zimbabwe.
- Date
- June 2025
- Publisher
- BUSE
- Keywords
- Climate Change
- Classification Algorithm
- Time Series
- Supervisor
- Dr Pedzisai
- Item sets
- Department of Disaster Risk Reduction
- Media
-
Matamande, Thania.pdf