Research Article
Integration of IK, Satellite Imagery Data, Weather Data and Time Series Models in Season Behaviour Predictions. Case of Swayimane, KZN, South Africa
@INPROCEEDINGS{10.1007/978-3-031-63999-9_10, author={John Nyetanyane}, title={Integration of IK, Satellite Imagery Data, Weather Data and Time Series Models in Season Behaviour Predictions. Case of Swayimane, KZN, South Africa}, proceedings={Emerging Technologies for Developing Countries. 6th EAI International Conference, AFRICATEK 2023, Arusha, Tanzania, December 11--13, 2023, Proceedings}, proceedings_a={AFRICATEK}, year={2024}, month={6}, keywords={Indigenous knowledge certainty level LSTM SARIMA TES Holt Winter's model}, doi={10.1007/978-3-031-63999-9_10} }- John Nyetanyane
Year: 2024
Integration of IK, Satellite Imagery Data, Weather Data and Time Series Models in Season Behaviour Predictions. Case of Swayimane, KZN, South Africa
AFRICATEK
Springer
DOI: 10.1007/978-3-031-63999-9_10
Abstract
Impacts of climate change continue to cripple the livelihood of many South Africans by sabotaging the rainfed agricultural systems that they rely heavily on. Despite the value of the indigenous knowledge (IK) in tackling the impacts of climate change, it continues to lose value and precision in season behaviour predictions. In this paper, the IK is integrated with scientific knowledge to enhance the season predictions by small scale farmers. This is achieved by quantifying the collection and processing of IK indicators. Secondly, collect and process the regional weather data (rain and average temperature) from the weather station close to the region of interest and satellite data(vegetation cover, waterbodies cover and soil moisture cover) collected within the region of interest. Thirdly, train the following time series models: Long Short Term Memory (LSTM) network, Seasonal Auto Regressive Moving Average (SARIMA) and Holt Winter’s model on the historical weather and satellite data and perform rainy season predictions for scientific perspective. Fourthly, categorize the historical data into warm and cold rainy season periods when normal, below normal and above normal rains were experienced. Fifthly, develop the mobile application that will use the categorized historical data to complement the observation of IK indicators. Lastly, integrate the predictions made via the IK’s perspective with the ones performed via the scientific perspective to come up with more robust season predictions.
