Agricultural Land Use Change in the Lowlands of Southern Mali under Climate Variability

Abstract

research investigated agricultural land use change in the lowlands of Southern Mali under climate variability. Four supervised classification techniques, Classification and Regression Tree (CART), Support Vector Machine (SVM), Random Forest (RF) and Gradient Tree Boosting (GTB) in Google Earth Engine (GEE), were used for the image classification. An integrated Cellular Automata-Artificial Neural-Network (CA-ANN) within the MOLUSCE plugin of QGIS was used for future Land Use and Land Cover prediction. The Mann-Kendall test, Sen’s slope, Pettit-test and change-point detection analyse were applied for climate variability assessment. Monthly rainfall and mean temperature extending over a period of 61 years (1960–2020) recorded at Sikasso District were analysed. Annual rainfall varied between 800 mm to 1600 mm and annual mean temperature ranged between 25 oC to 28 oC. Seasonal rainfall ranged between 37-387 mm, March-April-May (MAM), 400-1030 mm, June-July-August (JJA), 77-577 mm September-October-November (SON) and 0-45 mm for December-January-February (DJF). Mean seasonal temperature ranged from 29 oC to 32 oC (MAM), 26.5 oC to 28.5 (JJA) oC and 26 oC to 28 oC (SON). Annual and seasonal rainfall trends increased slightly. Temperature showed a significant increase in both annual and seasonal trends. Out of 395 respondents, 79 % were of the view that annual rainfall decreased while 83 % reported mean temperature increased. Again, respondents perceived late onset rainfall (97 %), early cessation of rainfall (96 %), increased in drought (83 %) and flooding (96 %). Also, 43 % of respondents adopted new varieties to cope with climate variability. The findings showed that physical and socioeconomic driving forces had impact on terrain patterns. Over the past three decades, the study revealed that apart from cropland area which increased from 43.81 % to 52.75 %, the size of the other land uses decreased, forest cover (19.93 % - 13.93 %) shrubs (16 % - 14 %), and streams (6 % - 4 %). However, the forecast for the 2020 to 2030 predicted an increasing trend in forest cover and decreasing trend in agricultural land in the study area due to the ongoing afforestation projects. The study demonstrates the need to reinforce regional land management policies and programmes.