Assessment and Prediction of Climate Variability Impact on Land Use Land Cover Change in Sikasso Region, Mali

Abstract

Climate change and variability are worldwide phenomena and their impact is different in nature from one region to the other. In that context, this study focused on the assessment and prediction of climate variability impact on Land Use Land Cover Change (LULCC) in Sikasso region, Mali with focus on agricultural lands. Three objectives were achieved in this study: (1) assess changes in LULC, (2) examine climate variability and its impact on agricultural LULC and (3) predict future changes in LULC by 2030 and 2050. The dataset composed of time series satellite images from Moderate Resolution Imaging Spectroradiometer (MODIS) Terra for the years 2000, 2008 and 2016, monthly rainfall and temperature from 1981 to 2016 for the four main meteorological stations across the study area and socioeconomic information. The Savitzky-Golay (SG) filtering process (smoothing) was performed on Normalised Difference Vegetation Index (NDVI) time series images with TimeSat software and an ISODATA classification scheme adopted for four main classes which are cropland, vegetation, water and others. Standardised anomaly, Coefficient of Variation (CV) and Modified Mann-Kendall (MMK) trend test were used to analyse rainfall and temperature data. Pearson's Chi-square test of association was performed on questionnaire data to determine whether climate variability has impact on LULCC and the prediction was carried out using Cellular Automata (CA)-Markov model. The LULCC analysis showed that agricultural lands increased by 4 % (129,665 ha) between the year 2000 and 2016 and the vegetation cover decreased by -1 % (30,000 ha) during the same period; water bodies increased and the class others decreased. The expansion of agricultural lands and decreases in vegetation cover are expected to continue. Furthermore, the mean temperature increased from 1981 to 2016 at the rate of 0.3 °C per decade and the minimum temperature recorded the highest rate of increase (0.44 °C per decade); on monthly basis, the highest deviations in the temperature were observed in the months of November (+1.24 °C), March (+0.69 °C) and October (+0.67 °C) while lowest was observed in the month of February (+0.15 °C). At 5 % significance level, an increasing trend was detected in the regional annual average rainfall and the amount of rainfall during the rainy season (for years after 2010) was considerably higher than the climatological mean-normal (1981-2016) except the years 2011 and 2013. The LULC model revealed that cropland will increase by 6.54 % (217,599 ha) between the period 2016-2030 and 18.58 % (618,179 ha) in 2016-2050. Vegetation will decrease by -11.14 % (-357,149 ha) between 2016-2030 and by -34.49 % (-1,105,814 ha) by 2050. Generally, the observed increment in annual and seasonal rainfall was not the primary factor for the expansion of agricultural lands as questionnaire analysis revealed that farmers' decisions to bring changes in their farms size was rather a function of market prices, changes in production systems, access to improved seeds and number of male workers. The intensification of LULCC as apparent from the model predictions and spatio-temporal climatic pattern signals the need for the development of mitigation and adaptation strategies that will minimize the sensitivity and exposure as well enhance the resilience of the Sikasso region to the anticipated changes. Further study should address rainfall variability in terms of its intra seasonal distribution and impact on agricultural production in the region.