Modelling future land-use change and assessing resultant streamflow responses: a case study of two diverse Southern African catchments.

Abstract

Land-use and land cover (LULC) is a crucial constitute of the terrestrial ecosystem, impacting on numerous fundamental processes and characteristics such as land productivity, geomorphological process and the hydrological cycle. Assessing the hydrological impacts of land-use and land cover changes (LULCCs) has become one of many challenges in hydrological research. LULCCs modify hydrological processes such as evapotranspiration, infiltration and interception, consequently impacting on the hydrological regimes of a catchment. Understanding the implications of LULCCs on catchment hydrology is therefore fundamental for effective water resource planning and management, and land-use planning. Globally, numerous studies have documented the impacts of LULCCs on catchment hydrology, however in Southern Africa there exists a knowledge gap on the impacts of LULCCs on catchment hydrology, specifically future land-use and land cover change (LULCC). Therefore, the aim of this study was to simulate potential future land-use within two diverse South African catchments using an appropriate land-use change model and thereafter to assess streamflow responses to these future land-use scenarios using the ACRU hydrological model. Future land-use was simulated utilizing the Cellular Automata Markov (CA-Markov) model. The CA-Markov model is a hybrid land-use change model that integrates Markov chain, CA, Multi-Objective Land Allocation (MOLA) and Multi-Criteria Evaluation (MCE) concepts. CA-Markov simulated future land-use through the creation of conditional probability and transition probability matrices, suitability images and the utilization of a CA contiguity filter and socio-economic and biophysical drivers of LULCC. The results illustrated that within both catchments, increasing growth of anthropogenically driven LULC classes such as urban, agroforestry and agrarian areas inevitability contribute to the fragmentation, modification and deterioration of natural land-cover types. The model’s reliability and capability was assessed by running a validation, which was conducted by simulating changes between t1 (1990) and t2 (2013/14) to predict for t3 (2018). The predicted map produced for 2018 was then compared against the actual 2018 reclassified map, which served as a reference map. The obtained kappa values (Kstandard, Klocation and Kno) achieved during the validation were all above 80%, thus indicating the model’s reliability and capability in successfully predicting future LULC in the study sites. The assessment of future LULCC impacts on streamflow responses was achieved by utilizing the ACRU model. Historical and future scenarios of land-use were utilized as inputs into a preexisting ACRU model where all input parameters (e.g. climate, soils) remained constant with only changes made to the land cover parameters and area occupied by each land cover. The results illustrated that due to anthropogenic induced LULCC, the hydrological regime within the uMngeni catchment has been altered when compared to the baseline hydrological regime. Patterns of low (1:10 driest year) and high (1:10 wettest year) flows have changed significantly between the baseline and 1990. However, between 1990 and the future hydrological regime (2030 LU scenario) only a slight amplification of these impacts was evident. Mean annual streamflow increases and decreases were present in majority of Water Management Units (WMU’s), however, the Table Mountain, Pietermaritzburg, and Henley WMU’s illustrated greater increases in mean annual accumulated streamflows compared to other WMU’s while the New Hanover New Hanover and Karkloof WMU’s illustrated the greatest decreases in mean annual accumulated streamflows. Furthermore, results indicated that streamflow responses significantly increase in the presence of urban land-use. The impacts become evident as streamflows cascade through the catchment. The results also illustrated that streamflow responses were due to the nature of LULCC, viz urban land-use, commercial forestry, and agriculture combined with the location and extent of LULCCs. These results are beneficial for the implementation of proactive and sustainable water resource planning and land-use planning. Moreover, considering the simulated streamflow responses in relation to varying land-use scenarios, it is essential that water resource planning incorporate land-use location, nature and scale from not only the perspective of land-use effects, but also on hydrological responses in a catchment. Given the interdependence between streamflow responses and changes in land-use, water resource and land-use planning should not occur in silos. Overall, this study illustrated the importance of understanding and assessing land-use and water interactions in a water stressed region such as South Africa. Keywords: land-use and land cover changes, hybrid land-use change model, streamflow responses, land-use and water interactions, sustainable water resource planning.