Novel Hybrid Integration Approach of Bagging-Based Fisher’s Linear Discriminant Function for Groundwater Potential Analysis

Natural Resources Research - Groundwater is a vital water source in the rural and urban areas of developing and developed nations. In this study, a novel hybrid integration approach of...     

Soil erosion prediction based on land cover dynamics at the Semenyih watershed in Malaysia using LTM and USLE models

This study attempts to identify and forecast future land cover (LC) by using the Land Transformation Model (LTM), which considers pixel changes in the past and makes predictions using influential spatial features. LTM applies the Artificial Neural Networks algorithm) in conducting the analysis. In line with these objectives, two satellite images (Spot 5 acquired in 2004 and 2010) were classified using the Maximum Likelihood method for the change detection analysis. Consequently, LC maps from 2004 to 2010 with six classes (forest, agriculture, oil palm cultivations, open area, urban, and water bodies) were generated from the test area. A prediction was made on the actual soil erosion and the soil erosion rate using the Universal Soil Loss Equation (USLE) combined with remote sensing and GIS in the Semenyih watershed for 2004 and 2010 and projected to 2016. Actual and potential soil erosion maps from 2004 to 2010 and projected to 2016 were eventually generated. The results of the LC change detections indicated that three major changes were predicted from 2004 to 2016 (a period of 12 years): (1) forest cover and open area significantly decreased at rates of almost 30 and 8 km², respectively; (2) cultivated land and oil palm have shown an increment in sizes at rates of 25.02 and 5.77 km², respectively; and, (3) settlement and Urbanization has intensified also by almost 5 km². Soil erosion risk analysis results also showed that the Semenyih basin exhibited an average annual soil erosion between 143.35 ton ha^?1 year^?1 in 2004 and 151 in 2010, followed by the expected 162.24 ton ha^?1 year^?1. These results indicated that Semenyih is prone to water erosion by 2016. The wide range of erosion classes were estimated at a very low level (0–1 t/ha/year) and mainly located on steep lands and forest areas. This study has shown that using both LTM and USLE in combination with remote sensing and GIS is a suitable method for forecasting LC and accurately measuring the amount of soil losses in the future.     

Surface runoff prediction regarding LULC and climate dynamics using coupled LTM,optimized ARIMA,and GIS-based SCS-CN models in tropical region

The effects of climate and land use/land cover (LULC) dynamics have directly affected the surface runoff and flooding events. Hence, current study proposes a full-packaged model to monitor the changes in surface runoff in addition to forecast of the future surface runoff based on LULC and precipitation variations. On one hand, six different LULC classes were extracted from Spot-5 satellite image. Conjointly, land transformation model (LTM) was used to detect the LULC pixel changes from 2000 to 2010 as well as predict the 2020 ones. On the other hand, the time series-autoregressive integrated moving average (ARIMA) model was applied to forecast the amount of rainfall in 2020. The ARIMA parameters were calibrated and fitted by latest Taguchi method. To simulate the maximum probable surface runoff, distributed soil conservation service-curve number (SCS-CN) model was applied. The comparison results showed that firstly, deforestation and urbanization have been occurred upon the given time, and they are anticipated to increase as well. Secondly, the amount of rainfall has non-stationary declined since 2000 till 2015 and this trend is estimated to continue by 2020. Thirdly, due to damaging changes in LULC, the surface runoff has been also increased till 2010 and it is forecasted to gradually exceed by 2020. Generally, model calibrations and accuracy assessments have been indicated, using distributed-GIS-based SCS-CN model in combination with the LTM and ARIMA models are an efficient and reliable approach for detecting, monitoring, and forecasting surface runoff.