Land suitability evaluation for rubber in GIS platform and multicriteria decision-based model

To sustain the growing population and the competitive demand for land, there is the need to develop optimal land evaluation approach to identify suitable locations for rubber crop that can provide high yield. Proper match of land quality requirements, crop growth and land capabilities will allow achieving maximum yield, and eventually high economic returns. The aim of this study is to develop a land suitability evaluation model to identify optimal locations for rubber farming using Geographic Information System (GIS) and Multi-Criteria Decision Method (MCDM).The land suitability assessment is based on FAO (Food and Agriculture Organization) framework of 1976; with some modifications to comply with the Malaysian rubber crop land specifications. The model is based on a classification structure rather than a set of guidelines provided in the FAO framework. Land characteristics, grouped into nine land qualities and their threshold values were considered using datasets (soil type, soil productivity and drainage, rainfall data, elevation and slope) obtained from different national agencies. Each of the data with their associated sub-criteria represents input layer integrated into GIS environment and analyzed using multi-criteria decision making (MCDM) technique. Weighting factors for the input layers were determined based on expert opinions through analysis of the feedback from the questionnaire administered to the experts at the Malaysian Rubber Board (MRB). The result is a model, rubber land suitability evaluation model (RLSEM), that produces rubber land suitability map of Seremban district, an administrative unit in Negeri Sembilan, Peninsular Malaysia. Performance and fitness analysis of the model shows that the model is sensitive to detecting suitable and non-suitable land for rubber cultivation with sensitivity and specificity values of 84.14% and 76% respectively. Overall, assessment of the detection accuracy using the area under the ROC curve yielded (80%) and p-value <0.0001. For performance evaluation using regression models, the corrected Akaike’s information criteria agrees at both the global ordinary least square (OLS) model and local geographically weighted regression (GWR) model with AICc of 521. Also, the adjusted R2 measures of both the OLS and GWR models produced the same value, 0.802811. Correlation of the generated and the predicted land suitability models shows high positive relationship with correlation coefficient of 0.99. This implies that the land suitability model developed remained consistent from global to local model. Quantitatively, a total of 35575 hectares, distributed among the three suitability classes: highly suitable 45% (16048 hectares), moderately suitable 43% (15399 hectares), and marginally suitable 12% (4128 hectares) was obtained. Based on the World Bank monthly rubber market price projection at national level of 1.858 USD per kilogram for the month of June 2017 (for Singapore/Malaysia), it is estimated that ~28.9 million USD can be generated annually, if the available suitable land is put to use.

Preview Text fk 2018 82 ir.pdf Download (1MB) | Preview

Actions (login required)

View Item