Monitoring of agricultural land abandonment using remote sensing technology

Abandonment of agricultural land is a global issue, it is a waste of resources and brings negative impacts on local economy. It is also a key factor in certain environmental problems, such as soil erosion and increasing carbon sequestration. In order to address such problems related to land abandonment, their spatial distribution must first be precisely identified. This study utilized the multi-temporal of Landsat, ALOS PALSAR-1 and 2, and SPOT-6 images together with crop phenology information, to identify abandoned paddy, rubber and oil palm areas. With the advancement of object-oriented classification, the identification was done semi-automatically by developing rules set. The results indicated that time series images were highly beneficial in identifying unique phenology of abandoned and non-abandoned paddy area whereas, a minimum of three time-series images during planting seasons were required. In identification of abandoned rubber area, the use of image acquired during defoliation (leaf-off) phase was preferable to avoid misclassification. However, in identification of abandoned oil palm, homogeneity measures obtained from high spatial resolution satellite image was more important than high spectral resolution satellite image. The study demonstrated the advantages of using multi-temporal Landsat imagery in identifying abandoned paddy and rubber areas wherein accuracies of 93.33% ± 14% and 83.33% ± 1%, respectively, were achieved. For multi-temporal of ALOS PALSAR-1 and 2, accuracies obtained for these abandoned lands of paddy, rubber and oil palm were 93.33% ± 0.06%, 78% ± 2.32%, and 63.33% ± 1.88%, respectively. For classifying abandoned oil palm, SPOT-6 showed an accuracy of 92% ± 1%. As a conclusion, remote sensing technology have shown the potential to monitor agricultural land abandonment and is greatly useful especially when large and inaccessible areas are involved. This study confirmed that the understanding of crop phenology in relation to image date selection is essential to obtain high accuracy for classifying abandoned and non-abandoned agricultural crops. Generally, this study had successfully developed a semi-automated technique of land abandonment identification using multi-resolution and multi-temporal satellite images. With the robustness of the techniques developed based on the image rules set and crop phenology, the monitoring of agricultural land abandonment can be expanded nationwide by adjusting the rules set.

Preview Text fp 2016 53 ir.pdf Download (1MB) | Preview

Actions (login required)

View Item