Monitoring of Inefficient Land Use with High Resolution Remote Sensing Image in a Chinese Mega-City

Low efficiency as a special kind of urban construction land, which inhibits the blind expansion of urban boundaries and achieves urban construction land intensive and high efficiency. In this paper, the information feature classification of the low-cost land use was carried out by combining the Worldview-2 high resolution remote sensing images in 2011 and 2016, which was located in the central business district (CBD) of Tianhe district of Guangzhou city, China. Based on the remote sensing information extraction results of low efficiency land use of the dynamic changes. The results showed that the multi-scale segmentation was better when the segmentation scale was 15, the color parameter was 0.7, and the compactness parameter was 0.6. The low efficiency land maintained its spectral accuracy and avoided excessive fragmentation. CART classifier and SVM method had good classification results for low efficiency land, yet the results of SVM method (the total accuracy is 87.66%) were more accurate than CART classifier (the total accuracy is 80.48%), especially in green area and river. Compared with the classification results of remote sensing images in 2011 and 2016, the range of inefficient land use changed locally, and the increase of green area was more significant, which indicated that the improvement of the ecological environment in the whole area was improved greatly. In conclusion, we can effectively identify the features such as spectrum, brightness, texture and spatial geometry of low efficiency sites by using high resolution remote sensing and low efficiency monitoring of regional scale. The study can be used to provide decision support for improvement of the urban low efficiency, the internal structure optimization and the landscape ecological service.