Spatio-temporal similarity analysis strategy of SAR image time series for land development intensity monitoring

Abstract

Land development intensity is one key indicator of Major Function Oriented Zoning (MFOZ). For land development intensity monitoring in a large area, SAR image time series with medium resolution provides an appropriate way, because of its large-scale and high-temporal frequency measurements. According to time-series characteristics of cultivated land and construction land pixels, a spatio-temporal similarity analysis strategy considering mixed pixels and noise is presented to extract change nodes and change pixels. This strategy mainly includes three components: (1) Construction of pixel-level SAR image time series; and (2) Iterative binary partition mean square error (MSE) model to ascertain change nodes; (3) Spatio-temporal similarity analysis based on pixel-level SAR image time series to determine the change range of cultivated land to construction land. Through the monitoring of conversion of cultivated land to construction land across multiple periods leveraging pixel-level SAR image time series in Chengdu, several conclusions can be drawn from this study. (1) This study has illuminated the utility of pixel-level SAR image time series for land development intensity monitoring, especially in those areas with cloud cover the majority of the time. SAR images are not affected by cloud cover and provide continuous time-series information. (2) The spatio-temporal similarity measure was able to effectively extract change nodes and change range of cultivated land to construction land. Generally, the correctness of 85.82% and completeness of 84.78% were achieved.