Monitoring Land-use Change in a Time Series of Sequential Landsat Imagery for the Pearl River Delta, China: an Econometric Technique

In this paper, we describe a new change detection method that determines the date at which land-use changes occur in a sequential series of Landsat TM images. We use this approach to extract annual rates of land conversion in the Pearl River Delta, China, for the period 1988-1996. The technique is a three-step change detection procedure that uses time series and panel econometric methods. First, regression equations are estimated for each of the six TM bands for unchanged, stable land-cover classes. Second, the regression equations for these stable land-cover classes are used to calculate DN values for change classes for each of the eight possible dates of change (1989-1996). Lastly, the year of land-use change is identified by comparing a pixel’s DN values against the eight possible dates of land conversion using tests for predictive accuracy. The accuracy of the dates of land-use change identified by the econometric technique is upwards of 85 percent. Additionally, the econometric technique may reduce efforts required to assemble the training data and to correct the images for atmospheric effects.