You are here:
Sub-Pixel Mapping of Tree Canopy, Impervious Surfaces, and Cropland in the Laurentian Great Lakes Basin Using MODIS Time-Series Data
Citation:
Shao, Y. AND R. S. LUNETTA. Sub-Pixel Mapping of Tree Canopy, Impervious Surfaces, and Cropland in the Laurentian Great Lakes Basin Using MODIS Time-Series Data. International Journal of Applied Earth Observation and Geoinformation. Elsevier BV, AMSTERDAM, Netherlands, 4(2):336-347, (2011).
Impact/Purpose:
Land-cover (LC) types, their distributions, and their dynamics are important landscape characteristics needed for the study of terrestrial ecosystem processes, climate change impacts, and human-environmental interactions [1-3]. Recently, the Moderate Resolution Imaging Spectroradiometer (MODIS) data has been increasingly used for regional and global LC mapping [4-6]. The moderate spatial resolution and high temporal resolution attributes are particularly important for many large-area mapping applications [7]. Currently, global and regional LC map products can be routinely generated at 500 m spatial resolution and researchers are actively developing maps and change detection products at 250 m resolution [5], [8-10].
Description:
This research examined sub-pixel land-cover classification performance for tree canopy, impervious surface, and cropland in the Laurentian Great Lakes Basin (GLB) using both timeseries MODIS (MOderate Resolution Imaging Spectroradiometer) NDVI (Normalized Difference Vegetation Index) data and surface reflectance data. Classification training strategies included both an entire-region approach and an ecoregion-stratified approach, using multi-layer perceptron neural network classifiers. Although large variations in classification performances were observed for different ecoregions, the ecoregion-stratified approach did not significantly improve classification accuracies. Sub-pixel classification performances were largely dependent on different types of MODIS input datasets. For sub-pixel tree canopy, the time-series MODIS-NDVI composite data generated the best overall performance (R2 = 0.66). The combination of MODIS surface reflectance bands 1 (620–670 nm; 250 m), 2 (841–876 nm; 250 m), and 6 (1628–1652 nm; 500 m) generated the highest accuracies for sub-pixel estimations of impervious surface (R2=0.59) and cropland (R2=0.60). The use of individual date MODIS images were also examined with the best results being achieved for Julian days 177, 209, and 113 for tree canopy (R2=0.56), impervious surface (R2=0.44), and cropland (R2=0.56), respectively. The R2 values for individual date results were much lower than those derived using the entire time-series MODIS data. Also, it was determined that the spatial aggregation from 250 m to 500 m generally improved sub-pixel classification accuracies.