The landscapes were classified into seven basic classifications that included three ecotypes to analyze the various landscape information in the images. The seven landscape classifications included marsh, river pond, meadow, forest, paddy field, dry farmland, and others. Among these, the selleck catalog river pond classification comprised natural rivers, ponds, and all other artificial water bodies. Few areas included residences in the study area between 1975 and 1989, although this increased in 2006. For an easier historical comparison of the different landscape classifications, residential areas, road areas, and other types of small landscapes were merged into one landscape classification termed ��other.��The data processing method for this research included constructing a new multiple-band file for georeferenced remote sensing images and a mask for the boundary of the study area within the ENVI 4.

0 platform. The mask was applied to the imagery data for the purpose of creating an image-based region of interest in the three specific dates. We utilized the layer stacking tool to construct a new file and then performed rapid filter enhancement on the images to meet the needs of image interpretation. The interpretation signs were then established, based on the images according to different colors, shapes, textures, and field investigation photographs. Manual interpretation was used to obtain the classification maps in raster format to describe the regional wetland landscapes in 1975, 1989, and 2006. The QuickBird image was used for reducing the uncertainty while manually delineating the similar landscapes, such as marsh and meadow.

After resetting the digital boundaries of four inner units as the HNNR and three farms within the study area, the three thematic maps of the wetland landscapes were reproduced for dynamic analysis purposes (Figures 2(a), 2(b), and 2(c)). An accuracy Batimastat estimation was made based on the confusion matrices generated from the database of ground truth and a variety of relevant maps (e.g., the previous land-use maps and a previous classification map of the wetlands) [16, 41]. The results of the accuracy assessment showed that the total classification accuracies reached 92.33%, 92.60%, and 90.41% in 1975, 1989, and 2006, respectively. The kappa coefficients (N = 365) were 86.66%, 89.47%, and 86.93%, respectively. Finally, a statistical analysis was performed to present the temporal and spatial changes of the regional dynamic landscapes using Excel 2003 software [41].Figure 2Changes of the wetland landscape within the past 30 years.3. Results and Discussion3.1.