[M-AG32] Satellite Land Physical Processes Monitoring at Medium/High/Very High Resolution

[MAG32-P06] Mapping Area Changes of Ponds Using Stacks of Optical Satellite Images

*Youg-Sin Cheng¹, Chi-Farn Chen¹ (1.Center for Space and Remote Sensing Research, National Central University)

Keywords:ponds, stack images, optical satellite images

Ponds are an important and dynamic component of terrestrial water storage, and the area change of ponds is related to climate and human activities. In Taiwan, Taoyuan City has thousands of ponds over the terrace area with the highest density of ponds. Official statistics from the 1910s reported approximately 8,000 ponds, a number that has dwindled to around 2,851 ponds in 2011. These ponds, responsible for collecting and storing precipitation and stream water, fulfill essential functions such as flood prevention, agricultural irrigation, groundwater recharge, and climate regulation. However, due to climate change and urban development, the change and disappearance of pond area has become one of the important issues in water resources management in recent years. From 2019 to 2023, Taoyuan City experienced the most severe drought event over the past 70 years and the increasing urban development. Therefore, this study explores a method for mapping and detecting changes in pond area using time series and object-based optical satellite image analysis through deep learning methods. The stack SPOT images spanning from 2019 to 2023 were utilized to construct a water area detection model. This model considered the composition of pixel mean values of objects, such as minimum or maximum values, brightness and entropy, to address common challenges associated with water mapping from optical satellite data, including issues related to clouds or shadows. Meanwhile, the study was designed for two temporal scales on the annual and seasonal basis for observation of the 2,851 ponds. The annual results indicated 2% of pond numbers transferred to building or urban planning area which decreased 26 hectares of pond area during 5 years. The impact of pond area decline was attributed to the long-term expansion of urban development and human activities mostly, and the changes were permanent. In contrast, the seasonal results revealed the short-term impacts of climate and seasonal variations in pond area, and these changes were cycles. For instance, from autumn of 2022 to the winter of 2023, the total area of ponds in sequence was 2,011 hectares (autumn), 2,209 hectares (winter), 2,094 hectares (spring), 2,095 hectares (summer), 2,129 hectares (autumn), 2,212 hectares (winter). The area of ponds would have the most in winter and decrease in spring and summer, then raised in autumn to winter. This study showcased how the proposed method can effectively detect changes in pond area and generate a comprehensive pond inventory, utilizing Taoyuan City, Taiwan, as an illustrative example and analyzing the spatiotemporal changing factors of ponds.