Optimized Surface Water Extraction for Planform Morphology Assessment of Pagsanjan-Lumban Watershed

Abstract. This study focused on quantifying changes in river morphology using a remote sensing approach. Historical river planforms were derived from Landsat images processed through Google Earth Engine platform. Five water indices were used: the MBWI, NDWI, MNDWI, AWEIshadow and AWEInoshadow. To automatically extract surface image, combined OTSU-thresholding and Edge-Detection Approached was employed. The analysis covered the years 1990 to 2023 at 5-year interval. Baseline river planform was established using a 1970 topographic map from NAMRIA. For the sinuosity analysis, the entire stretch of Pagsanjan River was divided into 10 1-kilometer segments, starting from the first river junction of the main river to the outlet. Results show that the most significant change is observed near the outlet between the baseline year 1970 and 1990. An eastward shift was also observed during this year. Aside from these observed changes, no substantial changes in sinuosity or channel migrations were detected from 1990 onward. This stability can be attributed to the implementation of riverbank stabilization in the early 1990s. These findings demonstrate the effectiveness of remote sensing in morphological analysis and emphasize the long-term importance of river stabilization measures in maintaining channel. This is particularly important in a country, such as the Philippines, which is frequently visited by Typhoon which often lead to increase in channel discharge and subsequent changes in channel planforms.