Abstract
This research established an empirical methodology for estimating higher-resolution soil moisture using GIS and frequency ratio (FR) modeling techniques. Soil moisture active passive (SMAP) Level-4 global 3-hourly 9 km spatial resolution surface and root zone soil moisture datasets were used as reference data. A total of 283 reference points were selected through spatial fishnet analysis with the root zone soil moisture over 0.35 and surface soil moisture over 0.30. Eighty percent (80%) of these reference points served as inputs to the FR model, with the remaining twenty percent (20%) reserved for validation. Key independent variables incorporated in the FR modeling process included land use land cover, soil texture, normalized difference vegetation index, land surface temperature, topographic wetness index, rainfall, elevation, slope, and distance from rivers. The study area encompassed the final drainage basin of the Markham River catchment, situated in the Morobe Province of Papua New Guinea. The high-resolution developed database on surface soil moisture was reclassified into five basic zones segmenting on the FR index value, namely very low (less than 6), low (6–7), moderate (7–8), high (8–9), and very high (More than 9). The result indicates almost 26.10% of the land area is classified as a high soil moisture class and 56.89% as a very high soil moisture class. The FR model evinced a prediction accuracy of 93.98% along with a succession rate of 91.59%. These results provide useful data for scientific applications in various domains, specifically in the agricultural sector, local government administrator, researcher, and planner.