5:15 PM - 6:45 PM
[ACG36-P16] Development of Soil Moisture Indices Using Satellite-sensed Spectra of herbaceous plants in Drained Peatland
Keywords:soil moisture , herbaceous plants, peat soil , vegetation index , satellite remote sensing
Peatland fires in Southeast Asia release an average of 1.4 billion tons of greenhouse gases annually and contribute significantly to climate change. The fires not only propagate through vegetation but also extend into peat soil. Therefore, a fundamental step in fire prediction involves comprehending soil moisture content (SM). While SM has historically been gauged through direct sensor measurements and groundwater level assessments, the extensive nature of peatland areas has fueled interest in remote sensing for SM estimation. In particular, a precise understanding of SM distribution in the surface layer (0-20 cm) is considered crucial.
Remote sensing techniques for SM assessment encompass synthetic aperture radar, infrared (1.2 to 1.5 μm) reflectance, soil surface temperature, and water indices derived from soil spectra. However, their applicability to vegetation-dense regions remains to be a challenge. Although methods have been proposed to estimate SM in vegetated areas based on changes in the normalised vegetation index (NDVI), the NDVI incorporates information beyond SM and may lack high accuracy. Additionally, differences in root depths between herbaceous plants (approximately 20 cm) and trees (exceeding 1 m) may lead to varying responding times and correlating soil depths, reducing index accuracy.
This study focuses on herbaceous areas in peatlands, where we expect to observe a strong correlation between herbaceous plants and surface SM. Our objective is to formulate an operational index highly correlated with surface SM, utilising ground-measured soil moisture data and spectra derived from satellite imagery.
In a peatland oil palm field in Sumatra, Indonesia, soil moisture data at a depth of 10-20 cm, measured by HydroProbe sensors, have been conventionally obtained. Herbaceous areas were extracted by combining RGB images from Sentinel-2 and tree height information from Gedi. Leveraging four different bands of Sentinel-2, a new index was developed, exhibiting a correlation coefficient exceeding 0.7 with measured soil moisture (r = 0.786, n = 13, p = 0.0014). The outcomes of this study hold potential for applications in peatland fire-hazard mapping across Southeast Asia.
This research was partially supported by the Telecommunications Advancement Foundation under the "Development of Human Resource Education Methods with SDG Problem-Solving Capabilities Using Both ICT and Hands-on Approach (Shiga University of Medical Science, FY2022) and by the NPO Super Scientist Program Plus.
Remote sensing techniques for SM assessment encompass synthetic aperture radar, infrared (1.2 to 1.5 μm) reflectance, soil surface temperature, and water indices derived from soil spectra. However, their applicability to vegetation-dense regions remains to be a challenge. Although methods have been proposed to estimate SM in vegetated areas based on changes in the normalised vegetation index (NDVI), the NDVI incorporates information beyond SM and may lack high accuracy. Additionally, differences in root depths between herbaceous plants (approximately 20 cm) and trees (exceeding 1 m) may lead to varying responding times and correlating soil depths, reducing index accuracy.
This study focuses on herbaceous areas in peatlands, where we expect to observe a strong correlation between herbaceous plants and surface SM. Our objective is to formulate an operational index highly correlated with surface SM, utilising ground-measured soil moisture data and spectra derived from satellite imagery.
In a peatland oil palm field in Sumatra, Indonesia, soil moisture data at a depth of 10-20 cm, measured by HydroProbe sensors, have been conventionally obtained. Herbaceous areas were extracted by combining RGB images from Sentinel-2 and tree height information from Gedi. Leveraging four different bands of Sentinel-2, a new index was developed, exhibiting a correlation coefficient exceeding 0.7 with measured soil moisture (r = 0.786, n = 13, p = 0.0014). The outcomes of this study hold potential for applications in peatland fire-hazard mapping across Southeast Asia.
This research was partially supported by the Telecommunications Advancement Foundation under the "Development of Human Resource Education Methods with SDG Problem-Solving Capabilities Using Both ICT and Hands-on Approach (Shiga University of Medical Science, FY2022) and by the NPO Super Scientist Program Plus.