Exploring the potential of soil moisture maps using Sentinel Imagery as a Proxy for groundwater levels in peat

Degraded peatlands are extremely vulnerable to the threat of fires and have been a major source of national greenhouse gas emissions. Maintaining a certain level of water in peatlands is an essential measure of disaster vulnerability in peatlands. During the dry season, when the lower part of the peat still retains water, fires only occur on the surface and are relatively easy to extinguish. However, one of the limiting factors in peatland management and its more comprehensive application has been the availability of sufficient and spatially distributed Groundwater Level (GWL) data. This study explores the soil moisture map as a proxy for peat condition indicators that correlate with groundwater level. The case studies conducted at Tumbang Nusa Research Forest and Peat Hydrological Unit of Kahayan Sebangau show that peatland conditions can be estimated through biophysical parameters detectable from remotely-sensed data. Soil Moisture Map (SMM) can be produced with a higher resolution (Sentinel 1 = 10m) using the free and open tools SEPAL based on cloud computing infrastructure. The Support-Vector-Regression machine learning approach is used to estimate soil moisture. There is a correlation between SMM and GWL. However, the response to land cover varies. There is high uncertainty in densely forested areas where the sensors cannot penetrate the canopy. As a result, in its implementation, the SMM can be combined with the vegetation index, which can describe trends of land cover changes.