The water levels of peatlands hold a significant role in determining greenhouse gas emissions and holding the global climate system. Water level management in peatlands is critical to preventing peatland fires and greenhouse gas emissions[1]. Peatlands in Indonesia cover more than 7% of the country’s land, therefore the use of peatlands is unavoidable. Land clearing and construction of drainage networks can damage peat, resulting in a decrease in water level, subsidence of the peat surface, CO2 emissions, land fires, and total drought (irreversible drying).
Major forest fires in 2015 burned 2,611,411.44 hectares of forest, including peatlands[2]. Handling forest fires requires a ton of money[3]. The National Disaster Management Agency (BNPB) budget in 2019 is mainly for handling forest fires, reaching 50% of the total budget of 6.7 trillion Rupiah[4].
The standard water level value is 0.4 meters[5], and if it is more than that, the peatlands are declared vulnerable and prone to fire. Therefore, it is necessary to monitor the water level of peatlands, namely by predicting their value to estimate the water level condition for the next period.
This research complements previous research in terms of utilizing deep learning LSTM. This study uses the data on water levels in peatlands which sets it apart from previous research. The resulting model is expected to become supporting information in peatland monitoring efforts and in determining policies to reduce the potential for peatland fires.
