Increased fire hazard in human-modified wetlands in Southeast Asia

Accuracy of tropical peat and non-peat fire forecasts enhanced by simulating hydrology

Soil moisture deficits and water table dynamics are major biophysical controls on peat and non-peat fires in Indonesia. Development of modern fire forecasting models in Indonesia is hampered by the lack of scalable hydrologic datasets or scalable hydrology models that can inform the fire forecasting models on soil hydrologic behaviour. Existing fire forecasting models in Indonesia use weather data-derived fire probability indices, which often do not adequately proxy the sub-surface hydrologic dynamics. Here we demonstrate that soil moisture and water table dynamics can be simulated successfully across tropical peatlands and non-peatland areas by using a process-based eco-hydrology model (ecosys) and publicly available data for weather, soil, and management. Inclusion of these modelled water table depth and soil moisture contents significantly improves the accuracy of a neural network model in predicting active fires at two-weekly time scale. This constitutes an important step towards devising an operational fire early warning system for Indonesia.

Development of Spatial Peatland Fire Danger Index Using Coupled SWAT-MODFLOW Model

The Keetch–Byram Drought Index (KBDI) is a numerical value reflecting the dryness of the top layer of soils, deep forest litter, logs, and living vegetation. The KBDI is expressed as a scale from 0 to 200, where the number represents the amounts of rainfall (in millimeters) to return the soil to saturation. We proposed a method to integrate peatland groundwater as a key variable for the peatland forest fire hazard, and we called it mKBDI. The groundwater table was obtained from the SWAT-MODFLOW model simulation. The MODFLOW model uses a 200 m × 200 m grid. The SWAT-MODFLOW model was calibrated and validated using daily water level measurements in the river. The model failed to represent peak flow, but the model produced the average water level. Output from the simulation was read using the FloPy module, and then mKBDI was calculated. The daily calculations of the mKBDI for each grid for the catchment were saved in the NetCDF format using the x-array module. We applied this model in the Peatland Hydrological Unit (PHU) Merang-Kepahyang, South Sumatera, Indonesia, in 2015 (El-Niño year) and 2016 (La-Nina Year). The daily mKBDI index from all the grids in the catchment was classified into three classes: low (mKBDI < 100), moderate (mKBDI = 101–150), or high (mKBDI > 150). Then, the whole catchment was classified according to these classes. Therefore, every day the percentage of the area with low, moderate, or high class in the catchment dynamically changed. When these classes were verified with hotspot data, all hotspots only coincided with the high hazard classes, where more than 60% was area of the catchment. No hotspot data were reported on low/moderate levels throughout 2015/2016. In the larger area with high mKBDI classes, the frequency of hotspots substantially rose. Sixty-three hotspots occurred during August–October of 2015 when the area of high hazard classes was above 70%. Through this finding, we proposed to use a 60% area of the catchment with high mKBDI classes as a threshold value indicating that the area is prone to peatland fire. Therefore, the peatland restoration project in preventing the fire could be evaluated using this indicator. If the restoration projects could reduce the area with high mKBDI classes to less than 60% for the whole year, we could accept it as a successful project.

Satellite-based meteorological drought indicator to support food security in Java Island

A meteorological drought refers to reduced rainfall conditions and is a great challenge to food security. Information of a meteorological drought in advance is important for taking actions in anticipation of its effects, but this can be difficult for areas with limited or sparse ground observation data available. In this study, a meteorological drought indicator was approached by applying the Standardized Precipitation Index (SPI) to satellite-based precipitation products from multiple sources. The SPI based meteorological drought analysis was then applied to Java Island, in particular to the largest rice-producing districts of Indonesia. A comparison with ground observation data showed that the satellite products accurately described meteorological drought events in Java both spatially and temporally. Meteorological droughts of the eight largest rice-producing districts in Java were modulated by the natural variations in El Niño and a positive-phase Indian Ocean Dipole (IOD). The drought severity was found to be dependent on the intensity of El Niño and a positive-phase IOD that occurs simultaneously, while the duration seems to be modulated more by the positive-phase IOD. The results demonstrate the potential applicability of satellite-based precipitation monitoring to predicting meteorological drought conditions several months in advance and preparing for their effects.

Rewetting Tropical Peatlands Reduced Net Greenhouse Gas Emissions in Riau Province, Indonesia

Draining deforested tropical peat swamp forests (PSFs) converts greenhouse gas (GHG) sinks to sources and increases the likelihood of fire hazards. Rewetting deforested and drained PSFs before revegetation is expected to reverse this outcome. This study aims to quantify the GHG emissions of deforested PSFs that have been (a) reforested, (b) converted into oil palm, or (c) replanted with rubber. Before rewetting, heterotrophic soil respiration in reforested, oil palm, and rubber plantation areas were 48.91 ± 4.75 Mg CO2 ha−1 yr−1, 54.98 ± 1.53 Mg CO2 ha−1 yr−1, and 67.67 ± 2.13 Mg CO2 ha−1 yr−1, respectively. After rewetting, this decreased substantially by 21%, 36%, and 39%. Conversely, rewetting drained landscapes that used to be methane (CH4) sinks converted them into CH4 sources; almost twice as much methane was emitted after rewetting. Nitrous oxide (N2O) emissions tended to decrease; in nitrogen-rich rubber plantations, N2O emissions halved; in nitrogen-poor reforested areas, emissions reduced by up to a quarter after rewetting. Overall, rewetting reduced the net emissions up to 15.41 Mg CO2-eq ha−1 yr−1 (25%) in reforested, 18.36 Mg CO2-eq ha−1 yr−1 (18%) in oil palm, and 28.87 Mg CO2-eq ha−1 yr−1 (17%) in rubber plantation areas.

Litter Decomposition in Wet Rubber and Fruit Agroforests: Below the Threshold for Tropical Peat Formation

Peatlands are shaped by slow litter decomposition, but threshold decomposition rates that allow peat formation remain unclear. Can agroforestry in the tropics be compatible with paludiculture that allows peat formation? We explored the determinants of litter decomposition in wet agroforests adjacent to tropical peatlands in Central Kalimantan (Indonesia) by litterbag studies (up to 16 weeks) with standing litter sources to estimate rate constants, characterize litter quality (especially lignin (L), polyphenolics (Pp) and nitrogen (N)), and monitor temperature and groundwater levels. In litter transfer experiments we tested for home-field advantage (HFA) effects between land cover types. Mean residence times around 85 weeks at 27 °C were associated with a high (L + Pp)/N ratio. However, in the crossover treatments, mean residence times varied from 30 to 180 weeks and strong HFA effects (up to 80% faster decomposition at “home”) were found when litter from other sources was tested in old fruit-based agroforests. HFA indicates a local decomposer community well-adapted to its normal litter diet. Litter residence times of around two years are below the apparent peat formation threshold. Maintaining wet agroforest conditions adjacent to peat domes supports peatland rewetting and restoration but does not contribute to on-site peat formation processes.

Assessing costs of Indonesian fires and the benefits of restoring peatland

Deforestation and drainage has made Indonesian peatlands susceptible to burning. Large fires occur regularly, destroying agricultural crops and forest, emitting large amounts of CO₂ and air pollutants, resulting in adverse health effects. In order to reduce fire, the Indonesian government has committed to restore 2.49 Mha of degraded peatland, with an estimated cost of US$3.2-7 billion. Here we combine fire emissions and land cover data to estimate the 2015 fires, the largest in recent years, resulted in economic losses totalling US$28 billion, whilst the six largest fire events between 2004 and 2015 caused a total of US$93.9 billion in economic losses. We estimate that if restoration had already been completed, the area burned in 2015 would have been reduced by 6%, reducing CO₂ emissions by 18%, and PM_2.5 emissions by 24%, preventing 12,000 premature mortalities. Peatland restoration could have resulted in economic savings of US$8.4 billion for 2004-2015, making it a cost-effective strategy for reducing the impacts of peatland fires to the environment, climate and human health.

Landslide and Wildfire Susceptibility Assessment in Southeast Asia Using Ensemble Machine Learning Methods

Southeast Asia (SEA) is a region affected by landslide and wildfire; however, few studies on susceptibility modeling for the two hazards together have been conducted for this region, and the intersection and the uncertainty of the two hazards are rarely assessed. Thus, the intersection of landslide and wildfire susceptibility and the spatial uncertainty of the susceptibility maps were studied in this paper. Reliable landslide and wildfire susceptibility maps are necessary for disaster management and land use planning. This work used three advanced ensemble machine learning algorithms: RF (Random Forest), GBDT (Gradient Boosting Decision Tree) and AdaBoost (Adaptive Boosting) to assess the landslide and wildfire susceptibility for SEA. A geo-database was established with 2759 landslide locations, 1633 wildfire locations and 18 predictor variables in total. The performances of the models were assessed using the overall classification accuracy (ACC), Precision, the area under the ROC (receiver operating curve) (AUC) and confusion matrix values. The results showed RF performs superior in both landslide (ACC = 0.81, Precision = 0.78 and AUC= 0.89) and wildfire (ACC= 0.83, Precision = 0.83 and AUC = 0.91) susceptibility modeling, followed by GBDT and AdaBoost. The overall superiority of RF over other models indicates that it is potentially an efficient model for landslide and wildfire susceptibility mapping. The landslide and wildfire susceptibility were obtained using the RF model. This paper also conducted an overlay analysis of the two hazards. The uncertainty of the susceptibility was further assessed using the coefficient of variation (CV). Additionally, the distance to roads is relatively important in both landslide and wildfire susceptibility, which is the most important in landslides and the second most important in wildfires. The result of this paper is useful for mastering the whole situation of hazard susceptibility and proves that RF is a robust model in the hazard susceptibility assessment in SEA.