A data-driven prediction model for Fennoscandian wildfires

Wildfires are recurrent natural hazards that affect terrestrial ecosystems, the carbon cycle, climate and society. They are typically hard to predict, as their exact location and occurrence are driven by a variety of factors. Identifying a selection of dominant controls can ultimately improve predictions and projections of wildfires in both the current and a future climate. In this study, we applied a data-driven machine learning approach to identify dominant hydrometeorological factors determining fire occurrence over Fennoscandia, and produced spatiotemporally resolved fire danger probability maps. A random forest learner was applied to predict fire danger probabilities over space and time, using a monthly 2001–2019 satellite-based fire occurrence dataset at a 0.25° spatial grid as the target variable. The final data-driven model slightly outperformed the established Canadian fire weather index (FWI) used for comparison. Half of the 30 potential predictors included in the study were automatically selected for the model. Shallow volumetric soil water anomaly stood out as the dominant predictor, followed by predictors related to temperature and deep volumetric soil water. Using a local fire occurrence record for Norway as target data in a separate analysis, the test set performance increased considerably. This improvement shows the potential of developing reliable data-driven prediction models for regions with a high quality fire occurrence record, and the limitation of using satellite-based fire occurrence data in regions subject to small fires not picked up by satellites. We conclude that data-driven fire prediction models are promising, both as a tool to identify the dominant predictors and for fire danger probability mapping. The derived relationships between wildfires and its compound predictors can further be used to assess potential changes in fire danger probability under future climate scenarios.

Development of Google Earth Engine Fire Weather Index Calculator for Indonesian Fire Danger Rating System

Tropical peatlands make up almost ten percent of the land surface in Indonesia, making peat fires detrimental not only for global atmospheric carbon levels, but also to public health and socioeconomic activities in the region. Indonesian Fire Danger Rating System (FDRS) was developed based on the Canadian Forest Fire Weather Index System (CFFWIS), using three different fuel codes and three indices representing fire behaviour. Daily Fire Weather Index (FWI) calculation is done by the Meteorological Climatological and Geophysical Agency (BMKG) with data from its synoptic weather stations network. Distribution of such weather stations are sparse, therefore this paper reports on the development of Fire Weather Index calculator on Google Earth Engine, using high resolution weather data, provided by weather model and remote-sensing open datasets. The resulting application is capable of generating daily maps of FWI components to be used by the Indonesian Fire Danger Rating System.

PEAT SURFACE COMPRESSION REDUCES SMOULDERING FIRE POTENTIAL AS A NOVEL FUEL TREATMENT FOR BOREAL PEATLANDS

The wildfire regime in Canada’s boreal region is changing; extended fire seasons are characterized by more frequent large fires (≥200 ha) burning greater areas of land, whilst climate-mediated drying is increasing the vulnerability of peatlands to deep burning. Proactive management strategies, such as fuel modification treatments, are necessary to reduce fire danger at the wildland-human interface (WHI). Novel approaches to fuel management are especially needed in peatlands where deep smouldering combustion is a challenge to suppression efforts and releases harmful emissions. Here, we integrate surface compression within conventional stand treatments to examine the potential for reducing smouldering of near-surface moss and peat. A linear model (adj. R2=0.62, p=2.2e-16) revealed that ground cover (F(2,101)=60.97, p<0.001) and compression (F(1,101)=56.46, p<0.001) had the greatest effects on smouldering potential, while stand treatment did not have a significant effect (F(3,101)=0.44, p=0.727). On average, compressed Sphagnum and feather moss plots showed 57.1% and 58.7% lower smouldering potential, respectively, when compared to uncompressed analogs. While practical evaluation is warranted to better understand the evolving effectiveness of this strategy, these findings demonstrate that a compression treatment can be successfully incorporated within both managed and unmanaged peatlands to reduce fire danger at the WHI.

Developing fire danger models using logistic regression analysis for mid-hills of Himachal Pradesh

The present study intended to develop a climatic fire danger model for mid-hills zone of Himachal Pradesh using ten years weather data in relation with forest fire occurrence (2007-2016). Logistic regression technique was used to determine the relationship between fire occurrence and weather parameters viz., maximum temperature (°C), relative humidity (%), and wind speed (ms-1). The model was validated by calculating area under curve (AUC), coefficient of determination (R2) and root mean square Error (RMSE), with estimated values of 88.90%, 0.705 and 0.247, respectively. The fire danger model was verified with actual fire incidences in the study area during the year 2017. Wald's test was carried out to quantify impact climatic parameters on forest fire. Wald's test value was highest for maximum temperature (40.07) followed by relative humidity (1.15) and wind speed (0.75), respectively. In future such model can be utilized for prevention of forest fire hazards in the study area.

Analysis of drought index to asses land and forest fire season in Maros Regency

Forest and land fires occur almost every year, so they are a concern and priority in their control efforts. One of the important factors in the effort to control forest and land fires is knowing the times and locations that are prone to forest and land fires. This study aims to describe the level of drought and areas prone to forest and land fires in Maros Regency, South Sulawesi. This research was conducted in January to August 2017. The data collection was obtained from the Meteorology, Climatology and Geophysics Agency, the Climate Change Control Center, and the Bantimurung Bulusaraung National Park. The data were analyzed using the Polygon Thiessen method, the Keecth Byram Drought Index (KBDI) method, and the spatial analysis method. The results showed that extreme drought conditions in Maros Regency occurred from September to October based on observations of maximum rainfall and temperature. The forest area classified as moderate forest fire danger rating dominates the Maros Regency area, namely 73418.67 ha (45.77%). Maros Regency which is included in the area with a very high forest fire danger rating is Tompobulu and Cenrana Districts.

Efficiency evaluation of the weighted mean calculation of the forest fire hazard class

Nowadays, cost-optimization of aerial patrolling plays a key role in the context of limited aerial forest protection funding. Forest Fire Danger Class is the main indicator that regulates the work of forest fire services. Usually, it’s calculated by the nearest weather station data. Some information systems use the mean of several nearby weather stations to estimate large areas, such as the surveyed area of aerial forest protection. The idea of using the mean weighted index with the weather stations weighting factor is not new. Even though, this idea isn’t widespread due to the calculation complexity and questionable efficiency in practice, this study proposes a scientifically substantiated method of quantitative comparison of two approaches and the direct calculation method of the economic impact when transition to using the mean weighted Forest Fire Danger Class calculation algorithm. The first time such an indicator was used to obtain derivatives of analytical information products. A long-term analysis of forest fire rate showed that the weighted mean of the Forest Fire Danger Class value is 6.7% greater in correlation with the number of forest fires than the usual mean value. The use logarithmic transformation of the forest fire occurrence frequency and population density allows statistical criteria to be reasonably used.