Wildfires are expected to increase in the near future, mainly because of climate changes and land use management. One of the most vulnerable areas in the world is the forest in central-South America, including Bolivia. Despite that this country is highly prone to wildfires, literature is rather limited here. To fill this gap, we implemented a dataset including the burned area that occurred in the department of Santa Cruz in the period of 2010–2019, and the digital spatial data describing the predisposing factors (i.e., topography, land cover, ecoregions). The main goal was to develop a model, based on Random Forest, in which probabilistic outputs allowed to elaborate wildfires susceptibility maps. The overall accuracy was finally estimated by using 5-fold cross-validation. In addition, the last three years of observations acted as the testing dataset, allowing to evaluate the predictive performance of the model. The quantitative assessment of the variables revealed that “flooded savanna” and “shrub or herbaceous cover, flooded, fresh/saline/brakish water” are respectively the ecoregions and land cover classes with the highest probability of predicting wildfires. This study contributes to the development and validation of an innovative mapping tool for fire risk assessment, implementable at a regional scale in different areas of the globe.

<p>Understanding the relationships between different drought drivers and observed drought impact can provide important information for early warning systems and drought management planning. Moreover, this relationship can help inform the definition and delineation of drought events. However, currently, drought hazards are often characterized based on their frequency of occurring, rather than based on the impacts they cause. A more data-driven depiction of “impactful drought events”- whereby droughts are defined by the hydrometeorological conditions that, in the past, have led to observable impacts-, has the potential to be more meaningful for drought risk assessments.</p> <p>In our research, we apply a data-mining method based on association rules, namely fast and frugal decision trees, to link different drought hazard indices to agricultural impacts. This machine learning technique is able to select the most relevant drought hazard drivers (among both hydrological and meteorological indices) and their thresholds associated with “impactful drought events”. The technique can be used to assess the likelihood of occurrence of several impact severities, hence it supports the creation of a loss exceedance curve and estimates of average annual loss. An additional advantage is that such data-driven relations in essence reflect varying local drought vulnerabilities which are difficult to quantify in data-scarce regions.</p> <p>This contribution exemplifies the use of fast and frugal decision trees to estimate (agricultural) drought risk in the Volta basin and its riparian countries. We find that some agriculture-dependent regions in Ghana, Togo and Côte d’Ivoire face annual average drought-induced maize production losses up to 3M USD, while per hectare, losses can mount to on average 50 USD/ha per year in Burkina Faso. In general, there is a clear north-south gradient in the drought risk, which we find augmented under projected climate conditions. Climate change is estimated to worsen the drought impacts in the Volta Basin, with 11 regions facing increases in annual average losses of more than 50%.</p> <p>We show that the proposed multi-variate, impact-based, non-parametric, machine learning approach can improve the evaluation of droughts, as this approach directly leverages observed drought impact information to demarcate impactful drought events. We evidence that the proposed technique can support quantitative drought risk assessments which can be used for geographic comparison of disaster losses at a sub-national scale.</p>

<p>The central South American forest is one of the area most affected by wildfires in the world. Because of climate changes and land use management, these events are becoming more frequent and extended in the last years. For example, in 2019 Bolivia faced an extremely extensive wildfire event that had a serious ecological impact in the department of Santa Cruz. This region, called Chiquitania and characterized by a mosaic where wet tropical forests, dry tropical forests and savannas alternate, accounts for more than two-thirds of the total wildfires in the country. Despite Bolivia is between the top-ten countries with the highest expected risk in terms of annual burned forest area, the literature on wildfires here is quite limited, also because of the scarcity of available data and resources. To fill this gap, as part of the present study, we implemented an accurate dataset of burned areas, based on MODIS wildfire product, occurred in the entire Santa Cruz region in the period 2010-2019. Predisposing factors, such as topography, land use and ecoregions, were also collected in the form of digital spatial data. This information allowed assessing the susceptibility to wildfires on the entire region, with a special focus on the municipality of San Ignacio de Velasco. The analysis was performed using Random Forest (RF), an ensemble-learning algorithm based on decision trees, capable of learning from and make predictions on data by modeling the hidden relationships between a set of input and output variables. The goodness of fit was estimated by the area under the ROC (receiver operating characteristic) curve (AUC), selecting the validation dataset by using a 5-folds cross validation procedure. In addition, the last three years of observed burned areas were kept out during the medialization stage and used to test if the implemented model gives good predictions on new data. As result, we obtained a probabilistic output from RF indicating the probability for an area to burn in the future, which allowed elaborating the susceptibility maps. For San Ignacio de Velasco it resulted an AUC of 0.8, while for the entire Santa Cruz the AUC was of 0.73. Likewise, the predictive capabilities of the model gave quite good results, better at municipality that at regional level. The detailed investigation of the relative importance of each categorical class belonging to the variables ecoregions and land use reveals that “Flooded savanna” and “Shrub or herbaceous cover, flooded, fresh/saline/brakish water” are respectively the classes most related with wildfires. This important outcome confirms recent findings, that seasonally wet and dry climate, coupled with hydrologic controls on the vegetation, create in this ecoregion favorable conditions to the ignition and spreading of large wildfires during the driest period, when the biomass is abundant. The occurrence of large fires, initiated by slash-and-burn practice getting out of control, is predicted to increase in the near future and the development of new tools for fire risk assessment and reduction is thus needed. </p>

<p>Within the framework of the CIF financed “Pilot Program for Climate Resilience”, the Drought Monitoring and Early Warning System for Bolivia was developed and implemented. The system is operational since July 2020 and aims at detecting emerging severe drought conditions in the country, in order to trigger timely warnings to stakeholders and the general public.</p><p>The Bolivian Drought Monitor has two main components: a technical one (data gathering and analysis, performed through the multi-hazard early warning “myDEWETRA” platform) and an institutional one (creating consensus and disseminating warnings). The system design followed a participatory approach, involving since the early stages the Ministry for Water and Environment (MMAyA), the National Hydrometeorological Service (SENAMHI), the Vice-Ministry for Civil Defence (VIDECI). These institutions actively contribute to the monthly edition of the drought bulletin, each one for its own sector of competence, through a dedicated IT tool for synchronous compilation. Ongoing drought conditions are reported in a national bulletin, issued monthly and published on a dedicated public website: http://monitorsequias.senamhi.gob.bo/</p><p>Given the Bolivian data-poor context, analysis strongly relies on a large variety of multi-source satellite products, spanning from well consolidated ones in the operational practice to more experimental ones such as from the SMAP mission. This information is used to monthly refresh the spatial maps of 17 indexes covering meteorological, hydrological and agricultural droughts for different aggregation periods (from 1 to 12 months). Simulation of the system performance over a long period (2002-2019) and comparison with recorded socio-economic drought impacts  from the National Disaster Observatory (Observatorio Nacional de Desastres- OND) of the Vice-Ministry of Civil Defence (VIDECI) was used to define a most representative compound index, based on a weighted combination of a selection of 4 indexes with their related thresholds. The combination of 3-month SPEI, 2-month SWDI, 1-month VHI and 1-month FAPAR indexes performed the best in the comparison with impact records. This combination encompasses both the medium-term effects of meteorological and hydrological deficits (3-month SPEI and SWDI), both the short-term effects on vegetation (1-month VHI and FAPAR). This set of indexes proved to be a solid proxy in estimating possible impacts on population of ongoing or incoming drought spells, as happened for most significant recent drought events occurred in Bolivia, such as the 2010 event in the Chaco region and the 2016 drought event in the Altiplano and Valles regions, that heavily affected the water supply in several major cities (La Paz, Sucre, Cochabamba, Oruro and Potosí).</p><p>The design of the monitoring and bulletin management platform, together with its strong remote-sensing base, give to the system a high potential for easy export to other regional and national contexts. Also, the variety of the different computed drought indexes and the replicability of the procedure for the best compound index identification will allow for efficient evolutionary maintenance as new remote-sensing products will be available in the future.</p>