Managing forests has been demonstrated to be an efficient strategy for fragmenting fuels and for reducing fire spread rates and severity. However, large-scale analyses to examine operational aspects ...of implementing different forest management scenarios to meet fire governance objectives are nonexistent for many Mediterranean countries. In this study we described an optimization framework to build forest management scenarios that leverages fire simulation, forest management, and tradeoff analyses for forest areas in Macedonia, Greece. We demonstrated the framework to evaluate five forest management priorities aimed at (1) protection of developed areas, (2) optimized commercial timber harvests, (3) protection of ecosystem services, (4) fire resilience, and (5) reducing suppression difficulty. Results revealed that by managing approximately 33,000 ha across all lands in different allocations of 100 projects, the area that accounted for 16% of the wildfire exposure to developed areas was treated while harvesting 2.5% of total wood volume. The treatments also reduced fuels on the area that are responsible for 3% of the potential fire impacts to sites with important ecosystem services, while suppression difficulty and wildfire transmission to protected areas attainment was 4.5% and 16%, respectively. We also tested the performance of multiple forest district management priorities when applying a proposed four-year fuel treatment plan that targeted achieving high levels of attainment by treating less area but strategically selected lands. Sharp management tradeoffs were observed among all management priorities, especially for harvest production compared with suppression difficulty, the protection of developed areas, and wildfire exposure to protected areas.
Predicting where the next large-scale wildfire event will occur can help fire management agencies better prepare for taking preventive actions and improving suppression efficiency. Wildfire ...simulations can be useful in estimating the spread and behavior of potential future fires by several available algorithms. The uncertainty of ignition location and weather data influencing fire propagation requires a stochastic approach integrated with fire simulations. In addition, scarcity of required spatial data in different fire-prone European regions limits the creation of fire simulation outputs. In this study we provide a framework for processing and creating spatial layers and descriptive data from open-access international and national databases for use in Monte Carlo fire simulations with the Minimum Travel Time fire spread algorithm, targeted to assess cross-boundary wildfire propagation and community exposure for a large-scale case study area (Macedonia, Greece). We simulated over 300,000 fires, each independently modelled with constant weather conditions from a randomly chosen simulation scenario derived from historical weather data. Simulations generated fire perimeters and raster estimates of annual burn probability and conditional flame length. Results were used to estimate community exposure by intersecting simulated fire perimeters with community polygons. We found potential ignitions can grow large enough to reach communities across 27% of the study area and identified the top-50 most exposed communities and the sources of their exposure. The proposed framework can guide efforts in European regions to prioritize fuel management activities in order to reduce wildfire risk.
In this work, we provide a framework for assessing cross-boundary wildfire exposure and a case study application in the region of Macedonia, Greece. The required spatial layers describing topography, ...fuels/ vegetation and ignition location were retrieved from open-access international and national databases, while climate data were obtained from remote automatic weather stations. We processed the spatial layers to derive the required inputs for the Minimum Travel Time fire spread algorithm. Hourly and daily weather data were used to derive the most frequent wind directions and to characterize the extreme wind speed scenarios for the season with the lowest dead and live fuel moisture contents (July and August). The study region was divided into 30 zones of similar climatic conditions and historical wildfire activity (i.e., simulation scenarios that included simulation duration, wind data, fuel moisture content and spotting probability). For model calibration, we replicated the historical large wildfire size (>50 ha) distribution of each zone by simulating thousands of potential wildfires with the derived simulation scenarios. Ignitions were allocated within burnable fuels according to an ignition probability grid developed from a historical fire occurrence database. We simulated over 300,000 fires, each independently modelled with constant weather conditions considering a randomly chosen simulation scenario. Scenario selection was based on a predefined selection probability derived by the historical wind direction frequency and fire duration of the ignition location zone. Simulations generated a layer of fire perimeters and raster estimates of annual burn probabilities and conditional flame length. Results were used to estimate community exposure by intersecting simulated fire perimeters with community polygons. The number of exposed structures was assigned to each simulated fire ignition, estimating its influence on each community (one ignition to many communities). The post-processing of these ignitions generated community firesheds, which delineate the area around communities where large fires are likely to be transmitted to the burnable and populated community area polygons. We found that on 9,250 km2, or the 27% of the study area, potential ignitions can grow large enough to reach communities and cause structure exposure. The proposed framework can guide future efforts aimed at quantifying community exposure to large-scale wildfires and guide investments to prioritize fuel management activities to reduce fire risk.
Accurate measurements in forest inventories are crucial. Though manual measurements are precise, they are time consuming, while most of the times, the amount of collected data is limited in Diameter ...at Breast Height (DBH) and Height. Contemporary technological achievements such as Terrestrial Laser Scanner (TLS) and Unmanned Air Vehicle (UAV) can be used in forest inventories, where capturing detailed and highly accurate information, such as geometry, texture or color, can be obtained in great detail and speed. This study is focusing on semi-automatic tree and tree crown parameters extraction from Point Clouds (PCs) acquired by TLS and UAV, on a plot level. PC derived from TLS procedure and from combination of TLS and UAV were further studied. Tree and crown parameters were computed for both PC datasets, which are tree position, DBH, Stem Curve, tree planar projection, Crown position, Crown bottom height, Crown total height, Crown volume and surface area and Crown intersections. Tree height and DBH parameters of both datasets were further compared with manually derived field data measurements. TLS PC gave satisfying results regarding DBH (Pearson's r: 0.971, R2: 0.9427) while at Height there was a significant loss of information due to obstruction by lower branches (Pearson's r: 0.524, R2: 0.2747). TLS - UAV combined PC has better but not adequate results at Height (Pearson's r: 0.581, R2: 0.3380) where loss of information was also occurred due UAV camera sensor insufficiency in capturing the upper stem.
PDF
