Search

Forest biomass is a renewable energy source, more climate-friendly than fossil fuels and widely available in Europe. The wood energy chain has been suggested as a means to re-activate forest management and improve the value of forest stands in marginalized rural areas. However, wall-to-wall estimates of forest biomass, needed to design the location and size of power and heat biomass plants in any given territory, are notoriously difficult to obtain. This paper tests an algorithm to predict forest biomass using publicly available Landsat satellite imagery in the Liguria region, northern Italy. We used regional forest inventory data to train and validate an artificial neural network (ANN) classifier that uses remotely-sensed information such as three principal components of Landsat-5 TM spectral bands, the Enhanced Vegetation Index (EVI), and topography, to retrieve aboveground live tree volume. Percent root mean square error was -9% and -23% for conifers and broadleaves respectively in the calibration dataset, and -27% and -24% in the validation dataset. The reconstructed volume map was updated to present day using current volume increment rates reported by the Italian National Forest Inventory. A wall-to-wall map of forest biomass from harvest residues was finally produced based on species-specific wood density, biomass expansion factors, volume logged for timber assortments, forest accessibility, and topography. Predicted aboveground forest volume ranged from 81 to 391 m³ ha-1. In forests available for wood supply (70% of the total), planned volume removals averaged 25.4 m³ ha-1, or 18.7% of the average standing stock across. Biomass available for bioenergy supply was 1.295.921 million Mg dry matter or 8.95 Mg ha-1. This analysis workflow can be replicated in all mountain regions with a predominant broadleaved coppice component. [ABSTRACT FROM AUTHOR]

Forests provide direct protection to human settlements from hydrogeomorphic hazards. This paper proposes a method for assessing the effect of natural disturbances on the functionality of direct protection forests (DPFs) in order to prioritize management interventions. We georeferenced disturbance data for wildfires, wind and snow damage, avalanches, and insects and overlaid them to a region-wide DPF map. Within each disturbance polygon, we used a Landsat-5 TM image to identify DPFs with insufficient vegetation cover, by using a maximum likelihood classifier of 6 spectral bands plus 5 vegetation indices. For each disturbance agent, we fitted a generalized linear model of the probability of finding a forested pixel, as a function of topography, time since disturbance, distance from disturbance edge, summer precipitation, and drought in the disturbance year. DPFs covered almost half of total forest area in the study region. Disturbance by insects occurred in more than one sixth of all forests. Avalanche and wildfire occurred each in about one tenth of total forest area, and wind and snow disturbance in only 1%. In the last 50 years, disturbances had a recurrence rate of 3% every 10 years. Almost one sixth of DPFs are currently lacking sufficient forest cover. Wildfires resulted in the highest rate of nonforested pixels (42% of all DPFs), followed by avalanches (21%). Forest recovery was explained by time elapsed, distance from edge (for conifers), and aspect. Summer precipitation and drought had a mixed influence. Our approach to assessing the effect of disturbances on the functionality of DPFs is reproducible in all mountain regions using institutional or open-access geographic data and provides a tool to prioritize DPF management by indicating where restoration of protection is most urgent. [ABSTRACT FROM AUTHOR]