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.