Estimating VAIA Windstorm Damaged Forest Area in Italy Using Time Series Sentinel-2 Imagery and Continuous Change Detection Algorithms

Modeling windthrow effects on water runoff and hillslope stability in a mountain catchment affected by the VAIA storm

Windthrows seriously affect forest landscapes, causing several issues in hydrological and geomorphological terms. In this regard, Airborne Laser Scanning (ALS) topographic data recently increased the opportunity to investigate in detail physical processes at the catchment scale. Moreover, topographically based hydrological and geomorphological models allow quantifying runoff alteration due to windthrows-driven land cover changes and detect the occurrence of land degradative processes at the sub-catchment scale. In this connection, accurate investigations about windthrows role in varying local runoff regimes over time are still obscure, as well as the possibility of predicting terrain instabilities due to windstorm occurrence. This research aims to investigate the interaction between windthrows, runoff alterations and hillslope failures affecting a landslide-prone mountain catchment (northern Italy). Hydrological HEC-HMS and geomorphological RESS models were applied. Windthrows' role in altering runoff regimes and hillslope stability was investigated starting from the elaboration of ALS-derived points clouds acquired before and after the occurrence of the Vaia storm. Digital Terrain Models (DTMs) were elaborated for the two scenarios to compare daily runoff variations and predict the activation of terrain instabilities by looking at land cover changes driven by the blowdown event at the sub-catchment detail. Results attested the key role of windstorms in altering local runoff, with a maximum relative runoff increment equal to 2.56 % and a maximum runoff difference equal to 3.12 mmh-1, as well as in encouraging the activation of the observed shallow landslide. The correlation between windthrows occurrence and runoff alterations was validated by performing regression analysis (R2 = 0.76), while the accuracy of instabilities predictions was tested through the Distance to Perfect Classification (D2PC) index and True Skill Statistic (TSS) score, respectively resulted equal to 0.076 and 0.898. This research represents a valid tool for investigating similar issues at a wider scale, also providing suggestions for promoting interventions in wind-disturbed forest areas.

Machine Learning Methods to Estimate Productivity of Harvesters: Mechanized Timber Harvesting in Brazil

The correct capture of forest operations information carried out in forest plantations can help in the management of mechanized harvesting timber. Proper management must be able to dimension resources and tools necessary for the fulfillment of operations and helping in strategic, tactical, and operational planning. In order to facilitate the decision making of forest managers, this work aimed to analyze the performance of machine learning algorithms in estimating the productivity of timber harvesters. As predictors of productivity, we used the availability of hours of machine use, individual mean volumes of trees, and terrain slopes. The dataset was composed of 144,973 records, carried out over a period of 28 months. We tested the predictive performance of 24 machine learning algorithms in default mode. In addition, we tested the performance of blending and stacking joint learning methods. We evaluated the model’s fit using the root mean squared error, mean absolute error, mean absolute percentage error, and determination coefficient. After cleaning the initial database, we used only 1.12% to build the model. Learning by blending ensemble stood out with a determination coefficient of 0.71 and a mean absolute percentage error of 15%. From the use of data from machine learning algorithms, it became possible to predict the productivity of timber harvesters. Testing a variety of machine learning algorithms with different dynamics contributed to the machine learning technique that helped us reach our goal: maximizing the model’s performance by conducting experimentation.

Distinguishing Abrupt and Gradual Forest Disturbances With MODIS-Based Phenological Anomaly Series

Capturing forest disturbances over time is increasingly important to determine the ecosystem's capacity to recover as well as aiding a timely response of foresters. With changes due to climate change increasing the frequencies, a better understanding of forest disturbances and their role in historical development is needed to, on the one hand, develop forest management approaches promoting ecosystem resilience and, on the other hand, provide quick and spatially explicit information to foresters. A large, publicly available satellite imagery spanning more than two decades for large areas of the Earth's surface at varying spatial and temporal resolutions represents a vast, free data source for this. The challenge is 2-fold: (1) obtaining reliable information on forest condition and development from satellite data requires not only quantification of forest loss but rather a differentiated assessment of the extent and severity of forest degradation; (2) standardized and efficient processing routines both are needed to bridge the gap between remote-sensing signals and conventional forest condition parameters to enable forest managers for the operational use of the data. Here, we investigated abiotic and biotic disturbances based on a set of ground validated occurrences in various forest areas across Germany to build disturbance response chronologies and examine event-specific patterns. The proposed workflow is based on the R-package "npphen" for non-parametric vegetation phenology reconstruction and anomaly detection using MODIS EVI time series data. Results show the potential to detect distinct disturbance responses in forest ecosystems and reveal event-specific characteristics. Difficulties still exist for the determination of, e.g., scattered wind throw, due to its subpixel resolution, especially in highly fragmented landscapes and small forest patches. However, the demonstrated method shows potential for operational use as a semi-automatic system to augment terrestrial monitoring in the forestry sector. Combining the more robust EVI and the assessment of the phenological series at a pixel-by-pixel level allows for a changing species cover without false classification as forest loss.

Integrating GEDI and Landsat: Spaceborne Lidar and Four Decades of Optical Imagery for the Analysis of Forest Disturbances and Biomass Changes in Italy

Forests play a prominent role in the battle against climate change, as they absorb a relevant part of human carbon emissions. However, precisely because of climate change, forest disturbances are expected to increase and alter forests' capacity to absorb carbon. In this context, forest monitoring using all available sources of information is crucial. We combined optical (Landsat) and photonic (GEDI) data to monitor four decades (1985-2019) of disturbances in Italian forests (11 Mha). Landsat data were confirmed as a relevant source of information for forest disturbance mapping, as forest harvestings in Tuscany were predicted with omission errors estimated between 29% (in 2012) and 65% (in 2001). GEDI was assessed using Airborne Laser Scanning (ALS) data available for about 6 Mha of Italian forests. A good correlation (r² = 0.75) between Above Ground Biomass Density GEDI estimates (AGBD) and canopy height ALS estimates was reported. GEDI data provided complementary information to Landsat. The Landsat mission is capable of mapping disturbances, but not retrieving the three-dimensional structure of forests, while our results indicate that GEDI is capable of capturing forest biomass changes due to disturbances. GEDI acquires useful information not only for biomass trend quantification in disturbance regimes but also for forest disturbance discrimination and characterization, which is crucial to further understanding the effect of climate change on forest ecosystems.