https://greenmarked.it/wp-content/uploads/2026/03/Copia-di-Parabolic-recorder-scaled.jpg
1920
2560
Jennifer Lüdtke
https://greenmarked.it/wp-content/uploads/2022/01/LOGO-GREENMARKED-SITO-600x600.png
Jennifer Lüdtke2026-03-08 14:56:142026-04-07 23:17:05Behind the Lens and Beyond the Microphone: Studying Wildlife with AIThere is a statistical algorithm—well known to anyone with a bit of machine learning experience—called “Random Forest”.
As a forester by trade, not well versed in computer terminology, the first time I encountered this term I immediately thought it referred to some form of woodland structure classification and certainly not to a complex statistical data‐training mechanism.
Yet, in my mathematical ignorance, I wasn’t that far off. Because this statistical mechanism (may my computer-science colleagues forgive the oversimplification) does nothing more than classify a set of data in order to assign labels to it: it takes a collection of scattered and heterogeneous pieces of information and divides them into smaller groups, organized according to a logical scheme.
And how does it do it? In simple terms, it takes the original dataset and creates a multitude of random decision trees from which it derives the final decision. Visually, the mechanism appears as a series of random branches unfolding one after another—hence the term “Random Forest”.
For the forestry sector, as in many other technical fields, data collection, processing, and management are now of fundamental importance. And “Random Forest”, in forests as in other areas, is just one of the many ways to do what technology demands most: classify, organize, and deliver concrete answers.
Imagine an incredibly detailed, color aerial photograph in which you can distinguish forests, meadows, roads, lone trees, houses and industrial complexes, lakes, streams, and rocks. The technician’s job is to classify this photograph and turn it into a map—detailed yet easy to use—that represents the terrain. They can do this by hand, drawing what they see in the photo and investigating in the field what isn’t visible, or they can go digital and… let the computer handle it.
Once this map has been created, they can print it and turn it into a nice informational sign—the kind you post along trails that reads “YOU ARE HERE”—or they can make it available online, with geolocation features and real-time notifications of any trail closures.
Not only that, since the map also shows all the trees in the city—clearly visible in the photo—it can answer the question posed by the municipality regarding the number of trees the administration needs to manage. However, the funds allocated for managing these trees vary based on their height and size; therefore, a complete field census would still have to be carried out… or maybe not?
And what about the roads? Can the forestry company haul the timber by truck along the forest road that traverses the mountain, or is the gradient of the route too steep, or worse yet: is there even a road crossing that mountain? Which trees should be harvested, then—beeches, firs, chestnuts? And how much will the company earn financially from the timber sale? Wouldn’t it be better to log elsewhere, in a more easily accessible location?
And finally, there’s that new industrial shed under construction… on the only buildable strip of land between the protected park area and the stream. But are we sure that this structure, in that location, can be erected without obstructing amphibian reproduction?
These are just some of the questions a forestry technician is professionally called upon to answer. Clear, direct questions with concrete answers. Answers that the advent of digitization in this technical field is making far more reliable, rapid, and based on measurable, shareable evidence.
The management of the entire territory—from nature conservation to urbanization, from forestry operations to playground management—cannot disregard a detailed knowledge of geographical elements. In this case, geography is to land management what numbers are to mathematics: its foundational elements.
The digitalization of the forestry sector starts right here: by dematerializing, sharing, and managing large volumes of data (primarily geographic) with the aim of using them in concrete applications.

Figure 1: Drone taking off for a LiDAR survey of a forest parcel. Latemar Forestry School, Carezza/Karerpass, Nova Levante/Welschnofen – Bolzano/Bozen. Photo: Author. 09.06.2022.
Some examples?
It serves as the strategic instrument for understanding the status of Italian woodlands and defining effective policies on forestry, the environment, climate-change adaptation and sustainable development. It also enables monitoring the implementation of the National Forestry Strategy and supports the programming, planning and management of forests and the forestry sector [1].
SINFor is a complex, detailed tool that is continuously updated and lays the groundwork for a nationwide digitization of the entire forestry sector. Just consider that, thanks to this system, a Forest Map of Italy is finally available—comprehensive, coordinated, and up-to-date (the last comparable document was the Forest Militia Map, dated… 1936).
Decision Support Systems (DSS)
Decision Support Systems (DSS) – once also called Expert Systems – serve as platforms, often online, designed to support management, planning, and decision-making in a complex field such as forestry and environmental management.
As a result, through these tools, professionals—but also private citizens or public administrations—can view and extract data in (almost) real time, thereby making conscious and informed decisions. As in the example above of the forestry company’s request for timber extraction feasibility: in the past, processing such information would have required days of work with extensive field surveys; today, it can be obtained via a single, well-designed and up-to-date DSS.
And we’re not just talking about theory or experimentation; some DSS are already active and functional:
- Pri.For.Man DSS: The Decision Support System for Forestry of Friuli Venezia Giulia [2]
- GO-SURF: The Decision Support System for Forestry of Tuscany [3]
HeProMo is the wood‐harvesting productivity model developed by the Swiss Federal Institute for Forest, Snow and Landscape Research (WSL).
With HeProMo, it is possible to easily and quickly calculate the time requirements, performance, and costs of the various wood‐harvesting operations [4].
This tool enables the generation of a canopy height map (CHM) of trees from medium- or high-resolution aerial or satellite imagery [5].
Although difficult to implement in professional settings due to its complex usability (which relies on deep-learning algorithms), the results are often astounding, with accuracy comparable even to LiDAR surveys over large areas.
References:
[1] Direzione generale dell’economia montana e delle foreste del Ministero dell’agricoltura, della sovranità alimentare e delle foreste, “Sistema Informativo Forestale Nazionale (SINFor),” CREA, [Online]. Available: https://sinfor.crea.gov.it/#/. [03 08 2025].
[2] Università degli Studi di Udine, “PRI.FOR.MAN DSS 2.0,” Regione Autonoma Friuli Venezia Giulia, [Online]. Available: https://priformandss.it/. [03 08 2025].
[3] Regione Toscana, “GO-SURF. Sistema di Supporto decisionale alla pianificazione Forestale sostenibile,” [Online]. Available: https://go-surf.app/it/#/it. [03 08 2025].
[4] WSL, “HeProMo,” [Online]. Available: https://www.wsl.ch/it/services-produkte/hepromo/. [03 08 2025].
[5] J. Tolan e e. al., “Very high resolution canopy height maps from RGB imagery using self-supervised vision transformer and convolutional decoder trained on aerial lidar,” Remote Sensing of Environment, vol. 113888, p. 300, 2024.
Cover image: Three-dimensional digital model of Malga Campomandriolo – Asiago, Vicenza, Italy. It depicts grassland types, the heights and densities of trees and shrubs, and non-vegetative elements (water bodies, buildings, rocks). Created by the Author. 01.08.2025




















