Forestry × remote sensing

Tommaso
Trotto

Remote sensing scientist working in the boreal forests using lidar and satellites.

Email me -> Resume (PDF)
Portrait of Tommaso Trotto

40+

years of Landsat imagery in my time series

1M+ ha

of processed airborne laser scanning data for fuel monitoring

3

first-author papers, one more in review

About

I work on lidar and satellite remote sensing for characterizing and monitoring forest fire fuel distribution. I am interested in deep learning, model explainability, and uncertainty. I also write Python and Rust to make remote sensing fun, faster, and simpler to work with.

Integrated Remote Sensing Studio, UBC

Now

Continuous forest inventory for forest fire fuels

Enhanced forest inventories are updated every 5 - 10 years, but fuels change rapidly. I am building one that updates every 14 days from Harmonized Landsat-Sentinel imagery to track fine and coarse woody debris, canopy fuel load and crown base height, alongside height, basal area, biomass and volume.

  1. Measure

    Field fuel plots and provincial lidar build a baseline inventory with uncertainty layers.

  2. Composite

    Harmonized Landsat-Sentinel scenes since 2016, merged into biweekly spectral composites.

  3. Decompose

    Bayesian time-series decomposition (BEAST) separates trend, seasonality and abrupt change per pixel.

  4. Update

    Models link those signals to the baseline to map fuels every 14 days, checked against drone lidar.

Skills

Code

  • Rust, for fast geoprocessing tools
  • Python, my main language for work
  • R, for lidar processing and model validation
  • Linux, git and GitHub
  • HPC on the Digital Research Alliance of Canada (SLURM)

Data

  • Airborne lidar engineering and analysis
  • Landsat and Sentinel optical imagery
  • Random forests, boosted trees, deep learning
  • Statistical modelling
  • Forest inventory, ecology, and applied silviculture

People

  • Scientific project management
  • Collaboration with industry, academia and First Nations
  • Workshops and graduate mentoring
  • Self-directed learning

Experience

2026 - now

Postdoctoral Researcher

Integrated Remote Sensing Studio, Vancouver

Building a near real time framework to monitor fire fuel changes driven by natural disturbances and forest management across eastern and western Canada.

Details
  • Collaborating with UBC, Université Laval, and First Nations on the spatiotemporal effectiveness of fuel treatments.
  • Developing transparent machine learning models, including random forest and imputation, to track fuel changes at biweekly intervals.
  • Working with Harmonized Landsat-Sentinel data and provincial airborne laser scanning across 1M+ hectares.
  • Contributing to fire fuel modelling and monitoring at large scales.

2022 - 2026

PhD Candidate

Silva21, Vancouver

Researched how disturbances reshape forest resilience at scale, combining lidar and Landsat data with machine learning.

Details
  • Published on the role of insect infestations in structural change across the boreal forests of Quebec.
  • Used supervised and unsupervised machine learning to characterize infestations at stand and landscape level.
  • Processed large data volumes on the Digital Research Alliance of Canada (SLURM).
  • Wrote Rust software for very fast geometry rasterization for forest inventory.
  • Translated results into tactical management strategies with Silva21 partner universities and industry.
  • Led workshops on machine learning for remote-sensing data and mentored graduate researchers.
  • Mentored a diverge group of graduate researchers.

Research

  1. Trotto T., Coops N.C., Achim A., Gergel S.E. (2025). Spectral remote sensing reveals forest structural characteristics resilient to spruce budworm infestations. Ecological Indicators.

    Summary

    Forest resilience is central to forest management, yet frameworks to quantify it comparably across studies remain largely theoretical. We operationalized these frameworks using three decades of Landsat surface-reflectance composites and a two-stage clustering approach, computing spectral impact and recovery rate from indices of greenness, moisture and structure. Boreal forests proved more resilient in greenness and structure than in moisture, likely owing to post-infestation foliage recovery.

  2. Trotto T., Coops N.C., Achim A., Gergel S.E., Roeser D. (2025). Characterizing landscape configuration effects on eastern spruce budworm infestation dynamics. Landscape Ecology.

    Summary

    Landscape configuration is known to affect insect-infestation movement and severity, but a spatially explicit characterization of which configuration properties drive these dynamics is still missing. We used 13 years of Landsat composites and Random Forest to assess the role of forest fragmentation on infestation severity in Quebec. Increased fragmentation may raise the likelihood of severe infestations, potentially through landscape saturation effects.

  3. Trotto T., Coops N.C., Achim A., Gergel S.E., Roeser D. (2024). Characterizing forest structural changes in response to non-stand-replacing disturbances using bitemporal airborne laser scanning data. Science of Remote Sensing.

    Summary

    Understanding which forest structural attributes affect insect infestations at fine scale is challenging. We investigated a bitemporal airborne laser scanning dataset to extract structural characteristics linked to spruce budworm infestations in the boreal forests of Quebec using unsupervised machine learning. Canopy height and cover best characterized the effect of different infestation severities across the landscape.

  4. Trotto T., et al. (in review). Uncovering forest characteristics supporting resilience to spruce budworm infestation over Quebec, Canada using Landsat time series. Manuscript.

Projects

rusterize

author, 2024 - now

Extremely fast rasterization for every geometry type, written in Rust with ports to Python and R. No GDAL dependency. 

$ pip install rusterize
$ cargo add rusterize-rs

docs: python r rust

sgsPy

contributor, 2026

Structurally guided sampling for forest inventory, in Python.

DRI-EDIA F4A

contributor, 2025

Advancing equity in forestry through digital research infrastructure and deep learning.

3PG

contributor, 2025

Physiological Principles Predicting Growth, extended with spatial features.

Education

2022 - 2026

Ph.D., Forestry & Remote Sensing

The University of British Columbia

2020 - 2021

M.Sc., Forestry

The University of British Columbia

Theses on individual tree crown delineation and spectroscopic foliar traits in dense Douglas-fir stands. Supervised by Dr. Nicholas Coops.

2019 - 2020

M.Sc., Forestry

Università degli Studi di Padova

30 cum laude / 30. Transatlantic double degree with UBC.

2019

B.Sc., Forestry

Sir Sandford Fleming College

2016 - 2019

B.Sc., Forestry

Università degli Studi di Padova

30 cum laude / 30.

Awards & talks

2025
Academic workshop on AI in forestry, UBC. Workshops supporting diversity and inclusion in forestry and AI.
2024
Invited talk: mapping disturbances from space, Bogotá. Landscape-configuration effects on natural disturbances.
2024
Basic drone pilot certificate, Transport Canada. Small remotely piloted aircraft (VLOS).
2022
ESRI Canada Scholarship, UBC. Best map, on multi-temporal lidar tree-height growth in the Petawawa Research Forest.
2019
Transatlantic Forestry Master, Padova and UBC. Double-degree program across Europe and Canada.

Contact

Open to research collaborations, talks, and conversations about forests and data.

tommaso.trotto@ubc.ca