Nuno Garcia Let's talk
Work

Six problems, from a protected area to the whole of Europe

Each case answers the same five questions: what the problem was, what data existed, what intelligence was built, with which technology, and what changed because of it.

ForecastingVrije Universiteit Amsterdam2023–2025

Forecasting biodiversity futures

Multi-scenario models linking climate pathways, land-use allocation and 10,000+ species to forecast biodiversity across Europe.

  • 10,000+species in the nested SDM framework
  • ~1,200European species forecast at high resolution
Read the case
Decision platformMontObEO research project2021–2023

Biodiversity intelligence platform for a protected area

An R Shiny WebGIS mapping 1,312 species across the Montesinho/Nogueira SAC, used by 40+ stakeholders including ICNF and local authorities.

  • 1,312species mapped in the WebGIS
  • 40+stakeholders supported (ICNF, municipalities, local actors)
Read the case
Earth observationMontObEO · University of Porto2021–2025

Reading twenty years of habitat change from orbit

MaxEnt niche models driven by satellite time-series in Google Earth Engine, revealing where potential species richness rose or fell between 2001 and 2021.

  • 10,190species records used to train the models
  • 2001–2021years of satellite observations analysed
Read the case
Operational intelligenceFloresta Bem Cuidada, Lda.2025–2026

GIS → operational intelligence for a forestry company

Geospatial workflows, dashboards and work-order systems that cut manual mapping time by ~40% and operational errors by more than half.

  • ~40%less manual mapping time
  • >50%fewer operational errors
Read the case
AI × environmentLabelbox / Alignerr2026–present

Domain expertise inside AI systems

Reviewing AI outputs on land-use planning, impact mitigation and environmental decisions so that models learn what actually holds up in practice.

Read the case
Open scienceEarth Commission · VU Amsterdam2024

Making a global threshold explorable

A browser app that lets anyone explore where human-modified landscapes fall below the habitat share needed to sustain Nature's Contributions to People.

  • 20–25%(semi-)natural habitat per km² needed to sustain NCP
  • One EarthCell Press, 2024
Read the case
Geography

Where the work happened

Two countries, one protected area studied for four years, and a continent modelled at one-kilometre resolution. Pick a place.

interactive map loads when in view
    • Biodiversity forecasting for Europe
    • Climate & SSP scenarios
    • CLUMondo land-use modelling
    Open case study

The same question at six scales

Sensor → planet → continent → country → landscape → site. The tools change; the logic (observe, model, decide) doesn't.

  1. Satellite Time-series from MODIS/Landsat in Google Earth Engine Twenty years of habitat change read from orbit
  2. Global Functional-integrity thresholds across human-modified landscapes 20–25% natural habitat per km² — a GEE app to explore it
  3. Continent Europe at 1 km: species forecasts under climate and SSP pathways 10,000+ species in nested SDMs
  4. Country Portugal — from protected-area science to forestry operations National and European conservation-status reconciliation
  5. Landscape A protected area: richness, suitability, scenarios for 2050 Rewilding vs agroforestry in Montesinho
  6. Site A parcel and a crew: the work order that goes out tomorrow ODT templates, approvals, dashboards