Introducing Restor Disturbance Alerts

Field teams can't be everywhere at once across thousands of hectares of land. Fire, illegal logging crews, or disease outbreaks move faster than human patrols can track, quietly erasing years of hard-won progress in conservation and restoration before anyone can react.

Disturbance Alerts change that. Restor’s new offering acts as a fully automated security network, giving you near-real-time visibility so you can catch threats and act before damage spreads. 

An early warning system for your land

Powered by the OPERA DIST-ALERT algorithm (co-developed by the University of Maryland and NASA, published in Nature Communications (Pickens, Hansen et al., 2025)), the system processes NASA Landsat and ESA Sentinel-2 satellite data to check your site every 1 to 4 days.

How it works

• Every satellite pass is checked against your site’s unique 3-year rolling baseline, filtering out natural seasonal shifts

• Alerts are verified across repeated satellite passes before flagging, saving your team from chasing false positives.

• Delivered monthly or quarterly, each report maps the exact footprint of change, ranks its severity, and provides side-by-side before-and-after imagery 

Note: an alert tells you that land changed, not why. That's why every report recommends checking on the ground.

What does it detect?

Both sudden, direct vegetation loss and subtle environmental stress. The system flags rapid clearing driven by agriculture, logging, mining, fire, drought, and landslides. It also picks up gradual canopy changes, including insect infestations, disease outbreaks, delayed spring green-up, and shifted growing seasons.

Smart triage 

Not all disturbances are equal. By automatically sorting alerts into low, medium, and high severity tiers, field teams can prioritize their limited time and resources.

At one monitored site, 64 ha of loss was caught across 21 separate micro-events. Because satellites caught them early, the local team didn't have to wait around for an annual survey; they knew right away.

Try it free

Anyone on Restor can try Disturbance Alerts for free. The trial gives you one report for the previous month, for one site up to 100 ha, with no ongoing commitment. Disturbance Alerts is also fully integrated into Restor Insights.

Get a free trial | See an example report

Coming soon: Vegetation Change Reports

While Disturbance Alerts catch sudden loss, our upcoming Vegetation Change Reports will track long-term recovery at 10-metre resolution back to 2018, showing you exactly where restoration and regeneration are taking hold- a win!  Join the waitlist

————————————————

Key insights from our latest webinar

In our latest science webinar, Restor founder Tom Crowther, Lead Geospatial Scientist Andrew Cottam and PhD candidate Pauline Depoortere (University of Liège) walked through the science behind Restor.  You asked brilliant questions after the session. 

Here are a few of them, lightly edited for length.

Can you track positive "disturbances", like restoration, rewilding or natural regeneration? (John Cooper) Yes. That's exactly what our upcoming Vegetation Change Reports are for. They analyse how vegetation changes over the long term at 10-metre resolution, specifically to track growth and recovery.

How do Vegetation Change Reports handle tropical dry forests, especially when El Niño changes the usual pattern? (Catalina Mejia) 

In strongly seasonal habitats like tropical dry forests, the reports look at the long-term trend in vegetation health (NDVI) over about eight years. Fitting a single trend line across that period smooths out the normal seasonal peaks and troughs within each year. You're right, though, that El Niño years introduce real year-to-year anomalies, and these will influence the trend.

Can Restor predict areas vulnerable to forest fire over the next decade, and help governments act early? (Geraldine Donovan) 

Restor already has datasets that help you understand fire risk, such as the Aridity Index and a layer showing fires from the past 30 days. We don't yet have a predictive fire-risk layer. If the community wants one, we'll look at the leading research to build it.

Are your models cross-validated against data from the ground? (Arthur Parry) 

Our models go through extensive ground-truth validation using k-fold cross-validation. For example, with a million observations, you hold back 100,000, train the model on the other 900,000, and test its predictions against the ones you held back. You repeat this for every subset. When Restor users flag incorrect local values, that feedback also feeds into future versions of the models.

Global datasets are applied across very different ecosystems; for example, forest-based carbon values showing on mangrove sites. How should we interpret these values? (Lanie Esch) 

There's a well-known rule in this field: every model is wrong, but some are useful. No map is ever 100% accurate, yet without them we couldn't manage natural resources at all. Model outputs are valuable as long as you understand their uncertainty, and where uncertainty is high, you should hold your conclusions more loosely. Our job is to make that uncertainty clear. Satellite resolution is also improving fast, now down to 10 metres, which helps with tricky boundaries like the edge between land and sea.

Do your models capture human-caused changes, or predict what landscapes would look like without human influence? (Alexander Russell) 

Our monitoring layers show current land cover from satellite imagery. Tools like the SEED Index go further, comparing a site's current biological complexity with what it would be under natural, undisturbed conditions. Upcoming research using data from Restor users also suggests that clearing nature initially frees up room for agriculture. But past a critical threshold, which around 72% of the world has now crossed, losing nature starts to reduce crop yields, because farms lose climate regulation, soil stability and pollinators.

Can we add our own field data to improve the models' accuracy? (Maria Moncada) 

Yes, and it's one of the best ways to improve accuracy locally. You can upload your own field measurements to Restor using Self-Reported Data, and we'll keep adding more field survey variables over time.

Link to the webinar recording can be accessed at this link.