Optimization

Test every land decision for heat before the land is paved.

We think every zoning decision, new district and planting program should be tested for its effect on heat and green space before it is approved, because land decisions are hard to reverse. The WHO cites studies showing about 489,000 heat-related deaths a year between 2000 and 2019, and says cities are not being designed to limit the build-up of urban heat. FAO reports that deforestation removed 10.9 million hectares of forest a year between 2015 and 2025. extendfuture would model how land is used and covered, so planners can see these effects before they approve a plan.

Explore with AI

ChatGPT Perplexity Claude Google Grok

How can a city use data to reduce urban heat?

Map where heat builds up, then test what would cool it, street by street. We would combine satellite land cover, building footprints and weather data into a model of a city's surfaces, then use open tools such as UMEP to estimate how hot streets feel to people outdoors. Scenarios test where new trees, shade or cooler surfaces would help the most people, and what each would cost to plant and maintain. AI agents, software that sets up runs and drafts summaries, compare the scenarios. Officials decide.

Test land decisions before they are hard to undo

A model lets planners see the heat and green-space effects of a zoning decision before they make it, while the land can still be used differently. UN-Habitat data in the UN's 2026 SDG report show that across 378 cities, the share of urban residents who can walk to an open public space within 400 meters fell from 48.9% to 45.6% between 2020 and 2025. Open public space made up 2% of urban land in those cities in 2025. The same report calls for more compact, efficient and data-informed urban planning.

Start from satellites, not a new survey

Satellite data does most of the work, and much of it is free. The model would cover land cover, buildings, trees and paved surfaces.

Copernicus Sentinel satellite images, free under an open data policy
Dynamic World, a near real-time 10-meter land cover dataset from Google and the World Resources Institute
Open building footprints derived from satellite imagery and published by Microsoft
Weather station records for temperature and rainfall
Population data from the national statistics office, to see who lives in the hottest places
The city's zoning maps, tree inventory and planned projects

Report the error rate of every map

A map without a stated error rate should not decide where public money goes. We would classify each patch of land as trees, grass, water, buildings or paving, and track how that changes year by year. Land cover maps are checked against a sample of locations that people have labeled from high-resolution images, and the error rate is reported. Heat estimates from UMEP are compared with readings from weather stations or street sensors. For growth scenarios, UrbanSim forecasts where households and jobs may locate under different zoning rules, and those forecasts are checked against what happened over a past decade.

Judge planting by cooling and survival

A planting plan should be judged by how many residents it cools and how many trees survive. City and regional planners would get maps and options they can take to a council meeting.

Heat maps by street, with the number of residents exposed
Planting and shading sites ranked by expected cooling per unit of cost and by residents served
The green-space and forest effects of each zoning or development scenario
Alerts that flag tree loss or new paving between satellite images
A plain-language summary of each scenario, checked by a planner

Shade the people who have the least of it

Planting plans are checked for whether they reach the neighborhoods with the least shade and the most heat exposure, rather than the areas that ask most often. The model works with satellite images and published statistics, not with information about individual residents. Alerts about tree loss go to people, who confirm them before any enforcement. Every map records its data date and version, and officials sign off every plan.

Use open data so residents can check the maps

Maps used in a public consultation should be reproducible by anyone. Each item below is open source or open data.

Copernicus Data Space Ecosystem, for free access to Sentinel satellite data
Dynamic World, near real-time 10-meter land cover in nine classes, including trees and built-up area
Global ML Building Footprints, open building outlines derived from satellite imagery
UMEP, a QGIS plugin for urban climate work such as outdoor thermal comfort
UrbanSim, a platform for statistical models of cities and regions that forecasts development, households and jobs
QGIS, for mapping and analysis

Do trees cool a city?

Yes, in the streets around them. The US EPA says trees and other plants have a natural cooling effect and calls vegetation a simple and effective way to reduce heat islands. The effect differs between streets, so the model estimates it street by street instead of using one city-wide figure.

Can this track deforestation outside cities?

Yes. The same satellite land cover data covers forests and farmland, and change detection flags where tree cover has dropped between images. People confirm each alert before anyone acts on it.

How accurate are satellite land cover maps?

Accuracy varies by place and land type, so each map's error rate is measured and reported. High-stakes decisions, such as a protected area boundary, get a field check.

Does the plan count the cost of keeping trees alive?

Yes. Sites are ranked on the cost of planting, watering and care over the tree's life, not on planting cost alone, so the plan accounts for sites where young trees need more water.

Can residents see the results?

Yes, if the public body chooses to publish them, and the open-source tools let anyone check how the maps were made.

Sources

Talk to the people who build.

One call. An honest read on what AI can do for this, and the number it has to beat.

Start a project Book a call