Feeling the heat

Visualise deforestation driven temperature changes


The application visualises the effect of tropical deforestation on local temperatures across the Maritime Continent in South-East Asia and Oceania. It allows you to explore the warming effects of different deforestation scenarios. You can supply your area-of-interest (AOI) as a GeoPackage polygon file or use the drawing tools on the map to create an AOI.

How does forest loss affect temperatures?

There are several mechanisms through which forest cover and forest loss events affect temperature. These include modifications to the energy balance through changes in latent heat exchange, surface roughness properties, and albedo.

Illustration of mechanisms through which tropical forest loss affects temperatures (source: Cohn et al. (2019))

How are the predictions generated?

The predictions are generated using statistical models that relate different types of forest loss to changes in temperature. These models estimate the warming effect of different levels of local forest loss and forest loss at varying distances from the site of interest. The cumulative effect of each of these forest loss events represents the total forest loss-driven warming experienced at a location. These models also account for the nature of forest loss: clear felling or smallholder / fragmented. Smallholder / fragmented forest loss results in patchy forest cover and a reduction in the warming effect.


Land surface temperature

Land surface temperature (LST) is the temperature of the Earth's surface. LST measurements from the MODIS MOD11A product were used in this analysis.

Local forest loss

Here, local forest loss is defined as a deforestation event occurring at the same location where temperature is sensed (i.e. a 1 km spatial resolution MODIS LST pixel). Local forest loss is estimated using the Global Forest Change dataset.

Non-local forest loss

Non-local forest loss is forest loss that occurs in neighbouring regions to the location where temperature was sensed. Non-local forest loss is measured in circular regions with radii up to 8 km from the location where temperature was sensed. Non-local forest loss is estimated using the Global Forest Change dataset.

(A) A 1 km spatial resolution MODIS pixel which has experienced forest loss, (B) local forest loss and nonlocal forest loss, and (C) nonlocal forest loss but no local forest loss.

Clear fell forest loss

Clear fell forest loss is the complete removal of trees and woody cover at a location.

Smallholder / fragmented forest loss

Smallholder or fragmented forest loss events do not result in complete removal (clear felling) of forested stands. Often, what remains is patchy forest cover interspersed with other land uses. Our analysis has demonstrated that the patchiness or pattern of remnant cover affects forest cover-heat relationships and typically attenuates the warming signal associated with forest loss.

Upload a GeoPackage

Use the file upload widget to load a GeoPackage into the app to delineate your AOI. The GeoPackage should store a single geometry feature that intersects with the Maritime Continent in South East Asia and Oceania. If your GeoPackage stores more than one feature, the first feature will be used.

Drawing Tools

Use the drawing tools to draw an AOI on the map where deforestation-driven temperature change predictions will be generated.

Forest Loss Amount

Use the slider to specify the percentage amount of forest loss in each forested pixel within your AOI

Deforestation Type

Use the checkbox to select the type of deforestation event occurring inside your AOI. Clear fell deforestation results in complete removal of tree cover within the percentage amount specified. Smallholder / fragmented deforestation results in patchy forest loss within the area of deforestation.