We have drawn on data gathered by three different instruments orbiting Earth to develop a sophisticated computer model that is able to forecast how quickly forests will regrow—and thus how rapidly they will suck carbon from the atmosphere—at any given location on Earth. This is a potentially transformational technology in efforts to combat climate change because it will support the efforts of organizations regenerating forests in two ways. First, it will guide them in their decisions about where to focus their resources and efforts. Second, and even more important, it will allow them to finance their work in advance by calling on international markets in voluntary carbon credits; in essence, our technology makes the sale of carbon credit "futures" a realistic possibility.
We built our model in three main steps. First, we drew on NASA's LANDSAT 5, 7, and 8 time-series data—roughly weekly snapshots of the planet's surface that go back nearly twenty years. Searching these images for abrupt changes in the appearance of forested areas, our algorithms identified patches that have suffered large deforestation events. Another algorithm then worked to delineate precisely the edges of each patch, so that subsequent steps in the analysis would not be biased by fragments of remnant forest at the edge of each disturbance.
Second, we used data from the remarkable GEDI instrument on the International Space Station. This instrument bounces a laser off the surface of Earth, inferring the 3D image of the landcover. From this 3D image, it is possible to estimate the current accumulation of above-ground forest (or "biomass", the total amount of carbon-based material) for any given spot roughly 25m across. This accumulation of biomass, in turn, is an approximate measure of the quantity of carbon that has been pulled from the atmosphere during regrowth. GEDI data is available only for a relatively recent window of time. However, by examining numerous patches in a forested region, each disturbed at a different point in the past, we reconstructed a curve showing the accumulation of biomass over time in a particular region of forest (Figure 1).
In the third and final step, we drew on multiple global datasets characterizing climate and topography for all points on the planet. An especially important source of data at this stage was the European Space Agency's Copernicus mission, from which we acquired data on local temperature fluctuations, humidity, windspeed, soil moisture, slope, aspect, elevation, and over ten other variables. We then used an approach called "machine learning", in which a computer is allowed to learn from huge quantitates of data, to build a model that relates the climate and topography of any point on Earth to the regeneration of biomass that we inferred in step two of our analysis, based on the LANDSAT and GEDI data.
Built and fine-tuned based on one set of disturbed patches of forest, our predictive model was then tested on a separate set of disturbed patches in the same region. The result is shown in Figure 1, in which the blue curve shows the output of our predictive model for the same region of forest that yielded the empirically derived curve shown in red.
Figure 1. Red points relate time since disturbance (x-axis) to present accumulated above-ground forest (y-axis). Disturbed patches of forest were located in space and time using our algorithms processing LANDSAT data, while above-ground biomass (a measure of carbon sequestered) was estimated from GEDI data for each patch. The red line is the best-fit curve for empirical points. Blue points are predicted by our machine-learning model (see text for explanation) based on each site's topological and climatic characteristics. The blue line is the best-fit curve for predicted points.
The goal of our lab is to create a high-spatial resolution map of coastal forested wetlands at global scale. If we know precisely where these ecologically critical but fragile forests are located, we can manage freshwater flows to counteract saltwater introgression due to rising sea levels, and we can assist in their migration inland, preserving their critical function in protecting coastlines and sequestering carbon.
Across the continent, a number of first nations are in the process of reintroducing bison to the grasslands in which they were once the primary grazer and an ecologically vital species. Initial experiences and evolutionary considerations suggest that this may be ecologically beneficial in terms of grassland biodiversity, carbon cycle, and resilience to climate change. However, these questions have not yet been studied at scale. In this lab, we will leverage remote sensing to scale up from ground measurements, establishing the large-scale patterns of bison impact.
Beaver dams are known to result in greener, more drought-resilient waterways in semi-arid environments. We are using computer vision to spot dams in satellite imagery, generating a large dataset that we can use to train models that will tell us what the ecological effects of a dam will be at any point on a waterway. The goal is to create a tool to guide efficient restoration through the introduction of small dams.
Markets in voluntary carbon credits are increasingly providing a flow of capital for regenerating ecosystems. The problem is, thriving and resilient ecosystems are not just carbon. We need to find ways to structure credits to incentivize the diverse and functional ecosystems we want, not merely high-concentrations of carbon. We will design the technological tools to support a market in bundled ecological credits.
We are building an accurate and global model for predicting potential rates of reforestation and resulting carbon sequestration. Such a model could have a transformational impact on global reforestation efforts by opening new streams of financing in the form of carbon credit futures.
Leveraging The Earthshot Institute’s broad scientific and technical expertise, the Impact and Risk Lab helps investors and governments who earnestly want to forecast, measure, and address the socio-ecological risks to and/or impacts from their work. For a given system, we build simple process-based models to identify key socio-ecological risks and outcomes. We then draw on big data to improve and train our models, generating quantitative predictions and developing measurement systems for verification.