Science

Missing Forests Are Messing With Climate Targets

Countries are using forests to pad their climate commitments. New satellite images might call their bluff.
Image may contain Plant Tree Nature Weather Fir Abies Outdoors Fog Conifer and Mist
Getty Images / WIRED

All products featured on WIRED are independently selected by our editors. However, we may receive compensation from retailers and/or from purchases of products through these links.

In late 2018, the Vietnamese government submitted a document it thought would be worth $51.5 million. The country was expecting to be paid through a UN-backed scheme called REDD+ that pays countries if they reduce emissions by keeping forests standing. By the reckoning of the Vietnamese Ministry of Agriculture and Rural Development, the country had done pretty well: it calculated that forest cover in Vietnam had increased over the previous 25 years, from 28 per cent in 1990 to about 41 per cent in 2015. 

There was just one problem: satellite data told a very different story. Keiko Nomura, who at the time was getting her PhD in geosciences from the University of Edinburgh, compared the deforestation rates submitted by seven countries to satellite-based observations of tree cover change. Nomura, who had previously worked as a program officer for the UN Environment Program on several REDD+ projects in South-East Asia wanted to assess whether the interventions countries were planning would actually target those areas with the highest rates of deforestation. To her surprise she found that, contrary to what the country had claimed, deforestation of natural forests in Vietnam had actually increased over the time period, a trend other satellite-based studies have also noted.

This wasn’t the first time researchers had noted discrepancies between country deforestation estimates and satellite images. A few years earlier, Do-Hyung Kim, then a PhD student at the University of Maryland, was attempting to verify a claim made by the UN’s Food and Agriculture Organization (FAO) that deforestation in the tropics had reduced by 25 per cent between the 1990s and the 2000s, based in large part on estimates countries had provided to the FAO. Kim analysed 5,444 satellite images, comparing past and present forest cover. His study completely contradicted the FAO’s report. Instead of a decrease, he found there had been a 62 per cent acceleration in net deforestation over the same time period.

These studies hint at a crucial and contentious problem in climate policy: uncertainties in how we define and monitor forest cover and emissions could jeopardise global climate change commitments. But forests are a key component of the Paris Agreement. According to a 2017 estimate, a quarter of the reductions in greenhouse gas emissions that countries have pledged will come largely from forests; either by increasing their carbon sucking potential, or decreasing the rate at which they are being destroyed.

But calculating emissions from forests in an accurate and comparable way is riddled with uncertainties and methodological differences. These uncertainties in estimates could also hamper our ability to know how close – or far away – we are from achieving the Paris Agreement goal of staying below two degrees Celsius of warming.

Now scientists have a new and powerful tool at their disposal. Rapid advancements in space-borne technologies, high resolution imagining, as well as enhanced computing power has ushered in what some have called the “golden age” of forest monitoring, giving scientists unprecedented information about how much carbon is actually stored and released by forests. Experts say it is the first step in creating a global standard for calculating forest carbon and emissions, but it could also highlight large discrepancies between the amount of carbon countries say is locked up in their forests and what is actually there.

The main driver behind climate change is the burning of fossil fuels, which contributes around 75 per cent of global carbon emissions. It’s relatively easy to both calculate and compare fossil fuel emissions across the board; we can calculate how many emissions a coal power plant in India produces compared to the UK based on the type of coal and efficiency of the plant, and we know down to a pretty accurate level of precision the tailpipe emissions of gas-guzzling cars.

We are much, much less certain about the second largest source of greenhouse gas emissions: the 25 per cent of emissions that come mainly from deforestation and agriculture. Calculating how much CO2 an individual tree emits or how much carbon it sequesters over its lifetime is extremely difficult. The problems are multiplied when you scale things up to forests, which vary widely across geographies, seasons, species, and way they are managed. The field is subsequent to such intense debate that the journal Nature recently dedicated an entire edition to this topic.

This is especially true for tropical forests, which despite being the most important land sink for sucking up excess carbon are also “the least certain major component of the global carbon budget” according to a review article in Nature. Tropical forests contain thousands of different species and diverse ecosystems and lie in lower-income countries with sparse resources for monitoring and underfunded forestry departments. In many cases it’s too difficult, costly, and complicated for countries to accurately calculate, meaning that countries either completely leave out forests of their climate goals, or include them piecemeal or with uncertain numbers.

Claire Fyson, a climate policy analyst at the non-profit Climate Analytics, co-wrote a study that calculated the massive ambiguity that exists in how exactly countries use forests and land-based activities to reach their climate goals. She looked at how countries were including forests and land in their individual commitments to the Paris Agreement and figured that depending on what definitions and methods you used, the range of uncertainty in the global annual emissions from forests and the land sector would be almost three gigatonnes of CO2 per year in 2030. For comparison, that is half of the global emissions from the forest and land sector in 2019

For climate policy analysts, this is not surprising; the land sector has always been a contentious part of climate agreements. Fyson calls the land sector the “unwanted child” of global climate negotiations.“No one really knows how to deal with it, right? It's uncertain. It comes with capacity issues, there are governance issues,” she says. “So I think there's still a lot to be resolved in our ability to sort of be confident that we're on a good pathway.”

But these large data uncertainties shouldn’t be brushed aside, especially considering the outsized role forests play in achieving global climate goals. Forests cover 31 per cent of global land area – around four billion hectares — and they suck up around 25 per cent of the excess carbon we release. “The forests are really saving our bacon and they have been for a while,” says Daniel Hayes, a professor of geospatial analysis and remote sensing at the University of Maine. Barring any massive technological advancement to suck carbon out of the air, there is no scenario that exists to keep us below dangerous levels of warming that doesn’t depend on forests as a key component.

The real test will come to a head as early as this year when countries come together to start defining the rules about how they will measure their collective progress towards achieving the Paris agreement’s climate goals, known as the “global stocktake.” Under the Paris Agreement, every five years the parties conduct a massive bookkeeping exercise to look at what countries said they would do and what has actually been done. The first one, set to start next year and conclude in 2023, will be one of the most important benchmarks against which countries decide when and how much to ramp up their climate commitments.

“I would say the key engine of the Paris agreement is the global stocktake,”says Giacomo Grassi, a forest ecologist at the European Commission’s Joint Research Centre who leads the UN group responsible for helping develop guidelines for reporting on forests. But a growing number of studies have pointed out that large uncertainties about how we calculate the carbon in forests will likely constrain the global stocktake.

According to published study by Grassi and his colleagues, there’s a 5.5 billion ton difference – roughly the annual emissions of the United States – between what largely independent, satellite-based global models are telling us about what forests are emitting and what country-level greenhouse gas inventories are telling us. In large part, it’s because models differentiate between natural causes of forest emissions and human-causes, and countries don’t. The difference can be alleviated if models essentially reduce the complexity of their assessments, but “in the absence of these adjustments," the study states, “collective progress would appear to be more on-track than it actually is.

New advances in remote sensing, such as the widespread adoption of space borne LiDAR technology coupled with machine learning and faster computing power could further show just how much we have been over or underestimating forests' in staving off the worst of climate change. It could represent either a major obstacle or a major opportunity, but most scientists think it’s the latter. “I think we are approaching a revolution for the integration of the observations of remote sensing in the monitoring of forest resources,” says Grassi.

Until recently, remote sensing for forests was largely based on Landsat time-series imagery, 30-meter resolution satellite data that forms the basis of projects like Global Forest Watch (GFW), a platform that provides global-scale forest loss data. This kind of data can tell the difference between surfaces such as forest, grasslands and desert, but couldn’t tell you the height of trees or tree cover – a critical piece of information for estimating forest emissions. Those calculations could only be indirectly estimated, based mainly on ground sampling.

Now light detection and ranging (LiDAR) technologies, which shoots lasers down to earth and measures the distance to create a 3D image of the Earth’s surface are changing the game, says Greg Asner, the founder of the Global Airborne Observatory (GAO) and director of ASU's Center for Global Discovery and Conservation Science.

In 2017 and 2018, Asner and his colleagues trained machine learning models using forest carbon estimates from their LiDAR airplane data, planet satellites, and other auxiliary data to comprehensively measure every inch of aboveground carbon in Peru. Their initial results showed not only that it was possible to do so, they also revealed that carbon emissions caused by deforestation in Peru were 23 per cent higher than reported. When they added up the emissions based on quarterly calculations instead of annual ones, they found an 83 per cent increase in previously reported annual emissions.

“This is way more accurate and real compared to what's being done [up until] now,” says Nomura, who previously worked for Space Intelligence, a company that relies heavily on similar LiDAR data to create continuous maps of forest carbon for private and public entities. Until recently, LiDAR calculations were exclusively done using planes, making deployment both more expensive and time-consuming, and limiting the area that could be calculated. Now LiDAR is being deployed on satellites in space, such as NASA’s Global Ecosystem Dynamics Investigation mission (GEDI) and ICE-SAT-2. New satellites are being launched every year, many explicitly to measure forest carbon and vegetation, making this once expensive and area limited data available globally.

It’s this leap from airplanes to space borne satellites that has really heralded the "golden age" of earth observation science. Asner estimates that by this year or next he won’t need to rely on aircraft based approaches at all.“It'll all be space based,” says Asner. “And that's really the scale to take us global.”

LiDAR isn’t the only technology that has made measuring and monitoring forests easier and more accurate, although most experts say it’s the biggest. It’s been accompanied by much higher-resolution satellite images in general and advancements in machine learning and computing power.

“The technology and the data is already there, what is needed is more computing power to sort through it,'' says Asner, noting that boots on the ground and data analysis also need to work in tandem with LiDAR data to get a full picture of forest carbon and emissions. “Each satellite is only giving us a piece of this tree shape and size and change puzzle.They really have to work together to actually get the carbon stocks and the emissions.”

Of the sixty people on Asner’s team, he says, half of them are AI specialists. His study also depended on tens of thousands of high resolution optical images from Planet, one of a growing number of private companies that have sent hundreds of small satellites into orbit, providing constant coverage of forests at a resolution as high as 50cm.

“It’s really not a Research and Development issue anymore,” says Asner, who has helped map forest carbon in several countries already. “It's really about scaling up the effort.”

And these efforts are already being scaled rapidly. In January this year, scientists for the first time ever were finally able to map forest carbon emissions, removals, and fluxes for the entire globe. Using a mixture of plot data from more than 20,000 sites, over 700,000 LiDAR observations, and information from around 90 billion Landsat pixels they were able to assess how the changes in forests over the past two decades have impacted carbon concentrations in the atmosphere, pinpointing which forests had become a source of emissions, and where the world’s largest forest sinks remained.

“These kinds of maps are only now becoming even possible,” says David Gibbs, a geospatial scientist at the World Resources Institute who helped create the global forest carbon flux map. ”There's tremendous potential in using remote sensing for monitoring forest carbon, we're already so much further ahead than we were five years ago.”

COP26 is expected to lock in the role of forests for mitigating climate change. But countries, and the UNFCCC are a long way off from using this cutting edge information across the board. And many countries have a vested interest in being able to decide how much carbon is locked up in their forests and how much they are emitting.

“The question of what the possibility is of using satellites for monitoring forest carbon is not entirely technological,” says Gibbs.“The first step is technological. And I would say that we basically have the technology down. And a lot of the issues now are policy related and politics related.”

Experts are advocating for the better integration of Earth Observation Systems in how emissions are calculated. For example the Committee of Earth Observation Satellites recently called in a plenary session “to recognise the magnitude of the opportunity for satellite Earth observations in support of the Global Stocktake (GST) process.”

But most insiders believe that the widespread adoption of these technologies as a way of accurately calculating global forest carbon and carbon emissions will proceed at the pace of politics, not science.

If these technologies are scaled up to the level of nations, it could be a game changer not just for countries, but also for figuring out just how much we can, and should, rely on forests. “Once they create this kind of wall-to-wall better forest carbon [measurements], it's easy to extract deforested areas, or more accurate forest carbon figures and they can report much more accurate numbers with less uncertainty,” says Nomura. “And I think that's extremely important because the figures could be very different, and our understanding of emissions from forest and use change and climate change mitigation, that may all change.”

Updated 02.11.2019, 13:00 GMT: The article has been updated to clarify details of Giacomo Grassi's study.

More great stories from WIRED

This article was originally published by WIRED UK