Satellite Glow Revealed Forest Stress Before Beetles Killed Trees

By late summer 2020, something had already changed in parts of the Sierra Nevada, Klamath Mountains and Cascades. A satellite instrument was recording a weaker photosynthetic signal from forest pixels that looked, by the standard aerial record, not yet stricken by widespread insect mortality. The USDA Forest Service aircraft surveys would not mark substantial bark beetle damage there until 2022. Yet the spaceborne signal had begun to separate from comparable forests about two years earlier.

The signal is solar-induced chlorophyll fluorescence (SIF), a faint glow that plants give off after chlorophyll absorbs sunlight. Because that glow is tied to the machinery of photosynthesis, it can reveal changes in plant activity before leaves or needles visibly disappear. NASA describes SIF as a direct proxy for photosynthetic activity, one that can brighten or dim as water, temperature and sunlight become more or less favorable.

Lewis Kunik of the University of Utah and his colleagues asked whether that glow could expose forest stress early enough to matter. Their peer-reviewed observational study paired SIF measurements from the TROPOMI instrument on Europe’s Sentinel-5P satellite with wildfire records, vegetation products and U.S. Department of Agriculture Forest Service aerial mortality surveys. The analysis covered the TROPOMI era from 2018 through 2023 and worked at 0.05-degree pixels, roughly 5 kilometers across. The key bark beetle comparison was small: 29 forest locations represented by mortality pixels, against 1,315 control pixels, alongside 1,061 wildfire pixels.

The Glow Faded Before the Canopy Looked Dead

For the forests later classified as bark beetle affected, SIF fell to about 70 percent of its 2019 level in September 2020 and below 60 percent in August 2021. Nearby control forests were also stressed by drought, but their declines were weaker on average. After the researchers repeatedly resampled mortality pixels against biogeographically matched controls, the difference was already statistically distinguishable in late summer 2020. Across the analysis, the bark beetle group’s decline was about 10 to 20 percent greater than the controls.

That is earlier than the usual visual evidence. Forest Service aerial surveys are conducted by observers in aircraft who map tree damage and mortality, typically after crowns have changed enough to be seen. The satellite was measuring something different: not dead crowns directly, but a change in how vegetation handled absorbed light.

The researchers also calculated a fluorescence-efficiency measure called SIFyield. In simplified terms, it asks how much fluorescence comes back out after accounting for the light available to plants and a satellite proxy for how much light the canopy absorbed. SIF and SIFyield declined together before visible mortality, suggesting that physiology, not just missing foliage, contributed to the early signal.

Fire Put the Signal Through a Reality Check

Bark beetle outbreaks are difficult test cases because drought, insects and tree death unfold together over years. Wildfire offered a cleaner check. The researchers knew where fires occurred and had established severity maps. If SIF really responded to forest disturbance, its loss should deepen as burns became more severe.

It did. In the months after fire, SIF fell as low as 60 percent of pre-fire levels after low-severity burns, 40 percent after moderate-severity burns and 20 percent after high-severity burns. The pattern gave the researchers confidence that the satellite signal could register real losses in photosynthetic vegetation, even though fire and beetle mortality affect forests in very different ways.

SIF was not universally the most sensitive satellite measure. Some conventional products performed as well or better after wildfire. The advantage appeared before beetle mortality became visible: SIF showed greater sensitivity than the other canopy products the team tested during the 1 to 2 years before the first aircraft detections.

Drought Keeps the Warning From Becoming a Forecast

The hardest part of the result is also what keeps it from being a prediction system. Drought was everywhere in the signal. Across the region, many large SIF declines occurred where no mortality had been documented, and the broad spatial pattern often resembled precipitation and drought anomalies more closely than it resembled the later beetle map. Even after matching mortality pixels with similar controls, some drought differences remained. The authors therefore treat the extra decline in the later mortality areas as a subtle signal layered on top of a much larger climate stress pattern, not as a clean fingerprint of beetle attack.

There are other limits. A TROPOMI pixel blends kilometers of forest, including trees, shrubs and grasses. Open evergreen canopies can expose substantial understory vegetation to the satellite, and those plants may respond differently to drought or disturbance. The beetle analysis also rested on only 29 mortality pixels that met the study’s threshold for moderate to severe damage. This is a regional signal, not a tree-level diagnosis.

That distinction matters for what comes next. Sentinel-5P provides daily global coverage with TROPOMI, which made the long sequence of observations possible. The European Space Agency’s forthcoming Fluorescence Explorer (FLEX) mission is designed specifically to map vegetation fluorescence at 300-meter resolution, potentially narrowing the view from multi-kilometer pixels toward individual forest stands.

But a sharper image will not erase the biological ambiguity. Drought can dim fluorescence without leading to widespread death, and beetle outbreaks can develop unevenly across a landscape. As Kunik put it, “We still can’t predict if, or where, mortality will occur just based on SIF observations, but our work shows that SIF could be another powerful tool for identifying areas of concern.” For now, that is the evidence boundary: the glow can tell managers where a forest may deserve a closer look before the damage becomes obvious, but not which trees are going to die.

Frequently Asked Questions