Refuge notebook: Detecting insect damage by satellite

Extensive defoliation of deciduous shrubs and trees in Southcentral Alaska has received considerable attention over the last couple of years. Even for those who may not be attentive to what is going on in their natural surroundings, the conspicuous patches of dieback on Skyline and Fuller Lakes Trail on the Kenai National Wildlife Refuge, and at Summit Creek on the Seward Highway, would have been hard to miss.


Lately, the Bruce spanworm (Operophtera bruceata) and Epirrita undulata have been the main defoliators of willows, alders, and aspens in the sub-alpine. Birch leafroller (Epinotia solandriana) has been damaging birch on the lowlands, but other herbivorous insects are also involved.

Two Old World moths closely related to two of our defoliators (Operophtera brumata and Epirrita autumnata) have been well documented in similar tree line systems in Scandinavia. There, severe outbreaks are cyclical, occurring about every decade and lasting for 2-3 years. The cause of this cyclicity remains unknown, but increasing prevalence of parasitoids and disease within the moth populations over the course of outbreaks does contribute to mortality of moths.

In 2012, entomologists with the U.S. Forest Service began investigations on the present defoliation event in Southcentral Alaska, seeking to relate these outbreaks to changes in the Pacific Decadal Oscillation (PDO). The PDO is a 20- to 30-year cycle of variation in sea surface temperatures in the Pacific Ocean and is an important determinant of Southcentral Alaska’s climate. 

Outbreaks of spruce bark beetles in Southcentral Alaska have been linked to the PDO, tending to occur after a change from the predominantly cool phase to the warm phase of the PDO. The PDO has been mostly in the warm phase since 1978, but appears to have switched to a cool phase around 2008. It has been suggested that lower temperatures associated with the cool-phase PDO may favor external-feeding herbivores like the Bruce spanworm and Epirrita undulata.

It’s difficult to associate shrub and hardwood defoliation in Alaska to the PDO because we lack long-term quantitative records of defoliation other than the Forest Service’s annual aerial survey data. These data are useful for documenting outbreaks and determining the cause of outbreaks, but the surveys are limited in spatial extent, only covering the narrow path which was flown, and they are spatially inconsistent, not always covering the same areas every year.

To overcome these limitations, I analyzed the thirteen years (2000-2012) of available imagery from the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite to look for patterns of insect-caused defoliation over the Kenai Peninsula.  I used MODIS’ Enhanced Vegetation Index (EVI) product, a measure of the concentration of active chlorophyll (that is, greenness) that is updated every two weeks.

The idea behind the analysis is simple. Plants are less green in years when they are heavily damaged by foliage-eating caterpillars than in normal years. To detect defoliation, I first made a composite image representing near maximum productivity by selecting, for each pixel, the third highest value of EVI observed over the 13 years. Within each growing season, I identified pixels having EVI values at least 20 percent lower than the composite image as having been potentially defoliated. I removed water, snowfields, barren areas, and recent burns from consideration.

I was able to detect defoliation events generally in agreement with the Forest Service’s aerial survey data, though my current algorithm appears to overestimate the area damaged compared to the aerial survey data. One reason for this discrepancy is that my MODIS-derived maps provide no information as to the agents that reduced greenness, which could include diverse herbivores, plant diseases, persistent snow (retarded phenology), and urbanization.

Upon first perusal of the satellite-derived defoliation maps, I did not perceive evidence of cyclicity of defoliation events within the 13 years considered. The amount of area damaged was high in the years 2005, 2009, and 2011-2012, with 2012 showing the most area affected (680,000 acres) of all years. In 2012, there appeared to be large areas affected by defoliators in the Kenai Mountains, in the hills above Homer, and on the south side of Kachemak Bay.

I do not know why 2012 was such a big year for defoliation. Perhaps especially deep snow cover over the winter of 2011-2012 could have provided unusually good insulation for the moths’ eggs, which spend the winter on branches of the host plants where they would usually be exposed to extreme cold.

In 2013, I will be collaborating with Forest Service entomologists in their investigations of current defoliator outbreaks in the field. I will also be using MODIS imagery to identify areas of potential defoliation early in the season so that these places can be verified during aerial surveys.

Despite its limitations, this simple MODIS method appears to be a useful tool for looking at defoliation events, and it may also be of value for monitoring the dieback or recovery of defoliated areas.


Matt Bowser serves as Entomologist at the Kenai National Wildlife Refuge. You can find more information about the refuge at or


The Lake Iliamna cryptid saga

Many people do not realize that there has been thoughtful consideration taken of the presence of a creature(s) that needs explaining in the depths of... Read more