Monitoring Forest Health with Remote Sensing Tools

“Satellite-Based Remote Sensing in Alaska” on page 11 (pdf page 21) of the 2019 Conditions Report. This essay outlines early plans and successes at adapting satellite-based remote sensing methods for FHP use in Alaska.

“Integration of Satellite-Based Remote Sensing for Forest Health Monitoring in Alaska” on page 21 (pdf page 31) of the 2022 Conditions Report. This essay explains and compares methods and data from satellite-based change detection and aerial detection surveys in Alaska.

Alaska FHP created a publicly available map: Satellite-Based Change Detection Southeast Alaska, to show persistent change associated with five years of defoliation from hemlock sawfly and western blackheaded budworm, 2018 to 2022. This extensive defoliator outbreak is known to have caused some tree topkill and mortality. By combining results from the Landscape Change Monitoring System (LCMS) with magnitude change detected by LandTrendr, we created a color severity map showing high, medium, and low change. Forest Health Protection personnel used aerial detection surveys and ground verification where possible, to determine that most of the satellite-detected change in this map is associated with defoliator activity via. Blowdown trees, landslides, and harvest may show in this map as well, although such stands were removed where recent harvest was documented in USFS records.

One of the areas most severely defoliated by hemlock sawfly is the forested peninsula north of Chaik Bay on Admiralty Island, Alaska (a). The Satellite-Based Change Detection Map (b) clearly shows change for the peninsula. A close-up photo (c) reveals the gray patches to be trees without needles that may have died or have dead tops. Photos from 2023 Aerial Detection Survey, by Dr. Karen Hutten.

A new remote sensing tool called BugNet has been developed through collaborative work with the EMapR lab at Oregon State University and the Washington State DNR. This tool isolates change caused by forest insect and disease agents by removing fire and harvest detection and highlighting areas most similar to those observed during aerial detection surveys. The tool was developed for use in Washington and Oregon (USDA-Forest Service Region 6) and will be extended into Alaska Region in the coming years.
 

Dr. Robert Kennedy (Oregon State University) and Dr. Garret Meigs (Washington DNR)
view the forest from an overlook to compare with Landsat-based change-detection maps,
August 16, 2022. USDA Forest Service photo by Dr. Karen Hutten.

Alaska FHP has collaborated with the U.S. Forest Service Geospatial Technology and Applications Center (GTAC) to test a two-date Sentinel-2 change detection tool called DeltaViewer and assess its ability to detect damage from specific agents. We also tested approaches for obtaining ground verification. So far, DeltaViewer has proven effective at detecting conifer and alder defoliation and spruce beetle damage. It appeared to be more sensitive at detecting light damage from blackheaded budworm than early August aerial detection surveys, and perhaps less sensitive at detecting light alder and spruce beetle damage.

In 2020, aerial surveys were not conducted in Alaska due to pandemic risks. Instead, FHP obtained high-resolution (50 cm) WorldView satellite imagery from MAXAR for a limited number of areas in Alaska from the 2020 summer season. The size of the area ranged from about 1,000 to 10,000 square km. Using a method called Scan-and-Sketch, FHP surveyors viewed this imagery on their aerial survey tablets, drew polygons around unhealthy forest, and attempted to determine the causal agent, host, and severity of damage, similar to conducting aerial detection surveys from the standard low-flying aircraft. Nevertheless, Scan-and-Sketch was found to be an inefficient and tedious replacement for aerial detection survey (Hanavan et al., 2022). Surveyors took more time to accurately place detailed polygons around affected forest and only had access to a few images because high-resolution imagery is not captured everywhere in Alaska each year. Identifying tree species and damage was also difficult compared to what can be seen from aircraft. Even so, high-resolution imagery continues to be useful for viewing forest characteristics and conditions when ground visitation or use of aircraft is not possible. The relative detail makes it a nice supplement to Landsat or Sentinel-2 and helps with tree damage identification.

The areas in Alaska for which high-resolution satellite imagery was obtained and surveyed for a forest damage in 2020. Aerial imagery increases from statewide (top), to northern Prince of Wales and Etolin Islands (bottom left), to a specific surveyed stand (bottom right).

As we learn from different methods and approaches, we improve our methods and available tools for the next year. Change detection processes that were tested this past year will be streamlined and used to create timely maps for forest managers in 2024. A new Survey123 app will allow FHP personnel to easily collect ground data for areas detected as change by satellite.

One of the strengths of satellite imagery is that new advanced methods can be used to quantify changes which occurred in the past, provided cloud-free imagery exists for that date. Analyses can be repeated as methods improve, and all of Alaska is covered annually, or better by some sources.

Some of the disadvantages are that satellites orbit above the clouds, requiring clear skies for a good image. Resolution and frequency of imagery can also be challenging. A detailed image generally covers less area and is data heavy, whereas a large image may cover a lot of area with less detail, making it unlikely to detect small or subtle changes. Also, trees with insect and disease damage vary widely in color, timing, severity, and visibility, potentially requiring different settings to detect different damages. It takes time to develop new methods, but Forest Health Protection will continue to add useful remote sensing tools as fast as we can test them.

Additional links and resources coming soon.