Mapping the 1998 ice storm in Maine.

In January 1998, a devastating ice storm hit northern New England, northern New York and eastern Canada. Unprecedented in its size, duration and severity, the storm deposited up to three inches of ice during five days of rain and freezing ground-level conditions. The weight of the ice downed trees, power lines, telephone poles and transmission towers, closing roads and knocking out power for weeks. Trees—depending on their species, general health and location—sustained the greatest injury. Ice loading levered roots out of the ground, bent and tore out limbs, broke crowns and snapped trunks or split them top to bottom. Over 25 million acres of rural and urban forests were damaged during the storm.

Since then, the USDA Forest Service, the Canadian Forest Service, and various state and local organizations have collaborated on studies to gauge the immediate and long-term impact of the storm on tree stands, plant communities and wildlife. In Maine, where over 11 million forested acres were affected, the Maine Forest Service (MFS) has taken the lead and set the standard for assessment, focusing first on the degree and distribution of stand damage and later on the pattern of harvest response and the percentage of tree recovery.

With the support of the USDA Forest Service and federal emergency funding, the MFS commissioned GIS consulting firm James W. Sewall Company, Old Town, Maine, to photograph, interpret and map ice-affected areas in Maine during the spring of 1998, with a second flight proposed in the spring of 1999. The mapping project was designed to produce an accurate picture of the amount and patterns of damage to forest stands immediately following the storm, and to provide natural resource consultants, local government officials and landowners with a tool for future assessment. In 2002, the MFS contracted Sewall to refly and map a subset of the original area to determine the amount and distribution of harvesting during the five-year period. The mapping would also provide the MFS with a cross-reference for assessing the degree of stand recovery. In sum, both projects produced over 22,000 aerial photographs and comprehensive GIS coverages that have been used extensively to analyze the short-term and long-term effects of ice damage to rural and urban trees.

Ice storm aerial photography coverage areas for 1998, 1999 and 2002.

Mapping Damage Patterns: The 1998-1999 Project

According to Maine State Entomologist Dave Struble, the decision to use high-resolution aerial photography for the assessment effort was driven by the complexity of the pattern of ice damage in the state. “We could delineate the major area of damage,” he says, “but the base stands, varying in species type, age and general health, were impacted by ice in patterns so complex that it was difficult to sketch map. High-resolution, leaf-off aerial photography would capture damage patterns and forest composition data, plus valuable site-specific information for landowners.”

Aerial Surveys. Under the direction of Struble for the MFS, Sewall aerial photographers captured 15,500 high-resolution, true color aerial photographs of 4.3 million acres of land during the spring flying seasons of 1998 and 1999. After a round of experimentation, they flew the photography at a scale of 1:9,000 (1"=750'), which enabled broad geographic coverage with sufficient detail. The large size of the project area—a swath from York County to Washington County, and parts of Franklin and Oxford counties—and the large number of exposures—59 rolls of color positive film—presented an unusual technical challenge in the interpretation process.

Interpreting the Photography. In conventional photo interpretation, an analyst examines every other frame of exposed film on a roll using a reel-to-reel stereoscope to identify tree stands. To delineate stand classes, the analyst overlays clear acetates directly onto the film, marking on the acetate the outline of each class (polygon) and a code that identifies each class. Each interpreted acetate is then transferred to a large mylar transparency, which is scanned to digital format and registered to an accurate base map (which corrects distortion in photographs originating from the tip and tilt of the airplane and irregular ground terrain).

To accommodate the scale of this project, Sewall analysts had to develop a new method of controlling (registering) the photography. Instead of transferring marked acetates to a larger transparency, which was impossible given the number of images, the analysts scanned and registered points on each individual acetate and stitched them together digitally to produce a clean GIS coverage. The method required close attention to edgematching during interpretation so that the delineation of polygons would align seamlessly in the final digital coverage.

In addition to interpreting tree damage classes, Sewall analysts identified the distribution of tree crown damage. Working closely with the MFS, they fine-tuned a classification matrix that included heavy, moderate and light damage in affected areas within 25 acres or more and blanket, scattered and diffuse patterns of damage distribution within the same footprint.

According to Sewall Project Manager Mary McDonald, 1:9,000-scale aerial photography was a far superior tool to use in capturing detail in a 25-acre area than satellite imagery. “Although the resolution of satellite imagery has improved, it still doesn’t provide the detail you need to see individual trees.” In the interpretation process, Sewall analysts were able to detect “white wood” (fresh breaks) and limbs bent at unnatural angles on the film and to assess polygon by polygon the percentage of crown damage to specific stands. Second, they were able to study the pattern of damage distribution within each polygon, whether continuous, scattered or scattered diffuse.

Developing the GIS Coverage

To control the interpreted photography, GIS Analyst Kyle Avila explains, Sewall used an automated process in ESRI ArcInfo software (Redlands, Calif.) to register points visible on the photography (such as roads, buildings, mountain tops and streams with known x, y coordinates) to corresponding points on USGS quadrangle maps. The stand lines from the acetates were vectorized (converted into arcs in ArcInfo) and georeferenced to the base map. After editing, Sewall delivered the master coverage to the MFS in ArcInfo format for field verification and further analysis.

The deliverables clearly indicated that the storm was a hardwood event. Sewall Analyst Scott Robinson, who assisted MFS staff with field checks of the data, or ground-truthing, reports that such species as birch, oak, maple, ash, beech and poplar were the hardest hit, suffering the greatest crown loss. Pine was the only softwood that incurred significant injury. The structure, age and overall health of the trees were factors that influenced the degree of damage as did the amount of ice loading. At higher altitudes and in open areas around lakes, ice loading and subsequent damage appeared the most severe. The pattern of damage distribution across affected areas was predominantly scattered and diffuse—another defining characteristic of this particular ice event.

In addition to the digital data produced, Sewall distributed over 20,000 9"x9" aerial photographs to affected landowners and forestry consultants on a cost-share basis. This measure was part of a federally supported MFS program to provide small landowners with technical and management planning assistance in the aftermath of the storm.

Mapping Harvest Response: The 2002 Project

With the first round of data as a benchmark for comparison, the MFS structured a reflight of ice-affected areas in the spring of 2002. During the mission, Sewall photographers acquired 6,681 true-color aerial photographs of 1.83 million acres, 1 million of which comprised a subset of the area flown five years before. In addition to interpreting the pattern of tree crown damage, Sewall analysts mapped the percentage of tree harvesting for approximately half of the acreage. Utilizing photography from both the 1998-1999 and 2002 projects, they intensified the interpretation and mapping down to a 2-acre minimum on approximately 75,000 acres of the total area. The enhanced spatial resolution was critical to identifying areas where selective cutting and tree recovery had occurred.

During the interpretation, Sewall interpreters found crown breakage, after five years, far less visible than on the earlier photography. Areas logged by manual or mechanical means were readily apparent. Holes in the canopy and straight lines or artificial stand breaks pointed to fresh cuts across ice-damaged areas. Stumps, skidder trails, brush piles, and open yards—where loggers haul wood for loading onto trucks—also signified harvesting in the aftermath of the storm. Experienced from the first project, Sewall GIS analysts refined their methods of automating and controlling the data and programmed in a new interface to streamline the edgematching and editing of both stand data and harvest data layers.

Detail of 2002 harvest response assessment to 1998 ice storm damage in East Machias, Maine.

Assessing the Data

Following the delivery of GIS coverages in September 2002, the MFS conducted rigorous ground checks of the data, cross-referencing baseline data pre-measured from forest health plots, and targeted three specific geographic areas—each roughly a township in size—to assess stand recovery and harvest response during the five-year period.

Tree Recovery. “Patterns are emerging on how well trees are recovering,” Struble says, “and we are beginning to identify the breakpoints at which recovery is good and not good. Most trees with moderate damage are recovering. Where damage was beyond the breakpoint, recovery has been virtually nil. If the pattern of ice damage was complex initially, we now have complexity a whole magnitude greater with the pattern of recovery, which varies depending on the species, site conditions and amount of damage.”

According to MFS Entomologist Henry Trial, the most severely damaged trees—with higher than 70 percent crown damage—have the lowest percentage of recovery. Trees with the least damage—10 to 20 percent crown loss—have not shown a strong sprouting response. Trees require “a threshold of damage to produce this response,” Trial explains. Thus, trees with 30 to 70 percent crown loss have the highest percentage of recovery.

For hardwoods that incurred substantial crown loss (white birch, yellow birch, red oak, sugar maple, red maple, white ash, American beech and trembling aspen), recovery has also depended upon their ability to sprout. According to Trial, trembling aspen, for example, does not sprout as well as red or sugar maple, white birch, yellow birch, red oak, white ash and American beech, which proliferate more freely. “Very few trees with crown breakage have died,” he says. They have experienced some aspect of recovery by sprouting, depending on how well they sprout.” In general, the speed of recovery has been surprising. “What we expected to see in 10 years, they have done very well in five.”

In addition to the percentage of damage and sprouting ability, a third factor—site location—has affected recovery. Study of the three targeted regions in relation to the photography has provided the MFS with another means of comparison. In the Buckfield and Sumner region of western Maine, hardwoods experienced high levels of damage due to the higher elevation. Drought conditions and rocky soil have allowed some species to desiccate. In the Liberty area, which is still somewhat hilly, a lower grade of the same species experienced damage in the middle range. A high percentage of these trees were sufficiently stimulated to restructure their crowns. The lower grade hardwoods in the Columbia and Epping region of eastern Maine—a wet area—experienced the lowest levels of damage. The availability of water during the drought has assisted these trees in the recovery process, which is comparable to that of Liberty.

Harvest Response. Using the Sewall mapping as a reference, the MFS is evaluating the degree of harvest response in relation to areas of greatest damage. Of the trees severely damaged, a high percentage has been salvaged by both small landowners and by companies, according to Trial. Those severely damaged trees left to recover, if they do survive, will never be a viable commercial commodity. The MFS anticipates the most significant effects of the ice storm will be a reduction in timber quality in the affected areas, rather than reductions in overall growth rates.

Future Applications

Driven by the need to assess forest damage and to provide landowners with recovery assistance, the MFS increased its use of GIS significantly following the storm. Sewall’s mapping of the damage footprint was a major part of this effort. Integrated in the MFS’s GIS at Augusta headquarters, the mapping is now being used with other data to assess timber volume and wood quality, to evaluate community land use, to address water quality problems, and to study salmon rivers and other wildlife habitat. As a chronicle of land use and land cover, it provides a basis for comparison against future events. The data will continue to play an expanding role for the MFS in monitoring and assessing forest conditions within the state and to educating landowners about Maine’s important forest resource.