HURRICANE HUGO--DAMAGE ASSESSMENT USING PERMANENT PLOT REMEASUREMENT
Raymond Sheffield Supervisory Research Forester Southeastern Forest Experiment Station P.O. Box 2680 Asheville, NC 28802 ABSTRACT: A network of 3,500 permanent ground plots was utilized to assess damage from a major hurricane that struck the Southeastern United States in 1989. Hurricane Hugo inflicted varying degrees of forest damage across an area of 4.4 million hectares. This paper summarizes data collection and analytical procedures used to assess forest damage across this area. Results that are easily provided from remeasurement of permanent plots are highlighted. Advantages and disadvantages of ground-based permanent plot assessment systems are highlighted. INTRODUCTION Hurricane Hugo struck the Southeastern United States in September 1989 with winds of 217 kilometers per hour, or 60 meters per second (Powell, Dodge and Black, 1991). Hugo extensively damaged forest stands in North Carolina and South Carolina. It significantly altered forest conditions and inventory volumes and rendered many short- and long-term forestry resource programs obsolete. Forestry agencies in the two States conducted aerial and ground reconnaissance surveys to determine the extent of the damage. Their observations and subjective estimates were adequate for planning initial responses, but they needed more quantitative measures for long-term planning. The South Carolina Forestry Commission asked the Forest Inventory and Analysis (FIA) Research Work Unit at the Southeastern Forest Experiment Station to conduct a special inventory of forests in that State's affected area. This damage assessment was later expanded to include a large portion of North Carolina ( Figure 1 ). The area of timberland in the study region totaled 4.4 million hectares. Utilizing a previously established network of permanent ground plots throughout this region, FIA designed damage inventory procedures to: (1) quantify the extent of damage and describe associated stand conditions, (2) determine the volume of hurricane-related mortality and damage to live trees, and (3) provide an estimate of current inventory volume. METHODS FIA is a research sctivity of the USDA Forest Service. Its mission is to comprehensively inventory the status and trends of the country's diverse forest ecosystems, their use, and their health (U.S. Department of Agriculture, Forest Service, 1992). Six regional FIA units maintain a network of permanent samples on forest land across the United States. More than 3,500 of these permanent plots were in North and South Carolina counties damaged by Hurricane Hugo. Prior to the hurricane, these ground plots had been visited and measured last in 1984 in North Carolina and 1986 in South Carolina. Field Procedures Data collection procedures are briefly summarized in this paper; details are available in a report by Sheffield and Thompson (1992). Two-person crews visited each sampling location in 1990, 5 to 9 months after Hurricane Hugo struck. Each tree above a minimum diameter at 1.4 meters above the ground was relocated. The minimum diameter used was 7.6 centimeters for trees in natural stands and 2.5 centimeters for trees in planted stands. Each tree was assigned one of the following six tree histories: <> tree alive without hurricane damage <> tree alive with hurricane damage <> tree dead (hurricane related) <> tree dead (other causes) <> tree cut, associated with hurricane salvage efforts <> tree cut, associated with normal harvest operations For all live trees with hurricane damage, specific classes of damage were noted. Damage classes were: <> leaning or bending (degrees from vertical) <> root system damage <> damage to tree bole <> loss of tree crown Any combination of these could be assigned to a tree; multiple damages, such as severe lean and exposed root system, were common. Analytical Procedures Standard protocols for the analysis of data from a damage inventory are not available. One must adjust methods to account for the magnitude of loss and the specific goals that have been specified. After Hugo, we wanted to quantitatively estimate damage to individual trees and to stands of trees. Little published data were available in the literature to assist in this process, so various approaches to solving these problems were used. In a ground-based damage inventory, the types of damage to individual trees must be categorized. Damage may be so minor that the tree's value, growth, and probability of survival are not affected. Conversely, damage may be so severe that the tree's value has been destroyed, its growth potential reduced to a fraction of normal, or its survival chances severely reduced. Guidelines for assigning meaningful categories of damage were not found in the literature, so we developed our own classification framework. Our design permits decisions about individual trees to be made on the basis of value, likely contribution, and development in the future. Detailed criteria for assignment of softwood and hardwood trees into four damage categories were developed and modified based on review by more than a dozen individuals. Sheffield and Thompson (1992) describe the process and the criteria. Here I present the damage/risk classes used and a brief description of levels of tree damage in each class. Class 1. High-risk tree with a probability of dying in the near future. damage and value loss are severe enough that this tree should not be retained in the stand. Class 2. Moderate-risk tree with elevated risk of mortality. Damage and value loss are substantial enough to make the retention of this tree in the stand questionable. Retention depends upon management objectives, stand and rotation age, and product objectives. Class 3. Low-risk tree with a high probability of survival. Damage elevates the risk of mortality slightly, but reduced growth and value degrade are minimal. This tree should be retained in the stand under most management scenarios. Healthy. No obvious hurricane damage. A tree with hidden or internal damage would be classed as healthy. The assessment of stand conditons depends on the condition of individual trees remaining in the stand. Thus, the decision to regenerate or salvage requires sound judgments about individual trees and about the level of residual stocking needed for an acceptable stand. One must also determine which categories of damage are acceptable for inclusion in minimum residual stocking. In the analysis of damage from Hurricane Hugo, I used the three broad categories of damage to trees in different ways to estimate lower- and upper-bound values for timberland area by condition class and treatment opportunity. For example, in South Carolina the estimate of timberland area needing regeneration because of Hurricane Hugo ranged from 328,000 to 623,000 hectares depending on which live tree damages were used to discount existing stocking in the stand (Sheffield and Thompson, 1992). RESULTS AVAILABLE FROM PERMANENT PLOTS Damage inventories utilizing permanent plot remeasurement can provide a wealth of information to the resource manager or analyst. The classifications and resource items than can be displayed depend in part on the classifications that have been made prior to the damage event. Forest type and age, for example, are sometimes difficult to assess after severe damage or salvage has occurred. The following is a brief summary of the kinds of results that were available in the Hurricane Hugo damage assessment. (1) AREAL EXTENT OF DAMAGE. Timberland area damaged can be estimated and displayed for political or administrative units, ownership categories, forest types, or any other stand classification available from the existing inventory. Information about the impact on the age structure of the forest is often needed to evaluate timber supply impacts. Figure 2 shows how the age structure of natural pine stands in South Carolina was affected by Hugo. Maps showing the geographic extent and distribution of forest damage are also easily produced ( Figure 3 ). If the coordinates of each permanent sampling point are known, a geographic information system can overlay information about stand damage severity with other layers. In Figure 3, State and county boundaries and the path of the hurricane's eye are shown along with damage. Inventory results such as this map are similar to products developed from satellite imagery. Ground-based inventories, however, can provide more quantitative estimates of timber volumes associated with conditions on the ground. (2) INVENTORY UPDATES. The remeasurement of all merchantable-size trees facilitates accurate estimates of current inventory volume, change since the previous sample measurement, and volume loss due to the damaging event. Figure 4 summarizes volume change for softwood species in South Carolina. In this example, softwood volume totaled 136 million cubic meters in 1986 and had dropped to 107 million cubic meters in 1990. Mortality and timber salvage due to hurricane damage accounted for 29 million cubic meters. Inventory updates such as this can be summarized by ownership class, stand condition, political or geographic unit, species, or tree size. (3) DAMAGE SEVERITY IN LIVE TREES. Estimates of the prevalence of damages to the current inventory of live trees are needed to predict future mortality. If substantial numbers of live trees are classed as severely damaged (class 1), mortality is likely to remain high for a long time ( Figure 5 ). In this example, 68 percent of the hardwood volume in South Carolina was classed as healthy and 32 percent damaged. About 37 percent of the damaged trees were classed as severely damaged (class 1). If these trees die, an additional reduction in inventory volume of around 12 percent can be anticipated. (4) ESTIMATES OF TREATMENT OPPORTUNITIES. For long- and short-term planning, estimates of stand treatment needs for regional areas are essential after catastrophic events such as hurricanes. The classifications described earlier help in estimation of the forest areas needing regeneration, salvage, and other treatments. PERMANENT PLOT DAMAGE ASSESSMENT -- ADVANTAGES AND DISADVANTAGES Advantages A network of permanent ground samples offers five obvious advantages over other methods such as remote sensing for forest damage assessment: (1) Forest conditions that existed before the catastrophic event are known. (2) Quantitative estimates of volume loss and damage are direct outputs from the system, and can be provided for critical subsets of the overall resource. (3) Up-to-date estimates of inventory volumes are provided. (4) Ground-based inventories set the stage for subsequent remeasurements to estimate recovery. (5) Ground-based inventories provide precise values needed to validate the rough estimates that can be made with remote sensing data. Disadvantages (1) Depending on the inventory objectives, ground-based activity can be more costly than other inventory methods. (2) Results may take longer to produce because field visits are so time consuming. (3) A satisfactory network of permanent plots for producing reliable results may not exist. (4) Large numbers of samples may be needed. LITERATURE CITED Powell, Mark D., Peter P. Dodge and Michael L. Black. 1991. The landfall of Hurricane Hugo in the Carolinas: surface wind distribution. Weather and Forecasting 6:379-399. Sheffield, Raymond M. and Michael T. Thompson. 1992. Hurricane Hugo: effects on South Carolina's forest resource. Res. Pap. SE-284, Asheville, NC: U.S. Department of Agriculture, Forest Service, Southeastern Forest Experiment Station. 51 p. United States Department of Agriculture, Forest Service. 1992. Forest Service resource inventories: an overview. Washington, DC. 39 p.