Forest Inventory And Analysis Data Research Articles

Visualizing forest landscapes using public data sources

Three-dimensional (3-D) visualizations of forest landscapes are quantitative ecological information-based techniques that can be used to visualize forest structure, dynamics, landscape transformations and regional plans. Visualizing forests and landscapes with limited ground observations are often the primary challenge for quality animations. Conducting stand-level field surveys over a large forest area is time consuming, labor-intensive and expensive. An alternative is to use existing public datasets. We used the Forest Inventory and Analysis (FIA) database and other existing vegetation data, classified Landsat TM imagery, measurements of tree architecture and Forest Vegetation Simulator (FVS) to visualize a forest landscape in the Washburn District of the Chequamegon National Forest (CNF) in northern Wisconsin to generate useful information for resource management at multiple scales. Realistic images for tree species were designed using the Tree Professional 5 software package. Empirical models were developed to calculate necessary information such as tree height and stand density from DBH, basal area and species composition. The 3-D visualizations were developed at stand and landscape levels within Visual Nature Studio 2.01. Different perspectives of the forest and landscape were visualized through zooming, variable latitudes and flying through. Potential applications of these animations are discussed within a context of alternative management of forests and landscapes (e.g., fire and harvesting), public education and decision-making processes. Results from our study demonstrate that public datasets are suitable for visualizing the dynamics of forests and landscapes, although precisely visualizing forest history is still challenging. It is appropriate to use FIA data for stand level visualization and existing vegetation data for landscape scale visualization. With these different public data sources, forests can be visualized at levels varying from a single stand to the landscape/regional level.

Some Theory for the Application of the Moving Average Estimator in Forest Surveys

Abstract The Forest Inventory and Analysis (FIA) and National Resources Inventory programs are two examples of large-scale survey programs that are in the process of moving from periodic surveys, where data for an area are collected once every 10–15 years, to annual surveys, where a smallersample of the data is collected every year. Each yearly sample can be referred to as a panel, and yearly estimates can be made with the data in each panel. With the change from periodic to annual surveys, there has been a substantial amount of discussion related to combining the data fromthe individual panels into a single estimator with an acceptably small mean square error. Numerous methods, each with certain advantages and disadvantages, have been proposed and tested. Testing with FIA data suggests that, in terms of mean square error, the equally weighted moving averageestimator is often as good as or better than other estimators when making estimates for individual counties or small groups of counties. This favorable performance is due to the relatively small sample sizes in these situations. One area of research that has not been addressed is the performanceof the moving average estimator when the sampling design deviates from its original protocol. Factors that cannot be controlled, such as weather, budgets, and politics, can alter the original sampling design. The purpose of this article is to propose and derive estimators that accommodatethese factors and give some guidelines on how to handle these situations when they occur. FOR. SCI. 50(5):672–681.

Vulnerability of Mid-Atlantic forested watersheds to timber harvest disturbance.

Forested watersheds of the Mid-Atlantic Region are an important economic resource. They are also critical for maintaining water quality, sustaining important ecological services, and providing habitat to many animal and plant species of conservation concern. These forests are vulnerable to disturbance and fragmentation from changing patterns of land use in the Mid-Atlantic Region, and from harvests of commercially mature and relatively inexpensive timber. The U.S. Department of Agriculture Forest Service (USDA-FS) Forest Inventory and Analysis (FIA) compiles data on forest condition by state and county. We have transformed these FIA data to a U.S. Geological Survey (USGS) 6-digithdrologic unit code (HUC 6) watershed base, and projected trends in timber growth, inventory, and harvest to 2025 using a timber economics forecasting model (SRTS). We consider forest sustainability from the perspective of timber production, and from the perspective of landscape stability important to conservation values. Simulation data is combined with FIA planted pine acreage data to form a more complete picture of forest extent, composition, and silvicultural practice. Early recognition of prevailing economic trends which encourage the fragmentation of mature forests due to increasing timber harvests may provide managers and policy makers with a planning tool to mitigate undesirable impacts.

Estimates of Down Woody Materials in Eastern US Forests

Down woody materials (DWMs) are an important part of forest ecosystems for wildlife habitat, carbon storage, structural diversity, wildfire hazard, and other large-scale ecosystem processes. To better manage forests for DWMs, available and easily accessible data on DWM components are needed. We examined data on DWMs, collected in 2001 by the US Department of Agriculture (USDA) Forest Inventory and Analysis (FIA) program on some plots in several states. We compiled DWM data from 778 plots to compute biomass for the following components: coarse woody material, fine woody material (three size classes), litter, duff, and shrub/herb cover. We developed regression equations to predict DWM components for extension to FIA’s more intensive plot network. Seven regression equations were applied to the FIA data to create maps of DWM biomass. As a first attempt to summarize FIA DWM measurements and extend them to plots without these data, our model produces reasonable results except possibly for duff and litter.

Fuzzy Classification of Ecological Habitats from FIA Data

Abstract Fuzzy c-mean classification (FCM), multi-layer perceptron neural network (MLP), fuzzy ARTMAP, and linear discriminant analysis (LDA) were used to classify Forest Inventory and Analysis (FIA) plots into six ecological habitats in the U.S. Northeast. Among the four classifiers, both FCM and MLP produced “soft” classifications based on fuzzy membership values. In contrast, fuzzy ARTMAP and LDA generated only “hard” classifications in which a plot is assigned to the dominant habitat class, based on binary logic, without considering any coexisting classes. The error matrix and several accuracy indices were calculated to assess the classification accuracy of the methods and to test the differences among the four classifiers. The classification accuracies of FCM and MLP were 98% and 97%, respectively, for overall classification, while fuzzy ARTMAP and LDA had an overall classification accuracy of 92% and 85%, respectively. The χ2 test showed that there was no significant difference between the FCM and MLP methods in classification accuracy. But they were significantly better than both fuzzy ARTMAP and LDA methods at a significant level of α = 0.05. This study showed that the fuzzy classifiers such as FCM and MLP were preferable in the classification of ecological habitats using FIA plots, especially for the classification of ambiguous plots with mixed overstory and understory species compositions. FOR. SCI. 50(1):117–127.

03/02221 Preparation of fullerenes from three high rank coals: Effect of mineral matters: Qiu, J. et al. Huagong Xuebao, 2002, 53, (11), 1117–1121. (In Chinese)

Outbreak of the emerging infectious disease sudden oak death continues to threaten California and Oregon forests following introduction of the exotic plant pathogen Phytophthora ramorum. Identifying areas at risk and forecasting changes in forest carbon following disease outbreak requires an understanding of the geographical distribution of host populations, which is unknown. In this study, we quantify and map the population density and carbon contents of five key host species for P. ramorum in California and Oregon, including four hosts killed by the pathogen (Notholithocarpus densiflorus, Quercus agrifolia, Quercus kelloggii and Quercus chrysolepis) and the foliar host Umbellularia californica which supports high sporulation rates. We integrate multiple sources of vegetation data, assembled from sparsely distributed (regional-scale) forest inventory and analysis (FIA) plots and more densely distributed (landscape-scale) plots for monitoring sudden oak death, and develop spatial prediction models based on correlation with environmental variables and spatial dependencies in host abundance. We estimate that 1.8 billion N. densiflorus trees (68 Tg C) and 2.6 billion Quercus host trees (227 Tg C) occur across 3.9 and 17.7 million ha of their respective habitat. A total of 436 million U. californica trees (14 Tg C) occur across 4.2 million ha which frequently overlap with Quercus and N. densiflorus host populations. Combination of landscape-scale data with FIA data resulted in more accurate estimation of host populations and their carbon contents. Forests of northern California and southwest Oregon have the highest concentration of the most susceptible hosts along with climatic conditions that favor pathogen spread. This study represents the first spatially-explicit estimate of P. ramorum host populations and their carbon contents which exceed previously published estimates. Our results will inform landscape- to regional-scale models of disease dynamics and guide management decisions regarding ecosystem impacts including risk of C release following widespread tree mortality.

Comparison of Neural Networks and Statistical Methods in Classification of Ecological Habitats Using FIA Data

Abstract Two artificial neural networks (ANN) and three traditional statistical classification methods are used to classify Forest Inventory and Analysis (FIA) plots into six ecological habitats in the U.S. Northeast. Four variables (overstory and understory species composition, hardwood basal area percentage, and current FIA forest type) are identified from a list of available stand variables as the most important discriminating variables for habitat classification. The error matrix and accuracy indices are used to assess the classification accuracy of the models and to test the differences between the five classifiers. The ANN models (MLP and RBF) are superior to the traditional statistical methods such as linear discriminant analysis and minimum-distance classification. The classification accuracy of the ANN models is 90% or higher for overall classification, and exceeds 92% in five of the six habitat categories. The K-Nearest Neighbor (KNN) method classifies the six ecological habitats as accurately as the two neural network models. This study shows that the ANN models and KNN method have a great potential for the classification of ecological habitats using FIA data, due to their flexibility of modeling algorithms and robustness to the problems in FIA data such as non-Gaussian distributions, nonlinear relationships, outliers and noise in the data. For. Sci. 49(4):619–631.

Modeling Forest Fire Probabilities in the South Central United States Using FIA Data

Abstract Factors influencing the probability of fire occurrence in the south central United States were investigated using a geographic information system (GIS) and a multinomial logit model. Forest Inventory and Analysis (FIA) data at the plot level were merged with census data at the census-tract level to create a data set containing demographic, geographic, and timber-related characteristics. A multinomial logit model was employed to estimate the relationships between plot characteristics and the probability of wildfires, prescribed fires, and fires of unknown origins. Wildfires occurred more frequently on public forests than industrial and nonindustrial private forests (NIPFs). The probability of wildfire increased with proximity to urban areas and “built-up” areas of 4 ha or more in size. Wildfires occurred more frequently in younger stands and in pine and mixed pine-hardwood types than in hardwood types. Prescribed fires occurred more frequently on public and industrial forests than on NIPFs. The probability of prescribed fires increased with proximity to roads, urban areas, built-up areas of 4 ha or more, and on flatter terrain, but was inversely related to population density. Fire was prescribed less frequently for pole-sized stands than sawtimber size stands and more frequently for pine and mixed pine-hardwood types than for hardwood types. Education levels and median household incomes of the surrounding census tract had no significant effects on the probability of any type of fire. South. J. Appl. For. 27(1):11–17.

Using biological criteria and indicators to address forest inventory data at the state level

The Santiago Declaration identified seven criteria and 67 indicators for assessing the conservation and sustainable management of temperate and boreal forests. Data collected by the Forest Inventory and Analysis (FIA) program of the US Department of Agriculture Forest Service can be used to directly address at least 11 biological indicators. The FIA program has conducted periodic inventories of forestland for several decades. These inventories provide statistical estimates of forest area, timber volume, growth, removals, and mortality. Recent legislation has mandated that the Forest Service inventory the forestland of the US on an annual basis measuring the entire set of national sample plots over a 5-year period. This paper identifies which criteria and indicators can be addressed by FIA data; the scale at which it may be appropriate to use these data; and how recent changes will impact the Forest Service’s ability to provide information needed to address these indicators. Data from the two most recent inventories of Iowa (1974 and 1990) and Minnesota (1977 and 1990) are used to show how criteria and indicators relate to trends in forest composition and extent, timber resource utilization, and the population size of native and non-native trees. These two states have distinctly different forest resources, ranging from sparse to dense, and provide a good test of the effectiveness of using forest inventory data to provide criteria and indicator information at the state level. Analysis of the data for the last two inventories of Iowa and Minnesota reveals that the area of timberland has increased by 34 and 8%, respectively, while growing-stock volumes have increased by 47 and 23%. Volumes of most native species increased over the period, especially for pioneer species such as Juniperus virginiana L. [ Little (1979). Agricultural Handbook No. 541, USDA Forest Service, Rocky Mountain Forest and Range Experiment Station, Ft. Collins, CO] (eastern red cedar). A notable exception was the 52% decline of Ulmus americana L. (American elm) due to the spread of Dutch elm disease. The number of non-native species also increased. In Iowa, the estimated number of live Ulmus pumila L. (Siberian elm) trees, a non-native species, went from 0 in 1974 to 675×10 3 in 1990. During the 1990 inventory of Minnesota another non-native species, Ailanthus altissima (Mill.) Swingle (tree of heaven), was sampled for the first time.

Evaluating Errors in a Digital Vegetation Map with Forest Inventory Data and Accuracy Assessment Using Fuzzy Sets

In large‐area mapping projects, existing reference data, often collected for a different purpose, are increasingly being used for map accuracy assessment. Multi‐attribute digital vegetation maps have been developed for all National Forest lands in California (8.1 million ha). We developed decision rules that could be applied to quantitative Forest Inventory and Analysis (FIA) plot data in order to score the fuzzy membership of plot locations in all possible map classes. We compare the accuracy of the vegetation map attributes estimated using this method to accuracy estimated from fuzzy class membership scores assigned by experts (inventory crews) during field work. Accuracy of the vegetation life form attribute was estimated to be higher when expert label assignments were used as reference data (76–87%), instead of FIA plot data (62–79%). This suggests that automated decision rules applied to detailed data from FIA plots, which have smaller area than map polygons, may systematically underestimate map accuracy. However, assignment of the actual map labels to FIA plot locations by inventory crews appears to be a robust method for using the FIA data for accuracy assessment.

BIOMASS AND NPP ESTIMATION FOR THE MID-ATLANTIC REGION (USA) USING PLOT-LEVEL FOREST INVENTORY DATA

As interest grows in quantification of global carbon cycles, process model predictions of forest biomass and net primary production (NPP) are being developed at an accelerating rate. Such models can provide useful predictions at large scales, but it has been difficult to evaluate their performance. Using the network of plots comprising the comprehensive and spatially extensive Forest Inventory and Analysis (FIA) data set collected and maintained by the USDA Forest Service, we applied methods typically used in field measurements to develop estimates of forest biomass and NPP for the mid-Atlantic region of the United States at a scale appropriate for comparison with model predictions. Plot-level and tree-level forest inventory data from a subset of plots were used together with species-specific biomass regression equations to calculate maximum current biomass and NPP values for the mid-Atlantic region. Estimates at the plot level were aggregated by forest type and to the 0.5° × 0.5° scale for analysis and comparison with process model predictions. Maximum current forest biomass averaged 248 and 200 Mg·ha−1·yr−1 in hardwood and softwood forest types, respectively; wood biomass increment averaged 559 and 460 g·m−2·yr−1 in hardwood and softwood forest types, respectively. Aggregated to the 0.5° × 0.5° scale, forest biomass ranged from 101 to 326 Mg/ha, while wood biomass increment ranged from 254 to 1050 g·m−2·yr−1. Biomass and NPP estimates for closed-canopy forests from this study were consistent with values reported in the literature but were as much as 50% lower than values reported for old-growth stands. NPP predictions from three process models were fairly consistent with the FIA-based estimates, but model predictions of biomass were higher than estimates from FIA data for the region. By describing upper and lower bounds on reasonable biomass and NPP values for closed-canopy forests, these FIA-derived estimates provide a foundation for model comparison and continued model development.

Diameter Growth Models Using Minnesota Forest Inventory and Analysis Data

Abstract The Forest Inventory and Analysis (FIA) program of the USDA Forest Service North Central Research Station (NCRS) has begun replacing the 12- to 13-yr periodic inventory cycles for the states in the North Central region with annual inventories featuring measurement of approximately 20% of all plots in each of the 11 states each year. State reports on summaries of the forest resources will be produced every 5 yr. As a method of updating information on plots not visited in the current year, NCRS is developing nonlinear, individual-tree, distance-independent annual diameter growth models for species groups. The models, formulated as the product of an average diameter growth component and a modifier component, were calibrated on Minnesota FIA data from stands that were generally undisturbed, of mixed ages and of mixed species. The dependent variable is annual diameter growth. The independent variables include crown ratio, crown class, stand basal area, stand basal area larger than the subject tree, physiographic class, and latitude and longitude of plot locations. The model predictions at both the individual-tree level and plot level have negligible bias, and the models may be easily recalibrated to include new data sets obtained from the annual inventories. For. Sci. 47(30):301–310.

FIAMODEL: A New Link for Geographic Analyses of Inventory Data

Abstract The USDA Forest Service Forest Inventory and Analysis (FIA) data are widely used throughout the United States for analyses of forest status and trends, landscape-level forest composition, and other forest characteristics. A new software product, FIAMODEL, is available for analyzing FIA data within the ArcView® (ESRI, Inc.) geographic information system. The software allows linkages with other natural resource information, such as watershed boundaries or soil maps, in a format widely used by natural resource organizations, which allows organizations to analyze FIA data in a spatial context with little new investment in hardware or training.

Evaluating FIA Forest Inventory Data for Monitoring Mexican Spotted Owl Habitat: Gila National Forest Example

Abstract The habitat of the Mexican spotted owl must be monitored because of the owl's “threatened” status under the Endangered Species Act of 1973. A possible data source for habitat monitoring is the network of permanent plots maintained by Forest Inventory and Analysis (FIA), USDA Forest Service, Rocky Mountain Research Station. This article demonstrates how FIA data might be used to calculate stand density measures for monitoring owl habitat. Test results for New Mexico's Gila National Forest showed 26% of the mixed conifer and 35% of the pine-oak as suitable owl habitat for a forest structure scenario. Ninety-five percent confidence intervals for these estimates were from ±20 to ±35% for the mixed-conifer habitat area; for pine-oak they were ±50 to ±73%. Although results are encouraging for using FIA data to monitor owl habitat, critical linkage needs to be established between the owl's home-range requirements and the stand-density metrics tested. West. J. Appl. For. 15(4):195-199.

Using a Mensuration Approach with FIA Vegetation Plot Data to Assess the Accuracy of Tree Size and Crown Closure Classes in a Vegetation Map of Northeastern California

The accuracy of the Lassen-Modoc vegetation map was assessed using data from a grid of permanent Forest Inventory and Analysis (FIA) plots in conjunction with a fuzzy logic approach. This paper presents the results for the Modoc National Forest. The main foci of this paper are the methods and results for assessing the accuracy of tree size and crown closure classes using a mensurational approach based on calculations made from the plot data. This approach proved very useful in conjunction with fuzzy accuracy methods. Higher error rates reported for crown closure classes may be due to a variety of factors including error associated with crown closure modeling methods, error associated with crown closure calculations for a plot, and the variability of crown closure within a plot. Using the FIA data has proven to represent a cost-effective means of generating accuracy assessment information based on a probability sample.

Relative stocking index: a proposed index of site quality

Site index is difficult to implement and interpret in multispecies, multiple-aged stands, and its relationship to site factors is obscure. Using data from the USDA forest inventory and analysis (FIA) for the Lake States, we developed log-log relationships between mean tree size and stand density for five cover types. Fits were good, with r2 from 0.96 to 0.98 and slopes from −0.948 to −0.995. We define an alternative index of site quality, the relative stocking index (RSI), as the ratio of a stand's measured density to that predicted using the log–log relationship for its cover type (the norm). We divided the range of RSI into three classes for each type (<0.9 of norm, >0.91 but <1.1 of norm, and >1.1 of norm). Based on analyses of the 1977 and 1990 FIA data from Minnesota, class assignments for individual stands remained constant over that 13-year period. Relationships between site factors and either RSI classes or analogous classes based on site index were examined in a subset of 169 stands. Temperature, precipitation, silt content of surface soil, and calculated annual water deficit all differed significantly among RSI classes, but not among site-index classes. The RSI is easy to apply, robust (resistant to change), and related to site factors. It merits additional examination as an index of site quality, especially in heterogenous stands.