Chris J. Cieszewski

K Nearest Neighbor Method for Forest Inventory Using Remote Sensing Data

The K nearest neighbor (KNN) method of image analysis is practical, relatively easy to implement, and is becoming one of the most popular methods for conducting forest inventory using remote sensing data. The KNN is often named K nearest neighbor classifier when it is used for classifying categorical variables, while KNN is called K nearest neighbor regression when it is applied for predicting noncategorical variables. As an instance-based estimation method, KNN has two problems: the selection of K values and computation cost. We address the problems of K selection by applying a new approach, which is the combination of the Kolmogorov-Smirnov (KS) test and cumulative distribution function (CDF) to determine the optimal K. Our research indicates that the KS tests and CDF are much more efficient for selecting K than cross-validation and bootstrapping, which are commonly used today. We use remote sensing data reduction techniques—such as principal components analysis, layer combination, and computation of a vegetation index—to save computation cost. We also consider the theoretical and practical implications of different K values in forest inventory.

Closest Spectral Fit for Removing Clouds and Cloud Shadows

Completely cloud-free remotely sensed images are preferred, but they are not always available. Although the average cloud coverage for the entire planet is about 40 percent, the removal of clouds and cloud shadows is rarely studied. To address this problem, a closest spectral fit method is developed to replace cloud and cloud-shadow pixels with their most similar nonclouded pixel values. The objective of this paper is to illustrate the methodology of the closest spectral fit and test its performance for removing clouds and cloud shadows in images. The closest spectral fit procedures are summarized into six steps, in which two main conceptions, location-based one-to-one correspondence and spectral-based closest fit, are defined. The location-based one-to-one correspondence is applied to identify pixels with the same locations in both base image and auxiliary images. The spectral-based closest fit is applied to determine the most similar pixels in an image. Finally, this closest spectral fit approach is applied to remove cloud and cloud-shadow pixels and diagnostically checked using Landsat TM images. Additional examples using QuickBird and MODIS images also indicate the efficiency of the closest spectral fit for removing cloud pixels.

Spatial Clusters and Variability Analysis of Tree Mortality

Spatial patterns of tree mortality are important factors in analysis of forest health and sustainable forest management. We analyzed spatial cluster characteristics of forest tree mortality among hardwoods, softwoods, and all species groups between 1989 and 1997 in Georgia. We used for this purpose the 1989 and 1997 USDA Forest Service Forest Inventory and Analysis (FIA) permanent sample plot data, spatial scan statistic and semivariogram analysis methods, and geographic information system (GIS) techniques. In Georgia, forest tree mortality for softwoods and all species groups formed clusters in North Georgia in 1989. These clusters became more widespread, but the aggregation intensity of tree mortality was decreased compared with that of the 1989 clusters. Another finding is that the locations for mortality clusters of softwoods were almost the same as those of all species combined both in 1989 and in 1997. In 1989 the forest tree mortality of softwoods and all species groups agglomerated with small clusters forming in north Georgia. In 1997, these clusters became more widespread, but the magnitudes of tree mortality within them decreased compared with those of 1989 clusters. Semivariogram analysis suggests that hardwoods, softwoods, and all species combined followed a similar trend for spatial variation in that they fit a spherical model of tree mortality. However, the strengths of spatial dependence were significantly different among the hardwoods, softwoods, and all species in 1989 and 1997. The higher the agglomeration of tree mortality, the stronger the spatial dependence. Spatial analysis of tree damage indicates that agents of tree damage were also agglomerated in the state. Eight kinds of tree damage, possibly combined with other kinds of damage, directly determined the spatial patterns of forest tree mortality. Out of these eight, only vegetation competition, insects, and weather caused the formation of the north Georgia tree mortality clusters.

Synopsis of Forest Management Plans of North America

Abstract In this book, we have collected 48 case studies of forest plans from throughout North America. We believe this is the most comprehensive set of forest planning case studies gathered using a common template. We are very grateful for the effort that many people (authors, editors, and support staff) put into preparing these chapters describing their approach to forest planning, especially since they were constrained by our timeline and our common template for all of the examples. The landowners were selected through outreach to professional organizations, calls to management organizations and to forestry consultants, and contacts within our personal networks. While perhaps not a representative sample of all potential management plans, we believe that these describe the range of management plans being created in the North American region. In other words, we employed a case study approach rather than a particular statistical design to sample these examples from the larger population of all forest management organizations in the region. Therefore, it is important for one to avoid making strong and significant inferences about the larger population of all forest plans in North America. As a result, this synthesis is limited to the case studies included in the text. The properties described in this book have a tremendous range of size, from 58 acres (ac) to 17 million ac (23 hectares (ha) to 6.8 million ha). They represent a variety of ownership types such as family forests, municipal forests, city and county forests, state forests, public and private university research forests, urban forests, federally managed forests, tribal forests, and large publicly traded corporate forests. Vertically integrated companies were once very numerous in the United States, but now they are a rare organizational structure, mainly due to changes in the U.S. tax codes over the last two decades. The two examples of this organizational structure included are the Pike Lumber Company ( Chapter 4 ) and the Green Diamond Resource Company ( Chapter 35 ).

Spatially explicit biomass supply sustainability analysis for bioenergy mill siting in Georgia, USA.

Forest production sustainability is a broad and controversial subject that is frequently argued but rarely computed. Especially in the context of private forest ownership the results of erroneous assessment of forest production sustainability, such as in the case of woody biomass production for local mill operations, may result in economical losses, in lower regional employment rates, and decreased prosperity and competitiveness. We describe in this article a simulation-based quantitative approach to sustainability analysis of forest biomass production and utilization in the context of new bioenergy mill siting. The analysis is based on the best available forest inventory data and on the most up- to-date knowledge of natural resource growth and yield dynamics as modelled by various studies available in the literature. The data used includes the USDA Forest Service FIA forest survey data as well as an enhanced analysis of data indicating locations of Intensive Management Plantations (IMP) at a county level, since such information, while not publicly available, has a significant impact on biomass production expectations. Using these data, simulated according to the state-of-the-art knowledge of regional growth and yield characteristics, we determine sustainable harvest levels (SHLs) for the purpose of siting bioenergy mills for 10- to 20-year production cycles. The simulations are conducted for each individual county of Georgia for four radii of procurement areas. The derived county level information on sustainable levels of biomass production, which vary for different units of analysis, can be used as a reference for effective forest utilization planning and for mill siting.

Proceedings of the fourth annual Forest Inventory and Analysis symposium

Abstract. —We provide here a short description of theorigin, current work, and future outlook of the FiberSupply Assessment program at the D.B. WarnellSchool of Forest Resources, University of Georgia,whose work includes various analyses of FIA data.Since 1997, the program has intended to assist theimplementation of the new Southern Annual ForestInventory System through related data analyses.Currently its projects include basic problems in theoryof equations and parameter estimation and variousanalyses of ground inventories and remote sensingand GIS data. We describe some of these projects andassociated software, hardware, and information dis-semination problems and solutions. The Fiber Supply Assessment program (FSA) at the Daniel B.Warnell School of Forest Resources was initiated by DeanArnett Mace, Jr. in 1997. The establishment of this programcoincided with the beginning of the implementation of theSouthern Annual Forest Inventory Analysis System (SAFIS)and was intended to provide the school’s input into solving thevarious problems of timely and accurate fiber supply assess-ment in Georgia. Parties interested in creation of this program includedmembers of the forest product industry, and others. Theirexpectations were directed toward finding new relationshipsand revealing information in the new annual measurement dataproduced by SAFIS. The new design of the continuous annualinventory was attracting many questions about statistical accu-racy, the possibility of monitoring growth, and differencesbetween current and former periodic estimates. Notwithstanding the above, in the beginning the FSAcould not focus on the annual data analysis due to the unavail-ability of such data. Furthermore, there was also little pragmat-ic value in work on new estimators because it seemed ratherfutile to begin changing the barely conceived statistical design,which was so new that it was not even quite implemented yet.Thus, at the outset the FSA focused initially on theoreticalstudies of inventory projection equations. Subsequent effortsconcentrated on building collaborative studies with other pro-grams and exploring funding opportunities related to the gener-al mandate of the program. Presently the program iscollaborating with forest biometrics, the quantitative forestmanagement and wood quality programs, the center for forestbusiness, forest finance, forest economics, and a number of for-est product industry partners.