Assessing The Impact Of Urbanization On Timberland Availability In Southeastern Louisiana by

This study illustrates how remotely-sensed data and GIS can be utilized to evaluate the impacts of urbanization and land use on timberland availability. GIS coverages were developed for land use, population density, public or reserved lands, site productivity, and timber volume. Econometric models then were developed to estimate changes in land use and the resulting impacts on timber availability. The study was conducted in St. Tammany Parish, Louisiana, located northeast of New Orleans. The parish has experienced the fastest population growth of any area in the state over the past decade and has served historically as a major timber production area for southeastern Louisiana. INTRODUCTION Projecting the future availability of timber in the United States has been problematic historically, primarily due to fluctuating markets for timber, changing landowner objectives, and a number of data and analytical problems. During the past two decades, however, social and institutional factors have emerged as important determinants of timber availability nationwide. The increased development and urbanization of areas that traditionally have been important sources of raw materials is one of the more obvious social factors. The impact has involved not only the obvious effects of a reduced forest land base, but it also has resulted in increased public concern regarding management activities and often new regulations limiting timber harvesting. Most often, these new regulations are not federal policies. Currently, however, policy makers, regulators, and landowners have little information on how changes in land use or regulations could affect resource availability from southern forests. A Geographical Information System (GIS) allows for a graphical depiction of the spatial and temporal dimensions of the biological, economic, legal, and social attributes related to land use change and resource availability. A GIS facilitates the assessment of the relationships between land use, regulations, and the provisions of a myriad of goods and services from southern forests. Moreover, it also allows policy makers and analysts a means of evaluating the effects of future land use scenarios on forest distribution. The primary purpose of this study was to develop the GIS capability to examine the relationships between land use patterns and timber availability in St. Tammany Parish. Three specific objectives were addressed by the study: 1. Develop a GIS that incorporates spatial data on the biological, economic, legal/political, physical, and social attributes of the area for analysis. 2. Assess the temporal variation in land use patterns, forest cover distribution, and timber availability. 3. Project changes in timber availability under various future land use scenarios. Associate Professor, Department of Forestry, Wildlife and Fisheries, The University of Tennessee, P.O. Box 1071, Knoxville, TN 37901-1071. Cartographic Engineer, Senior Cartographic Engineer, and Freeport-McMoran Professor and Director, Tulane University Medical Center, Entergy Spatial Analysis Research Laboratory, 1430 Tulane Avenue SL1, New Orleans, LA 70112-2699. Funding for this research was provided by the USDA Forest Service Southern Research Station, Law and Economics Research Work Unit, New Orleans, Louisiana. METHODS Accomplishing the objectives required initially obtaining the data and developing a GIS for the study site. The GIS then was utilized to assess the impacts on timber availability brought about by urbanization. The steps involved in developing the GIS and utilizing the system for the evaluations are described below. Study Site St. Tammany Parish is located northeast of New Orleans and adjacent to Orleans Parish. In 1974, the parish was a sparsely populated area (63,700 individuals). By 1992, the population had more than doubled to 156,200 individuals. Due primarily to migration of workers from New Orleans to outlying communities such as Slidell and Mandeville in St. Tammany, the parish’s economic base was transformed and land use patterns altered dramatically (U.S. Department of Commerce 1994). The increased urbanization has resulted in tremendous land use conflicts between residential and commercial, agricultural, and forest interests. GIS Development Geographic data sets of land cover, census blocks and block groups, primary and secondary roads, preserved lands, major access points to New Orleans. All geographic data sets were projected to Universal Transverse Mecator (UTM) zone 16. Additional databases were created in the census block coverage to facilitate the inclusion of demographic information from other data sources. Land cover data were obtained from two sources: Stennis Space Center/Environmental Enterprises, USA, Inc. (Environmental Enterprises, USA, Inc. 1995); and the United States Geological Survey (USGS) National Wetlands Research Center (United States Geological Survey National Wetlands Research Center 1997). Land cover classification developed at Stennis was the result of a commercial space technology project by NASA and the private firm Environmental Enterprises, USA, Inc. of Slidell, Louisiana. Stennis used the cluster modules of Erdas 7.2 with 1981 multispectral satellite (MSS) image to categorize land cover into 9 groups. Hydric soil groups digitized from the general soils maps of St. Tammany Parish aided in the determination of wetland (cypress/wet deciduous forest, high density marsh, low density marsh) and riparian (deciduous forests) cover types. The resulting land cover image created had 80 meter resolution. Land cover classifications were field verified by Stennis personnel. The USGS National Wetlands Research Center developed land cover for St. Tammany Parish as part of the National Geographic Assessment Program (GAP). 1993 Thematic Mapper (TM) satellite scenes with 30 meter resolution were categorized into 21 groups by the USGSThe USGS also incorporated information from the 1989 National Wetlands Inventory to aid with land cover classification. Hydric soils were not used. ESARL personnel performed extensive field verification of the land cover at 190 randomly selected locations throughout St. Tammany Parish. Both the Stennis Space Center and the USGS images were imported into Grid (ArcInfo’s raster module). The USGS image was filtered to remove small areas of land cover consisting of only a few pixels (specks) resampled (with the specks removed) to provide a land cover grid with 80 meter resolution. The resulting USGS grid was then spatially comparable to the Stennis Space Center grid. Next, the land cover of both grids was reclassified into 7 groups: pine forests, deciduous forests, mixed forests, wetlands, agriculture/grasses/crops, urban/inert/barren, and water. Most of the generalizations from the USGS categories to this new classification scheme are obvious. However, the field verification of the USGS land cover classification resulted in several needed changes. Field verification indicated that both young pine plantations and thinned pine forests appear as a mixed land cover type on the satellite images. In the reclassification of the USGS land cover, the mixed as well as evergreen shrub/scrub cover types were included in the pine forest category. Pine forests better reflects the land use than mixed forest, and evergreen or mixed shrub/scrub. With both land cover grids now spatially compatible and with similar land cover, a change detection analysis was performed. Areas that were pine or mixed forested in 1981 and remained pine or mixed forests in 1993, and areas that were pine or mixed forests in 1981 but were urban/inert/barren in 1993 were identified. The hydric soils mask used by Stennis to determine wetland areas was inherently applied in the change detection analysis (i.e., no pine forest was located on hydric soils). Two covers resulted from the change detection analysis: a point cover containing CHANGE, our dependent variable, and a polygon cover containing CAREA, the area of contiguous pine/mixed forest land. The point cover contained the centroids of all grid cells that either remained pine or mixed forests from 1981 until 1993, or that changed from pine or mixed forests to urban/inert/barren during that time. These points were used to define the cover containing the dependent variable CHANGE (1 if a point changed to urban occurred, 0 if it remained forested). We randomly selected 12,751 points for use in the land use change modeling effort. A polygon cover, CAREA, was created from the grid that contained our change detection results. CAREA contained the contiguous area of pine/mixed forest cover. We associated CAREA with CHANGE estimated by intersecting the covers. Demographic coverages were developed from U.S. Bureau of the Census (1993) data. Variables were developed for population and density, income, education, and type and location of employment. The primary and secondary roads coverage contains the high speed highways (interstates, US routes, and some state routes) in St. Tammany Parish. The coverage was derived from the USGS I:100000 digital line graph files of primary and secondary roads for Louisiana (LAGIC 1998, USGS 1989). This coverage was used to develop variables representing the Euclidean distance to the nearest major road. Many St. Tammany Parish residents commute to New Orleans to work. Lake Pontchartrain separates the parish from New Orleans and must be crossed by one of three bridges by St. Tammany commuters. Two bridges (U.S. 11 and Interstate 10) connect Slidell to New Orleans (these are collectively referred to as the twin spans) and the third connects Mandeville to New Orleans (Causeway). We marked the entrance of the twin spans and the causeway bridge to form a coverage. Model Development The probability that a particular parcel of pine or mixed forest would be developed was estimated using logistical regression. The dependent variable was binary, with zero (0) indicating that the parcel of land was pine or mixed forest in 1981, and remained pine or mixed forest in 1993. A one (1) indicates that the parcel of land was pine or mixed forest in 1981 but was developed (urban/inert/barren) in 1993. Based on prior research (Munn and Evans 1998. Alig et al. 1988, Alig 1986), the model that was evaluated for the analysis was: CHANGEi = ƒ(POPDENi, ROADi, INTERi, ACCESSi, CAREAi) (1)