Brian Lee, Department of Landscape Architecture
Identifying the Extent of Kentucky's Forest Fragmentation
Forest fragmentation refers to the severance of tracts of forested land as a result of harvesting practices and clearing for agricultural lands, as well as the development of human habitat, roadway construction, and other human-influenced landscape modification such as resource extraction, and is an important consideration in a comprehensive landscape conservation portfolio. Habitat loss due to forest fragmentation is often cited as one of the most important reasons for the decline of biological diversity around the world as well as in Kentucky.
The original intent of the Forest Fragmentation project was twofold. The first objective was to develop a framework for automating a process to classify forest fragmentation and to quantify (size) interior forest blocks of greater then 1000 acres in Kentucky using the Environmental Systems Research Institute's (ESRI) ArcGIS ModelBuilder. The second objective was to apply that process to three classified datasets at two temporal points to characterize potential change: 1992 Kentucky GAP Analysis data (performed using Landsat Thematic Mapper (TM) imagery); 1992 National Land Cover Data; and 2001 National Land Cover Data.
The spatial approach used to conceptualize the model was similar to the way a sculptor might create a subtractive clay block sculpture. The form is a result of removing material from the whole, in this example, from forested lands. A moving window analysis was used to characterize the forested cells as patch, transitional, edge, or interior forest. Three maps (one for each dataset) were generated to illustrate the extent of forest fragmentation. Two charts illustrate the quantitative characteristics of the forests from the three datasets: amount of fragmentation by type (Patch, Transitional, Edge, and Interior), and detailed quantitative characteristics of Interior Forest. It is important to recognize that there is known error in each of the land cover/land use datasets according to the data originators; however it is believed that these issues would amount to small errors and not substantially change the results.
In 2008, these geospatial models were applied to the Kentucky Land Cover Change Detection 2001/2005 data to update Kentucky's forest characterization with the most recent classified data publicly available. A conditional statement was used to combine the 2005 data with the KYNLCD 2001 where change was not detected. As with the previous three datasets, large forest blocks (>1000 acres) were identified and forest fragmentation was classified, and in addition, forest blocks greater than 5,000 acres were queried as well as forest blocks greater than 20,000 acres were queried to show spatial distribution of the largest of the large forest blocks. When quantitatively comparing the 2005 forest data to the 2001 forest data, there were reductions in all of the basic descriptive statistics, and the forest fragmentation characterization showed an increase in the amount of patch and transitional forest while showing a decrease in edge and interior forest.
Determining Landscape Areas for Targeted Reforestation Efforts
In 2010, work is underway to utilize this forest block data in combination with other publicly available data in a modified ELESA approach for forest land (FELESA) to identify critical locations in Kentucky for focusing reforestation efforts. This project will use historical land cover data and other data to predict the probability of changes in the forested landscape for the next 100 years under a variety of scenarios. The study area chosen to depict potential alternative forest futures is the Eastern Kentucky Coal Field. This natural resource management project also has land-use planning implications.
The FELESA process evaluates land consistently based on stakeholder input to model guidelines. The system is flexible so that suitability analysis can be based on a variety of stakeholder objectives such as water quality protection, biomass/carbon accumulation, forest connectivity, interior forest expansion, or physical access to name a few. For a given land area, factors are rated, weighted, and combined, resulting in a single numeric reforestation suitability score. The score indicates the relative land value in relation to other forested and non- forested land for reforestation activities.
In a second suitability approach, another software application, Marxan, will be used to identify reforestation opportunity areas. Marxan is the most widely used landscape conservation planning software (Ball et al., 2009; Watts et al., 2009). It provides decision support to a range of conservation planning problems including developing multiple-use plans for natural resource management. Marxan is efficient and repeatable providing a number of options and encouraging stakeholder participation. These features provide users with decision support to achieve an efficient allocation of resources across a range of different uses.
Once the most suitable reforestation areas are identified, LANDIS will be used to simulate forest change. LANDIS is a spatially explicit landscape model designed to simulate forest landscape change over large spatial and temporal scales (Mladenoff et al., 1996; He et al., 1999). LANDIS has been used to simulate the dynamics of forest succession, seed dispersal, wind, fire, biological disturbance (insects and diseases), harvesting, and fuel management. Differing from most landscape models, LANDIS simulates multiple landscape processes in combination with the simulation of succession dynamics. The forest simulations will be quantitatively evaluated and subsequently evaluated by the experts involved with the reforestation suitability model.
This project will utilize and build upon precision resource management capacity for the state in
four specific ways:
1 . Capitalize on the previous remote sensing and other geospatial data investments that are available from the Kentucky Geography Network.
2 . Apply land use suitability analysis relying on an Analytic Hierarchy Process in the context of a geospatial Delphi framework for identifying locations for the planting of 175 million trees in eastern Kentucky.
3 . Utilize the capabilities of one of the most widely used software applications that can identify potential conservation reserve locations (Marxan) to compare to the locations identified using a geospatial Delphi framework.
4 . Once suitable reforestation locations are identified, multiple simulated forest change scenarios over large spatial and temporal scales will be performed using LANDIS.
|• 2005 Kentucky Forest Characterization|
|• A Process to Identify Kentucky's Large Forest Blocks|
|• A Process to Identify Kentucky's Forest Fragmentation|