Progress 10/01/02 to 09/30/06
Outputs
The first phase of this study examined the correlation between parcelization and forest fragmentation. Maps depicting states of forest fragmentation were obtained for the years 1985, 1990, 1995, and 2002. The maps classified forest as one of five types: interior, perforated, edge, transitional, and patch forest. Interior forests are not fragmented whereas the remaining forest types have some form of fragmentation. In the first test, the total areas of each forest type were calculated at each map date. A parcel map was used to determine the total number of parcels that existed at each of the forest fragmentation map dates. A significant inverse correlation was found between the number of parcels and interior forest area. Significant positive correlations were found between the number of parcels and the areas of fragmented forest types. In the second test, parcels established on or after 1985 were grouped, by creation date, into three categories: 1985-1989, 1990-1994, and
1995-2002. Areas of forest types were determined for each group of parcels. Rates of change in forest type area were greatest during the time period in which parcelization was occurring. Interior forest area declined over time while fragmented forest type areas increased over time. The second phase of this study attempted to predict quantitatively how forest area and fragmentation would be affected in the future by building construction. Multi-Criteria Evaluation (MCE) was used to create suitability maps for the study area. These maps depict the suitability of a given area for development and indicate where future development is likely to occur. The Scenario360 ArcGIS extension was used to perform Buildout and TimeScope analyses for the study area. The buildout analyses placed points at all potential future building locations based on zoning regulations. TimeScope analyses were performed to assign build dates to the future building points. Build dates were assigned according to site
preference as determined from suitability maps. Square buffers were created around each building point to simulate the land clearing associated with building construction. Buffer areas were selected randomly and ranged from 3600 to 48,400 sq ft. The buffers were assigned the build dates of their corresponding building points. Buffers, with the appropriate build dates, were used to modify a land cover map to depict the predicted land cover states at several future dates. The modified land cover maps were input into a forest fragmentation model to characterize forest types in terms of degree of fragmentation. A state of forest fragmentation model was used to summarize the results of the forest fragmentation model. Forest cover and agricultural land, in the study area, are predicted to decline by 3.0% and 5.6%, respectively between 2002 and 2036. Urban and turf areas related to development are predicted to increase by 17.9%. Interior forest is predicted to decline by 27.7% between 2002
and 2036. Fragmented forest types are predicted to increase substantially. Predictions are likely to be conservative since the analysis did not account for future road construction.
Impacts
This study showed a strong correlation between parcelization and forest fragmentation which exposes parcelization as a socioeconomic factor that is contributing to forest fragmentation. Knowledge of the causal factors will assist efforts directed at minimizing future forest fragmentation. This study also developed a method for predicting future states of forest fragmentation for regions in which urban growth is a factor driving the fragmentation. The results of the analysis will be useful for guiding managers in land use decision-making. In addition, the maps generated will be useful for identifying areas in which conservation efforts may be a priority. The ArcGIS models created to perform the analyses used in this study can provide researchers with templates for use in similar applications across the country. The models also facilitate the repetition of the analyses with varying parameters. This repetition would be useful in researching the efficacy of various zoning
strategies (i.e. smart growth) aimed at minimizing future forest fragmentation.
Publications
- Civco, D.L., Hurd, J.D., Wilson, E.H., Arnold, C.L., Prisloe, S. 2002. Quantifying and describing urbanizing landscapes in the Northeast United States. Photogrammetric Engineering and Remote Sensing 68(10):1083-1090
- Civco, D.L., Hurd, J.D., Wilson, E.H., Song, M., Zhang, Z. 2002. A comparison of land use and land cover change detection methods. Proc. 2002 ASPRS Annual Convention, Washington, D.C. 12 p.
- Hurd, J.D., Wilson, E.H., Civco, D.L. 2002. Development of a forest fragmentation index to quantify the rate of forest change. Proc. 2002 ASPRS Annual Convention, Washington, D.C. 10 p.
- Parent, J. 2006. Simulating Future Forest Fragmentation in the Salmon River Watershed of Connecticut. MS Thesis, Department of Natural Resources Management and Engineering, University of Connecticut, Storrs, CT. 115 p.
- Hurd, J.D., Parent, J., Civco, D., Tyrrell, M., Butler, B. 2006. Forest Fragmentation in Connecticut: What do we know and where are we headed? Proc. 2nd Annual CT Forest Conservation Forum Program, Granby, CT. 16 p. (in press)
- Parent, J., Civco, D.L., Hurd, J.D. 2007. Simulating future forest fragmentation in Northeastern United States. Proc. ESRI User Group Meeting, San Diego, CA. 6 p. (in press)
- Parent, J., Civco, D.L., Hurd, J.D. 2007. Predicting forest fragmentation in the Salmon River Watershed in Connecticut. Proc. ASPRS Annual Convention, Tampa, FL. 12 p. (in press)
|
Progress 01/01/05 to 12/31/05
Outputs
Data collection and analysis have continued as have efforts to refine urban growth prediction models. A thorough analysis of factors believed to influence urban growth has been performed, including topographic slope and soil properties such as on-site septicability, engineering properties, agricultural production potential, wetlands status, flood zone designation, surficial materials, and erodibility. A high degree of correlation among factors was observed, thus violating the assumptions of multi-criteria evaluation (MCE). Therefore, it was decided to use soil unit types in place of slope and soil property factors. Other suitability factors analyzed included distance from roads and land cover type. To determine the relative importance of each factor's degree of influence on urban development, the percentage of each soil type class developed was calculated. Areas developed were determined from Landsat-derived land cover data. It was assumed that classes that were
proportionally more developed had a higher potential for development. It was found that soil unit types and distance from primary roads were the most important factors in determining development suitability. These two factors along with two constraints, hydrography and open space, were used to create suitability maps using MCE. Suitability map values ranged from 0 to 255 with 255 being theoretically the most suitable value and 0 being the least suitable. Several suitability maps were created by varying factor weights. For each suitability value the percentage of raster grid cells that were developed was calculated. The suitability map that had the highest positive correlation between development and suitability value was selected for use in further analysis. Several attempts were made to predict urban land cover change with GEOMOD, a land change prediction and simulation tool within the Idrisi raster GIS analysis software. Accurate predictions of land cover using GEOMOD were not
realized, and attention was directed towards 'build out' approaches to land use change. These techniques rely more on planning data, such as land parcels, existing building locations, zoning, as well as other factors and constraints. Community Viz's Scenario 360 has become the principal modeling tool to perform a build-out analysis based on a development suitability map, modified further by the aforementioned planning data. The build-out analysis populates the areas deemed suitable for development with buildings in accordance with the zoning constraints. High resolution (< 1 ft) digital orthophotographs acquired in 2004 were used to create a map of existing structures. Parcels already containing structures were excluded from the build-out analysis. In addition, the factors in our suitability map are being re-analyzed using the map of existing structures, in place of simply Landsat-derived land cover data. This will provide a better depiction of the developed landscape and will improve
the overall suitability and build-out modeling processes, leading to better estimates of future potential forest fragmentation.
Impacts
The development of a temporal dataset of parcel subdivision for a large study area is significant. These data illustrate change through time and development trends that are otherwise not apparent. The improvement of base spatial data to be used in both descriptive and predictive modeling is significant. Combination of temporal parcelization with physical and cultural geospatial data will lead to a characterization of forest fragmentation. The use of a GIS-based build-out tool (CommunityViz Scenario 360) will enable modeling of potential future complexions of the urban landscape, and as a consequence, the extent of forest loss and degree of forest fragmentation. The model, based on physical and cultural landscape data, will be incorporated into a decision support system (DSS) that will be of value to land use planners, forest managers, and decision makers.
Publications
- Civco, D.L., Chabaeva, A., Angel, S., Sheppard, S. 2005. The Urban Growth Management Initiative-Confronting the Expected Doubling of the Size in the Developing Countries in the Next Thirty Years-Methods and preliminary results. Proc. 2005 ASPRS Annual Convention, Baltimore, MD. 12 p.
- Angel, S., Sheppard, S.C., Civco, D.L., Buckley, R., Chabaeva, A., Gitlin, L., Kraley, A., Parent, J., Perlin, M. 2005. The Dynamics of Global Urban Expansion. Transport and Urban Development Department, The World Bank, Washington, D.C. 200 p.
- Song, M., Civco, D.L., Hurd, J.D. 2005. A competitive pixel-object approach for land cover classification. International Journal of Remote Sensing. 26(22):4981-4997
|
Progress 01/01/04 to 12/31/04
Outputs
The second year of this project has focused on refinement of data generation and collection (Phase I) and on the development of land use change and urban growth prediction models (Phase II). A digital elevation model (DEM) was created by way of spatial interpolation of a dense set of elevation points, originally acquired from USGS 1:24000 scale quadrangle maps. Geostatistical Kriging was used to generate a surface with 10 meter horizontal and one meter vertical resolution. To avoid curvature in the planar surface of water bodies, hydrographic features (lakes and ponds) were rasterized into the same coordinate space as the interpolated DEM, attributed with their elevation, and overlain on the DEM. Slope computed from this DEM is an improvement over that derived from NED or SRTM data. Land cover data for 1985, 1990, 1995, and 2002 were imported into an Idrisi format. These land cover data formed the basis of a linear forecast (r=0.98) of urban land extent for the years
2010, 2015, 2020, and 2025 (17855, 18531, 19206, and 19881 acres, respectively). These estimates are being used to drive urban land use scenarios for these dates using the multicriteria evaluation land use change simulation functionality of Idrisi. To accommodate GEOMODs requirement of a two-class land cover projection, land cover data were recoded into a binary data set (1=urban and 2=non-urban). Specific factors (continuous variables) deemed influential on urban land use suitability include topographic slope, proximity to roads and existing developed lands, on-site septicability, and parcelization history. Constraints (binary variables) to urban growth include wetlands, water, protected areas (natural diversity sites, municipal and private open space, federal and state lands), and already-built land. Source data for these factors and constraints have been acquired from existing sources, in either raster or vector format, and gridded into a 30-meter resolution dataset, consistent
with the Landsat-derived land cover data, and imported into Idrisi. Constraints were recoded as binary (0=unsuitable for urban development, and 1=suitable). Factors were scaled, using linear and non-linear transformations, over a suitability range of (0,255). An urban suitability map was generated to serve as the composite of all drivers for possible future urban growth. The logic of incorporating on-site septicability and degree of parcelization is still being investigated. Initial trials with GEOMOD have involved the prediction of 1995 urban land use extent from the drivers discussed above and extant 1985 land use. This process is aiding in the validation of the predictive model. Results from this phase of the research will allow the model to be calibrated by iteratively running GEOMOD with various suitability maps depicting differing growth scenarios. The suitability map that generates the best results in GEOMOD will be used in predictions of development for the years 2002, 2010,
2015, and 2020, which in turn will be used to erode the forest landscape to assess degree of forest loss and fragmentation, using models developed previously by this investigator and his colleagues.
Impacts
The development of a temporal dataset of parcel subdivision for a large study area is significant. These data illustrate change through time and development trends that are otherwise not apparent. The improvement of base spatial data to be used in both descriptive and predictive modeling is significant. Combination of temporal parcelization with physical and cultural geospatial data will lead to a characterization of forest fragmentation. The adoption of an accepted protocol (GEOMOD) for land use change forecasting is important to the application of the methodology developed in this project to other geographic regions. The model, based on physical and cultural landscape data, will be incorporated into a decision support system (DSS) that will be of value to land use planners, forest managers, and decision makers.
Publications
- Hurd, J.D., D.L. Civco. 2004. Temporal characterization of impervious surfaces for the state of Connecticut. Proc. 2004 ASPRS Annual Convention, Denver, CO. 12 p.
- Song, M., D.L. Civco. 2004. Road Extraction Using SVM and Image Segmentation. Photogrammetric Engineering and Remote Sensing. 70(11):1365-1372.
- Civco, D.L., A. Chabaeva, S. Angel, S. Sheppard. 2005. The Urban Growth Management Initiative: Confronting the Expected Doubling of the Size of Cities in the Developing Countries in the Next Thirty Years. Proc. 2005 ASPRS Annual Convention, Baltimore, MD. (accepted November 2004).
|
Progress 01/01/03 to 12/31/03
Outputs
Research in the initial year has focused largely on Data Generation and Collection (Phase 1 of the original Plan of Study). Parcel maps were obtained for the six towns in the study area. The parcelization history for the forty-year period of 1960 to 2000 was manually recorded using the property maps at the town hall, with a specific year of subdivision attributed to each parcel polygon. The parcel maps for the six towns were merged into a master coverage illustrating the subdivision history across the study area. Ancillary data were collected for the study area, including roads, hydrography, and zoning. Finally, a coverage of protected forest lands was compiled, incorporating tracts of land held by state and local government, including state and local parks, state forests, and wildlife sanctuaries. Two elevation models for the area were analyzed. The more recent Shuttle Radar Topography Mission (SRTM) elevation data represents the covered earth surface, including
vegetation and man-made structures. This surface was found to be significantly different from the "bald earth" model of the traditional National Elevation Dataset (NED), particularly in regards to slope. As a result, the derived datasets from the SRTM were found to be less appropriate for the purposes of this study. The 30-meter NED was chosen as the digital elevation model, and was used to derive the additional datasets of slope, aspect, and hill shade. Initial analysis on land parcelization has begun. The date of subdivision data was segmented into five-year increments to illustrate the subdivision trends, graphically illustrated with an animation of the parcelization over the forty-year period. The slope of subdivided areas was also investigated; initial results indicate that slope of subdivided land is comparable to slope of the area overall. Further investigation will compare the size of subdivided areas to the slope in the area. The statewide land use and change information has
been completed. Land use maps are available for 1985, 1990, 1995, and 2002. A change detection map was created as well to compare land use in 1985 to that of 2002. A subset of this map is being used to determine change trends in the study area. Forest change will also be determined from these data. Initial comparison of the raster change detection data to the date of subdivision maps indicates a clear correlation; parcels, which were subdivided within a particular time period, are typically illustrated as having changed from non-developed to developed. In some cases, the delay between subdivision and development is clearly illustrated by comparing the data sources. With the majority of parcelization data collected and processed, the fragmentation analysis component of the study (Phase 2) is underway.
Impacts
The development of a temporal dataset of parcel subdivision for a large study area is significant. These data illustrate change through time and development trends that are otherwise not apparent. Combination of temporal parcelization with physical and cultural geospatial data will lead to a characterization of forest fragmentation. Further, these data will prove useful in critiquing other land use change models.
Publications
- Wilson, E.H., J. D. Hurd, D.L. Civco, S. Prisloe, and C. Arnold. 2003. Development of a Geospatial Model to Quantify, Describe and Map Urban Growth. Special Issue of Remote Sensing of Environment Remote Sensing to Urban Planning and Urban Ecology 86(3):275-285.
- Hurd, J.D., D.L. Civco, E.H. Wilson, S. Prisloe, and C. Arnold. 2003. Temporal characterization of Connecticut's Landscape: Methods, results, and applications. Proc. 2003 ASPRS Annual Convention, Anchorage. AK. 12 p.
- Holdt, B., D.L. Civco and J.D. Hurd. 2004. Forest Fragmentation due to Land Parcelization and Subdivision: A Remote Sensing and GIS Analysis. Proc. 2004 ASPRS Annual Convention, Denver, CO. (accepted December 2003).
- Hurd, J.D. and D.L. Civco. 2004. Temporal Characterization of Impervious Surfaces for the State of Connecticut. Proc. 2004 ASPRS Annual Convention, Denver, CO. (accepted December 2003).
- Chabaeva, A., D.L. Civco, and S. Prisloe. 2004. Development of a Regionally-calibrated Impervious Surface Model. 2004 ASPRS Annual Convention, Denver, CO. (accepted December 2003).
|
|