Forest Stewardship Spatial Analysis Project Maryland Methodology January, 2005 Project Summary The purpose of the Spatial Analysis Project (SAP) is to create a series of GIS data layers and tools for the state that represents various levels of potential benefit from, and suitability for inclusion in the Stewardship Program as delivered by state forestry agencies and the U.S. Forest Service. Private land programs and GIS staff from the four states involved in the pilot SAP effort (Connecticut, Maryland, Massachusetts and Missouri), along with U.S. Forest Service program identified 12 factors which help identify the “Stewardship potential” of a given piece of land. The factors were differentiated into two groups: resource potential and resource threats. The resource potential factors include: Riparian Zones Priority Watersheds Forest Patch Size Natural Heritage Data Public Drinking Water Supply Sources Private Forest Lands Proximity to Public Lands Wetlands Topographic Slope The resource threat factors include: Forest Health Development Level Wildfire Assessment Certain lands within any state are not eligible for inclusion in the Forest Stewardship program. Land use / land cover factors which identify these areas are open water, industrial and urban areas. In Maryland open water was eliminated from the data sets and a mask was created to exclude industrial and urban areas from the analysis. Publicly owned lands were not entirely removed from Maryland’s considerations as some programs could be applied under certain conditions. Those publicly owned lands that were identified as not being eligible were removed from the data set through the same process as the open water. Once the 12 factors were identified, each state could determine the relative importance of each of the criteria based on their state-specific conditions or treat all areas as equal weights. Maryland Department of Natural Resources Forest Service staff decided to treat each layers 30 meter grid resolution intersect as an equal weight of 1 for a hit or miss summation model. A weighted 32-layer model was also attempted during the early planning phases to determine the feasibility of such an effort. It was determined to be too complex and difficult to interpret by the reviewers in the Stewardship advisory committee and Maryland DNR Forest Service staff. After reviewing the work of the other three states in the SAP project Maryland returned to the basic 12 layer equal weighted model as described above and used by the other project states. The 12 layers were then combined in a GIS overlay analysis which took into account the equal weight for each factor. The final product was a single data layer which represents the suitability of the land for inclusion in the Forest Stewardship Program. Possible values from this analysis range from 0 to 12, with a value of 12 representing the highest level of suitability. Actual values for Maryland ranged from 0 to 12. A natural breaks classification algorithm was used to break the values into low, medium and high classes. The result is shown below. Summary statistics were calculated and a series of maps were then created to display the data as required by the U.S. Forest Service SAP project team. While the process outlined above was taking place a parallel effort was occurring. In order to understand where the Forest Stewardship Program has been previously implemented, the centroids and property boundaries for ownerships with a Stewardship plan were digitized. For Maryland the centroids were the easiest to develop as we had 4,800 Stewardship plans for the years 1990-2000. These plans existed only as hard copy records scattered across 23 counties. In order to get this part of the task done no less than seven personnel were hired during the projects operational period to collect and scan data on 4,800 potential plans. There was significant turnover in staff and only two of those staff remain in service today. The centroid based database was completed Fall of 2003 and used in the SAP project maps buffered to a size matching their acres of impact under the Stewardship Programs. The process of digitizing polygons to match the centroids began Summer 2003 and the 1990-1999 polygons were completed in November 2004 with the exception of those from Mongomery County Maryland where record no longer used tax maps as the locator. This difficulty made it hard to determine where exactly the polygons belong. Through an intensive effort Montgomery County’s polygons were completed and incorporated into the database in January 2005. This complete set of Stewardship plan polygons were overlaid on the Stewardship potential layer to assess Stewardship efforts through 1999. Plans and polygons for the period from 2000 through 2004 will be incorporated into the Stewardship Database by December 31, 2005. Data Development Riparian Zones: Riparian zones were created by buffering (300 feet each side) the streams data set from the Maryland Office of Planning 1:60,000 source scale files. The buffers are shown in green. Priority Watersheds: These watersheds are considered priorities for various reasons including: non-point source pollution focus areas (SALT AgNIPS), presence of threatened / endangered species, Farm Bill program focus areas, etc. Watershed polygons were created by aggregating polygons from 12-digit Maryland DNR Watershed Services Division shapefiles. Source scale for this data is 1:24,000. The priority watersheds are shown in green. Forest Patches: All forest cover used in this project was extracted from the early 1990s National Land Cover Database (NLCD). This landcover classification was based on Landsat Thematic Mapper 30 meter satellite imagery acquired in 1992 & 1993. Five NLCD classes were used to create a "forest" layer: deciduous forest, coniferous forest, mixed forest & woodland, shrubland, and woody wetlands (NLCD classes 41, 42, 43, 51, 91). Large roads create discontinuities in forest cover and reduce forest patch size for some wildlife species. Accordingly, the state-maintained roads were buffered by road type: 50 feet (each side) for interstates, 50 feet (each side) for U.S. and MD highways. This buffered road layer was then erased through the private forest lands layer. The erase process has the effect of cutting a hole in the forest layer wherever the buffered roads occur. Minimum forest patch size was set at 50 acres. The result is the forest patch layer. This 30m dataset was used in the analysis. Forest patches are shown in green. Natural Heritage Data: Individual plant and animal records, and high quality community records from the Maryland Natural Heritage Database were used to represent areas of biological importance. Point and polygon records were both buffered by random amounts by the Maryland DNR Wildlife and Heritage staff. Old, general historic records were not used. Records are created using 1:24,000 scale USGS topographic maps as the base. These areas are shown in green. Public Drinking Water Supply Areas: Shapefiles from the Maryland Department of Environment Public Drinking Water Program were provided per our request. Source scale for this data is 1:24,000. Lake intakes and buffered (1 kilometer) surface and well intake points were unioned into one file and clipped with the Maryland boundary polygon. The MDE data set appeared to be missing some vital data areas but attempts to obtain further information proved fruitless as MDE had reduced it’s GIS staff and seemed unable or unwilling to comply with data requests over a two year period. Water resources and their buffers are shown in green. Private Forest Lands: All forest cover used in this project was extracted from the early 1990s National Land Cover Database (NLCD). This landcover classification was based on Landsat Thematic Mapper 30 meter satellite imagery acquired in 1992 & 1993. Five NLCD classes were used to create a "forest" layer: deciduous forest, coniferous forest, mixed forest & woodland, shrubland, and woody wetlands (NLCD classes 41, 42, 43, 51, 91). Private forest land is created by erasing a public lands shape file through the forest layer, creating "holes" in place of the public lands. This 30m dataset was then used in the analysis. Private forest lands are shown in green. Proximity to Public Lands: A single shapefile of all public land ownership in Maryland was created by merging individual agency files for Federal, State and county land holdings as maintained by the Maryland DNR and the Maryland Office of Planning. The source scale for all the individual shapefiles is 1:24,000. The public lands were buffered by 1000meters in an effort to show the priority to develop contiguous forest blocks adjacent to public lands. These areas are shown in green. Wetlands: Digital National Wetlands Inventory shapefiles originally compiled at 1:24,000 scale by the U.S. Fish and Wildlife Service were used by the Maryland DNR Watershed Management Division as the source for this layer which was modified by new data field sampled by state biologists and ecologists. All state designated wetlands classes were used. These areas are shown in shades of pink to red based on the wetlands importance. Topographic Slope: A statewide 30 meter Digital Elevation Model (1:24,000 scale source) from the National Elevation Dataset (USGS) was used to select areas where slope is between 15% and 35%. 35% represents the general maximum operability limit for a skidder operating in Maryland. Between 15% and 35% the Stewardship Program can be used to help control soil erosion. These areas are shown in green. Forest Health: The most extensive, addressable forest health concern occurring in Maryland over the period addressed is Gypsy Moth. In future efforts, maps for southern pine beetle, emerald ash borer (EAB), hemlock woolly adelgid (HWA) and sudden oak death (SOD) may be considered for inclusion. Statewide data was collected from the Maryland Department of Agriculture records for ten years of Gypsy Moth Treatment from 1990-2000 showing treatment zones. The area described are shown in green. Developing Areas: Based on U.S. Census Bureau 1990 and 2000 data. Census block group polygons from 1990 and 2000 were unioned to deal with differing census geography boundaries and the number of households per square kilometer was calculated. The change in number of households / sq. km. from 1990 to 2000 was then calculated. Block group polygons where the change was >= 1 and <= 8 households / sq. km. and not in urban areas were selected. These areas are shown in green. Wildfire Assessment: This layer is based on the Maryland DNR Forest Services Geomatics Lab THARSFE Wildland Urban Interface (WUI) Fire Model. The model is a composite of: wildfire risk based on wildland urban development level, wildfire hazard fuels density and types of ground cover and wildfire sensitivity based on landuse and perceived value. Fire equipment and fire response accessibility were used in the model to show the ability of the local governments to deal with WUI fire. The WUI THARSFE model was broken out originally by 12-digit watersheds and natural breaks of the summed zones model values into five categories ranging from very low to extreme fire impact. The high to extreme categories were used for SAP from the original assessment to create the "1" or hit cells for this layer. These areas are shown in color based on level of fire impact. Green is the least vulnerable and dark brown is the most vulnerable to wildfire and are rated as follows: Green 1, Light Green 2, Yellow 3, light brown 4, dark brown 5. Analysis Mask: The analysis mask contains urban/developed areas that were not considered in the analysis. Urban & developed areas were compiled from the urban and industrial - commercial lands from the NLCD land cover classification. The final layer was created by erasing the composite shape file of unavailable areas through the Maryland state polygon, creating a state with holes in it corresponding to the unavailable areas. The masked areas are shown in grey. All Maryland’s layers were clipped for open water and those public lands not eligible for Stewardship. Stewardship Plan Ownership Boundaries: Entire tract boundaries for which a Stewardship Plan has been written since 1995 - 2000. Tract boundaries were compiled at a source scale of 1:24,000. The Stewardship Plan tracts are shown in black. Weighting The twelve criteria identified as contributing to the potential Stewardship Program benefit of a given piece of ground were each given an equal weight of 1 for where they existed geographically and a 0 for where they do not exist. Each layer was converted into a 30meter grid of 1 and 0 valued grid cells. The analysis summed the intersected grids giving the analysis layer a range of grid values from 0 to 12 where 12 is the most important for Stewardship efforts and 0 is of no importance or ineligible. The following shows the layers and their weights for each factor. Criterion Riparian Corridors Priority Watersheds Forest Patches Forest Health Natural Heritage Sites Drinking Water Supply Private Forest Public Land Adjacency Wetlands Slope Development Risk Wildfire Assessment Weight 1 1 1 1 1 1 1 1 1 1 1 1 GIS Analysis The GIS data representing each of the twelve factors was converted to the ESRI Grid format with a cell size of 30 meters. The Grid of each factor was converted to a 0, 1 format. For example, all the 30m grid cells that fell within the riparian buffers were coded as a “1”, while all the cells that were outside the areas of the riparian buffers were give the value “0” in that layer. The analysis process is relatively straightforward. Each Grid is multiplied by its weight value, so that the cells coded as “1” take on the weight value while all the “0” cells retain a value of 0. Because all twelve grids were derived from the same source, the grid cells of each layer line up exactly with the cells from all the other layers. The overlay analysis procedure uses this fact to create a final result Grid whose individual cell values equal the sum of the values in the same location (on the same half-acre) from all twelve layers. The maximum possible cell value in the final Grid is 12. Result Grid values for Maryland could potentially range from 0 to 12. No single cell location was a “hit” in all twelve data layers, though many were “misses” in all twelve layers. The ESRI Spatial Analyst extension with model builder allows for the specification of an analysis mask. The analysis mask layer described above was used in this capacity to exclude areas of Maryland that don’t meet eligibility criteria for inclusion in the Forest Stewardship Program (open water, public lands, urban areas, etc.). To make interpretation of results easier and allow for computation of area statistics, three data classes were established to group the continuous cell values: Low, Medium and High Stewardship Potential. There are several possible methods for establishing class breaks. The project group decided to use the Natural Breaks classification algorithm available in ArcView. This was used by Maryland for Stewardship Potential maps and Stewardship Threat maps. The 12 data layer composite was grouped into three categories based on the number of intersects in a given pixel as shown in the table below. Maryland staff decided this gave a more useful picture as the ranges using natural breaks were quite small towards the low and high with over 80% falling within the natural breaks medium range creating an almost featureless image except around the Chesapeake Bay where the bulk of the high range appeared. Low Medium High 0-3 3-6 6-12 Creating a Maryland Stewardship Plan Database with Associated Polygon Locations Stewardship plans are not currently collected centrally in Maryland. Maps of each Stewardship plan tract were digitized at 1:12,000 scale on a Digital Ortho Quarter Quad based on the Stewardship plans written text description and the sketched maps based on the Maryland ADC road map books which were collected from each Forestry office in the state. Each tract was assigned a unique identifier based upon its legal description, and a database of plan attributes is assembled. Title searches and tax records had to be searched often to confirm the true location of the properties in question as often the locations were poorly defined or paper maps were entirely unavailable. Maryland has a high rate of mobile population which means tracts tend to change ownership often. This adds to the difficulties involved in locating poorly recorded tract data when the main source of location is the owners address and tax number. Often it was necessary to determine who the new owners were to locate the property and any recent subdivisions. Key Meetings And Project Genesis Much of the genesis of the Stewardship Spatial Analysis Project can be traced to a meeting on January 19, 2001 in Newtown Square Pennsylvania. At this meeting Steve Koehn, Don Van Hassent and John Wolf from Maryland DNR presented preliminary work on Maryland’s Strategic Forest Lands Assessment (SFLA), built on the ecologically-based modeling methodology for Maryland’s Green Infrastructure Analysis Project. USFS NA staff Susan Lacy and Mark Buccowich were present and could see the significant value of this modeling approach, and how a similar analysis might be able to benefit and focus Forest Stewardship efforts in the northeast. Maryland’s SFLA, funded through the USDA Forest Service NA for $80,000 started on October 1, 1999 and was completed December 31, 2003. With the seed planted, USFS staff sought to find other states willing to participate in a pilot project to build a model capable of guiding Stewardship efforts to areas where it would have the greatest benefit. The first meeting of the four state pilot group occurred at Newtown Square Pennsylvania on July 5-6, 2001. The meeting included representatives from the USDA Forest Service NA, and state forestry representatives from Maryland, Massachusetts, Connecticut, and Missouri. USDA FS funding for each of the four pilot states had already been secured through the Northeastern area, with Maryland’s grant for $61,000 beginning on June 1, 2001. A subsequent meeting of the principles from each of the four states and Forest Service staff was held in Annapolis, Maryland on November 1-2, 2001. Over the next three years the group met numerous times and spend countless hours pouring over data, evaluating rankings and honing the analysis. Maryland Stewardship Committee Involvement Maryland’s Stewardship Committee first received a briefing on the Stewardship Spatial Analysis Project at their summer 2001 meeting. The group received briefings on both SFLA and the exciting new SAP (than called Stewardship Initiative). During this meeting committee members were promised the opportunity to have input on the ranking process as it was developed for the SAP. In March 2003 the Stewardship Committee met again and had the opportunity to comment on the layers and the suggested rankings used in the SAP models. The Stewardship Committee felt strongly that the 32-layer approach presented in that meeting was overly complex. Following the meeting Pat Patterson revised the model to a 12-layer approach that was more closely aligned with the other three pilot states. A presentation of the final draft model and an opportunity for comment was provided at the December 17, 2004 meeting of the Maryland Stewardship Committee. Evolving and Potential Uses The Maryland Forest Service Operations Team made up of the four Regional Foresters and key headquarters staff met to discuss the potential use of the SAP analysis for Stewardship Program implementation on November 9, 2004. A subsequent meeting of the Regional Foresters and the rest of the Maryland Forest Service Leadership Team occurred on December 9, 2004 with an agreement to fully analyze the potential of using the SAP to prioritize Stewardship Program implementation in Maryland. The Leadership Team also approved a new operations order that will require all future Stewardship Plans and plan maps to be captured digitally and stored in the Stewardship Database. This new process will likely capture all Stewardship Plans and the maps created beginning in 2005. Maryland is also in the process of digitizing polygon maps for the Stewardship Plans developed between 2000-2004. Those plans should be completely incorporated into Maryland’s Stewardship Database by December 31, 2005. A corollary effort is also underway in Maryland to incorporate SAP and a product from Maryland SFLA, the Economic Model to build a decision support tool to help guide the acquisition and protection of forestland in Maryland. The “Economic Parcel Evaluation Tool” will use these two models to rank forest sustainability and justify the importance of individual parcels as they are considered for protection through fee simple purchase or conservation easement protection through Maryland Rural Legacy, Forest Legacy, Agricultural Preservation, Maryland Environmental Trust or Program Open Space. (Note: Questions about Maryland’s SAP Methodology should be addressed to Pat Patterson or Jeff Horan of Maryland DNR Forest Service.)