Buffer Labeled Impervious Data: An initial buffering of the impervious data produced the preliminary boundaries of the land uses for urban classes. Before buffering, small (< 20 m2) free-standing impervious clumps were erased. Most of these were identified as small isolated structures such as sheds, etc. Depending on the land use label assigned to the clumps, various buffer distances and priority rules were applied. Create Pervious Classification: The non-impervious (pervious) uses were classified separately through image segmentation using Definiens® software. The segmented classification was filtered to generalize / smooth the class boundaries, and features smaller than about 1 acre were dissolved into surrounding features. Label Pervious Data: Where logical combinations of segment labels and 1999 land use labels occurred, the segment was labeled with the 1999 class which contained the largest portion of the segment (using a zonal majority rule). When the combination was illogical, the pervious data were flagged and then corrected in a manual review. Secondary Buffer: The buffered and labeled impervious and labeled pervious data were combined through a second buffer using ESRI's ArcGIS®. This buffer incorporated natural classes into urban class delineations, e.g. lawns (pervious segment) into residential (buffered and labeled impervious). It allowed impervious features of certain classes to grow into pervious areas of specific classes. This buffer was allowed to grow a specified distance beyond the edge of the first buffer. The buffer distance and class hierarchy were specified in an automated model. Note: A significant result of the semi-automated method of production is that low and very low density residential areas are mapped so that the "useable area" (the residence and landscaped area immediately around it) is depicted. For example, a neighborhood of houses on relatively large wooded lots will be shown as polygons around the developed portions of the properties embedded in a matrix of Forest. Adding Wetlands: Wetlands from the DEP Wetlands datalayer were added to the land use classification, replacing any underlying interpreted polygons. The wetlannds polygons were not modified in any way, and the coastline was based on this dataset. No interpretation of wetlands was done except for cranberry bogs. Clean Data / Post-production: A series of filters was applied to smooth the polygon boundaries, and to maintain a minimum mapping area. (Isolated polygons with developed uses in rural areas were kept, even though the area may have been less than the 1 acre MMU.) A final manual quality control (QC) review of the data was performed by Sanborn, and polygons that did not delineate appropriately through the automated method were corrected in terms of geometry and labeling. During this phase of production, MassGIS provided input based on local knowledge and onsite field verification. Two areas were specifically addressed post production: the under-representation of multi-family housing in areas where there were no parcel data, and the residential class densities. Multi-family housing was delineated through the use of an assessment dataset geocoded to NAVTEQ road centerlines (Warren Group points), and identified manually in areas where there were significant concentrations of these points. MassGIS QA: Upon receipt of the data, MassGIS performed extensive QA steps. These included comparing the draft land use with numerous ancillary datasets, manual photo interpretation, and field visits. Draft data was graciously reviewed by several Regional Planning Agencies across the state. MassGIS carried out an additional post production step to further improve the mapping of residential, commercial, and industrial land use. In municipalities where assessors' parcels were available, the developed portions of residential, commercial, or industrial parcels with use codes that unambiguously corresponded to single land use codes were incorporated into the land use data. Additionally, in order to make the statewide dataset more manageable, polygons were simplified and slivers were removed. During QA, MassGIS combined the data into a single statewide layer, and then split it into more manageable County-sized pieces for distribution. For more efficient display and spatial indexing, the final statewide version includes the county boundaries. County boundaries are based on the Towns from Survey Points layer.

Information relevant to the GIS data encoding

This is a statewide, seamless digital dataset of land use / land cover for the State of Massachusetts derived using semi-automated methods and based on digital imagery captured in 2005 with 0.5 m pixel resolution. The project area encompasses the State of Massachusetts. The minimum mapping unit (MMU) for this dataset is 1 acre overall for the dataset. However, there are some exceptions: a MMU of 0.5 acre has been used for rural areas where there may be isolated residential, commercial, or industrial structures. In such cases, areas less than 1 acre and more than about 0.5 acre are mapped. In urban areas, the MMU is 1 acre. This implies that a water tower or a communication tower in the middle of a residential or commercial area may not be mapped out separately if it occupies less than 1 acre of land. In areas where assessor parcel data was available, Multi-family Residential, Commercial, and Industrial areas may be as small as 0.25 acre. The land use classification scheme used for these data is based on a coding schema used in previous land use datasets in Massachusetts with some modifications for the 2005 classification.Horizontal Coordinate System Name:NAD 1983 StatePlane Massachusetts Mainland FIPS 2001

Datum: D North American 1983
Reference Ellipsoid: Name: GRS 1980 Semi Axis: 6378137.0
Meridian: Greenwich
Projection Name : Lambert Conformal Conic