Distributions of land cover and forest formations for St. Lucia and St. Vincent and the Grenadines were mapped from satellite imagery. These products represent a successful collaborative effort between TNC, USDA-IITF, USGS, and the forestry and planning agencies on both St. Lucia and St. Vincent and the Grenadines. The local agencies have spent the last six months reviewing and verifying many of the derived land cover classes. Prior to field data collection, an extensive GIS database was assembled that was composed of ancillary geographic data collected from a variety of agencies from both countries. The final land cover classes were derived from multiple data sources including mosaicked ETM+ scenes (2001) representing the least amount of cloud cover, high-resolution IKONOS (1x1m) color imagery (2001), high resolution aerial photography (1999), other ancillary data, and local knowledge. The cloud cover present in the ETM+ mosaics was problematic. Cloud cover for St. Lucia and St. Vincent and the Grenadines was approximately 30%, and 25%, respectively. In order to map clouded areas we used prediction results from the See 5 rule-based classifier using ancillary data (elevation, distance buffers, etc). The IKONOS scenes were extremely useful for delineating socio-economic activities such as infrastructure (buildings and roads), black/white sand beaches, agriculture (bananas and coconut), tourism activities (hotels, marinas, golf courses) and industrial sites (quarries, mines, dumps). These features took precedence in final overlay classification due to the high accuracy that was used in the digitization process. Field work was conducted on Jan 2-10, 2004 and local agencies provided valuable field support and sampling strategies that permitted the team to collect over 150 GPS point locations throughout both islands. Local Caribbean botany experts accompanied the field team and provided valuable insight for assessing canopy characteristics and identifying general forest formations for woody vegetation classes. At each sampling point, extensive notes were collected as well as digital video and photos. Land cover reference data that was collected included urban/built-up lands, agriculture (herbaceous vs. woody crops), pasture/grass, emergent wetlands, water, barren areas, and 3-5 classes of woody vegetation as described in the proposal. All field notes and video/photos were catalogued and delivered to the Department of Forestry on both islands.

Satellite image-based mapping of tropical forests is often vital to conservation planning. Standard methods for automated image classification, however, limit classification detail in complex tropical landscapes. This study first tests an approach to image interpretation for mapping forest formations and land-cover classes in two islands of the Lesser Antilles, including St. Lucia and St. Vincent and the Grenadines. Secondly, it quantifies extents of land cover and protected forest by formation for five islands and asks whether cultivated land areas have declined over the last half-century. The image interpretation approach layered image data together with ancillary geographic data and classified the resulting stacks of raster data with decision trees. Finally, the approach used imagery from two seasons. The approach accurately distinguished several classes that would be confused with more standard methods. The seamless mosaics aided reference data collection. Having two seasons of imagery also improved reference data by allowing us to separate drought deciduous forests and woodlands from semi-deciduous ones. With the resulting detailed maps, we learned that higher-elevation forest formations are well protected on the islands. Drier lowland forests, however, are not well represented in the islands' reserve systems. Cultivated land areas have declined by 60 to 100 percent on the islands. Consequently, formerly cultivated lands could provide lowland habitat if new reserves were designated in lowland areas.

<?xml version = '1.0' encoding = 'ISO-8859-1'?> <eainfo> <detailed> <enttyp> <enttypl>svg_landcover.img.vat</enttypl> <enttypd></enttypd> <enttypds></enttypds> </enttyp> <attr> <attrlabl>Count</attrlabl> <attrdef></attrdef> <attrdefs></attrdefs> <atttype></atttype> <attprecision></attprecision> <attlength></attlength> <attnullsallowed>false</attnullsallowed> <attisprimarykey>false</attisprimarykey> <attisunique>false</attisunique> <attisindex>false</attisindex> <attrdomv> <codesetd> <codesetn></codesetn> <codesets></codesets> </codesetd> </attrdomv> <attrdomv> </attrdomv> <attrdomv> <rdom> <rdommin></rdommin> <rdommax></rdommax> <attrunit></attrunit> <attrmres></attrmres> </rdom> </attrdomv> <attrdomv> <udom></udom> </attrdomv> <attrvai> <attrva></attrva> <attrvae></attrvae> </attrvai> <attrmfrq></attrmfrq> </attr> <attr> <attrlabl>Value</attrlabl> <attrdef>Internal feature number.</attrdef> <attrdefs>Esri</attrdefs> <atttype></atttype> <attprecision></attprecision> <attlength></attlength> <attnullsallowed>false</attnullsallowed> <attisprimarykey>false</attisprimarykey> <attisunique>false</attisunique> <attisindex>false</attisindex> <attrdomv> <codesetd> <codesetn></codesetn> <codesets></codesets> </codesetd> </attrdomv> <attrdomv> </attrdomv> <attrdomv> <rdom> <rdommin></rdommin> <rdommax></rdommax> <attrunit></attrunit> <attrmres></attrmres> </rdom> </attrdomv> <attrdomv> <udom>Sequential unique whole numbers that are automatically generated.</udom> </attrdomv> <attrvai> <attrva></attrva> <attrvae></attrvae> </attrvai> <attrmfrq></attrmfrq> </attr> <attr> <attrlabl>VALUE</attrlabl> <attrdef>Internal feature number.</attrdef> <attrdefs>Esri</attrdefs> <atttype></atttype> <attprecision></attprecision> <attlength></attlength> <attnullsallowed>false</attnullsallowed> <attisprimarykey>false</attisprimarykey> <attisunique>false</attisunique> <attisindex>false</attisindex> <attrdomv> <codesetd> <codesetn></codesetn> <codesets></codesets> </codesetd> </attrdomv> <attrdomv> </attrdomv> <attrdomv> <rdom> <rdommin></rdommin> <rdommax></rdommax> <attrunit></attrunit> <attrmres></attrmres> </rdom> </attrdomv> <attrdomv> <udom>Sequential unique whole numbers that are automatically generated.</udom> </attrdomv> <attrvai> <attrva></attrva> <attrvae></attrvae> </attrvai> <attrmfrq></attrmfrq> </attr> <attr> <attrlabl>Class_Name</attrlabl> <attrdef>Internal feature number.</attrdef> <attrdefs>Esri</attrdefs> <atttype></atttype> <attprecision></attprecision> <attlength></attlength> <attnullsallowed>false</attnullsallowed> <attisprimarykey>false</attisprimarykey> <attisunique>false</attisunique> <attisindex>false</attisindex> <attrdomv> <codesetd> <codesetn></codesetn> <codesets></codesets> </codesetd> </attrdomv> <attrdomv> </attrdomv> <attrdomv> <rdom> <rdommin></rdommin> <rdommax></rdommax> <attrunit></attrunit> <attrmres></attrmres> </rdom> </attrdomv> <attrdomv> <udom>Sequential unique whole numbers that are automatically generated.</udom> </attrdomv> <attrvai> <attrva></attrva> <attrvae></attrvae> </attrvai> <attrmfrq></attrmfrq> </attr> <attr> <attrlabl>COUNT</attrlabl> <attrdef>Internal feature number.</attrdef> <attrdefs>Esri</attrdefs> <atttype></atttype> <attprecision></attprecision> <attlength></attlength> <attnullsallowed>false</attnullsallowed> <attisprimarykey>false</attisprimarykey> <attisunique>false</attisunique> <attisindex>false</attisindex> <attrdomv> <codesetd> <codesetn></codesetn> <codesets></codesets> </codesetd> </attrdomv> <attrdomv> </attrdomv> <attrdomv> <rdom> <rdommin></rdommin> <rdommax></rdommax> <attrunit></attrunit> <attrmres></attrmres> </rdom> </attrdomv> <attrdomv> <udom>Sequential unique whole numbers that are automatically generated.</udom> </attrdomv> <attrvai> <attrva></attrva> <attrvae></attrvae> </attrvai> <attrmfrq></attrmfrq> </attr> </detailed> </eainfo>