Short URL

Pine Plantation Mapping in the Boston Mountains

 

Other Remote Sensing areas:

Loading, please wait...

Project Overview

The overall goal of this project was to identify and map pine plantations within the Boston Mountains ecoregion and to integrate this new information into the existing Boston Mountains Terrestrial Habitat (TH) map and the Boston Mountains General Vegetation map. A secondary project task involved the revision of existing high resolution TH maps for three ANHC Natural Areas in order to more accurately reflect the land-cover of these areas.

It is often a painstaking process to distinguish pine plantations from naturally regenerating pine stands by way of on-the-ground, visual inspection, and a ground survey for an area as large as the Boston Mountains ecoregion would require a substantial investment in both time and money. Traditional remote sensing techniques employing one date of photography or satellite imagery, on the other hand, could be used to successfully map land-cover types at any given point in time, but pine plantations are a temporal, land-use phenomenon and can only be mapped by taking the historical satellite image record into account. Therefore pine plantations must be mapped both spatially and temporally, and this mapping process requires considerable first-hand knowledge of the study area’s historical and physical landscape.


For the pine plantation mapping portion of this project, 30 meter resolution satellite imagery from the Landsat Thematic Mapper (TM) was employed. Landsat TM’s uniquely consistent data continuity extends back to the Landsat 4 program in 1984, so the temporal timeframe for this study spans the 25 year period from 1984 – 2009. Two Landsat scenes (leaf-on and leaf-off) from the same year, or as close to the same year as possible, were selected and acquired. There is, roughly, a 4 year interval between each scene over the 25 year historical span for a total of 7 separate historical two-season “looks” at the study area.

Each Landsat scene was geo-corrected (rectified) to a standard UTM grid projection. The imagery was also color/contrast balanced. The classification process employed both traditional pixel-based analysis as well as object-based image analysis (OBIA). Results from each of these two procedures were merged and seven “pine existence” (i.e. Pine/Not Pine) maps were produced for the study area: one for each scene/date. Next, all of the pine existence maps were combined into one database for historical comparison. A rules-based, seven-case binary procedure was developed for the project, and the final decision rules for the procedure were developed by ANHC and CAST personnel.

Field data was conducted by CAST and ANHC staffers for the purpose of collecting a set of points and/or areas for accuracy assessment. CAST provided GPS equipment and the personnel to operate the equipment. ANHC provided personnel to identify and provide background information about the pine plantations to be mapped.

 

The overall accuracy of the final pine plantation map was an acceptable 82.5%. The accuracy assessment conducted for this study, however, is not fully conclusive because a new category called “Pine: Unknown” was added to the final category list. This addition caused slight categorical discrepancies in the accuracy assessment. Because pine plantations are, for the purposes of this particular remote sensing study, a temporally-defined, land-use type, the continued monitoring of pine land-cover in coming years throughout the ecoregion will produce a better pine plantations map layer.

 

Contacts

Turn on JavaScript!

Turn on JavaScript!Turn on JavaScript!