![](https://webarchive.library.unt.edu/web/20130219020826im_/http://www.afsc.noaa.gov/images/spacer.gif)
![](https://webarchive.library.unt.edu/web/20130219020826im_/http://www.afsc.noaa.gov/images/spacer.gif)
National Marine Mammal Laboratory (NMML)
Polar Ecosystems Program
Researchers Census Harbor Seals in the Gulf of Alaska
![Gulf of Alaska region map, see caption](images/NMMLfig3.jpg)
Figure 1. Map of the Gulf of Alaska region showing the twelve survey areas. (click map to enlarge) |
|
NMML’s Polar Ecosystems Program conducts aerial surveys of harbor seals in Alaska every August during the molt when seals spend much of their time out of the water. For logistical purposes, Alaska is divided into five regions and one region is surveyed each year. The five regions are: the Gulf of Alaska (surveyed this year) (Fig. 1), northern Southeast Alaska (2007), southern Southeast Alaska (2008), Aleutian Islands (2009), and the north side of the Alaska Peninsula and Bristol Bay (2010).
The 2006 surveys were conducted 5-17 August and used 12 aircraft and 12 researchers from NMML, the Alaska Department of Fish and Game, the Alaska Sea Life Center, and the Coastal and Marine Resources Centre (Cork, Ireland). We utilized four NOAA and eight charter aircraft. Five planes and crew were based in Cordova, four in Kodiak, two in Homer, and one in Moose Pass (near Seward).
All surveys were conducted within 2 hours of low tide. Observers spent the first few days flying reconnaissance, which entailed searching for haul-out sites along the entire coastline within each area. With new and historic sites mapped, surveys were flown daily from site-to-site to maximize the number of counts at each haulout. Each plane’s movement was recorded using a GPS while observers kept track of effort. Digital cameras provided a time stamp for each image so that it could later be geo-referenced. Most photographs were taken hand-held from oblique angles at elevations of 500-1,000 feet. Sites where seals were hauled out on ice were photographed using a down-looking digital camera linked to a GPS.
This year, weather conditions were unusually poor, particularly in Prince William Sound, where several flights were cancelled due to precipitation and low visibility. Despite the weather, at least four replicate surveys were conducted at most, if not all, of the larger seal haulouts. Analyses have shown that the larger haulouts (>20 seals) account for the vast majority of variability in counts across days. Image processing and counting will occur at NMML over the next several months.
The 2006 surveys were conducted under a new protocol designed to improve the spatial and temporal resolution of our surveys, data management, and workflow. The entire range of harbor seals in Alaska was divided into fixed areas (“polygons”) for documentation of seal counts and survey effort. In past surveys, seal haulouts and counts were accurately associated with geographic coordinates, but it was often difficult to manage the documentation of locations where seals were absent (i.e., to distinguish between true zeros and haulouts that were missed). Geo-referenced photographs, and the seals counted in them, can now be assigned to a particular polygon, and observers can easily confirm whether each polygon has been completely surveyed. This approach allows observers to focus on sighting seals and enhances continuity in the technique across years when observers and pilots may change.
For archival purposes, the coordinates calculated for each image are stored within the metadata of each image file. As the seals within images are counted (using Adobe Photoshop® and Adobe Bridge® CS2), abundance estimates will also be stored within the metadata of each image as custom XMP fields. Storing the survey data within the image metadata eliminates the need for an intermediate spreadsheet or database and thus reduces error. Prior to analysis, these metadata can be extracted in a batch process and stored directly to a SQL server database. Various analytical programs can then access the survey data directly from the SQL database.
By Dave Withrow, Josh London, and John Jansen
|