We conducted a series of analyses investigating the relationship between Steller sea lions, their prey, and various indicators of their physical habitat. We conducted these studies for two different spatial extents, at two spatial resolutions, and three temporal scales. To support this work, we built a collection of physical oceanographic data from satellite-borne remote sensors and ROMS model output. Given the level of processing required to make these various data sets suitable for use with standard, PC-based, analytical software (e.g., R, S-Plus, ArcGIS, IDRISI), we felt that making the processed data sets freely available to other researchers may benefit marine research in the North Pacific.
Studies of marine animals and how their habitats are distributed in the ocean are constrained by the availability of data on the state of the ocean. Data constraints apply to both the attributes that are available, and their spatial and temporal resolution. Habitat studies of widely distributed marine species require similarly extensive data. As part of our investigations into Steller sea lion habitat use and fisheries economics, we have therefore prepared a collection of physical oceanographic data for the eastern North Pacific and the Bering Sea.
One of the motivating factors of our work was the investigation of how spatio-temporal scaling affects predictive models. This required the preparation of a large number of data sets, at variable resolutions. Through studies of species-habitat relationships at different scales, we hope to gain insight into the ecology of marine species, as they respond to seasonal and inter-annual ocean dynamics operating at multiple spatial scales.
Data from Remote sensing are one of the few sources of the broad-extent, comprehensive data sets necessary for such analyses. While many of these data are now available via on-line servers, they are often available only as global coverages, at various temporal resolutions, and occasionally at different spatial resolutions. The data are also often provided at various levels of processing, from the raw satellite data to the end product (i.e., temperature), which is typically what is relevant to marine habitat analyses.
The work required to make such data sets compatible for analysis in typical Geographic Information System (GIS) tools begins after the data of interest are obtained. While there are on-going improvements to the accessibility and compatibility of the data products available from the live-access servers, different variables are often in different digital formats. Typically, the spatial extents of global coverages range from -180 to +180 degrees longitude, adding an extra step to the data preparation for North Pacific analyses.
We created rectified grids of chlorophyll-a concentrations, sea surface temperature, slope of sea surface temperature, sea level anomaly, wind speed. Data layers were created from available on-line sources as both monthly and long-term averages. Monthly averages were produced for all available years, for each sensor, at a spatial scale of 9×9 km² for the Gulf of Alaska and Bering Sea. We prepared long-term averages (climatologies) at 3×3 km² for the northern Gulf of Alaska, while for the eastern North Pacific we prepared each source data set at its native resolution.
We evaluated the sea surface temperature data at the 9×9 km² scale using available quality data, and improved the data provided by interpolating through low quality pixels. Considerable processing was required to create a continuous North Pacific perspective, and to ensure that the data sets were correctly aligned at the different spatial scales.
Ocean circulation models provide another comprehensive view of the ocean, in this case including ocean dynamics and vertical structure, typically at high temporal resolution. To investigate the suitability of such models for habitat studies, we included the output from a ROMS (Regional Ocean Modelling System) model of the northern Gulf of Alaska, developed by Al Hermann and colleagues. The volume of data created by circulation models means the output typically needs to be scaled for use in ecological studies. The use of such data for ocean habitat studies is rich with possibilities.
We present 2-week averaged data from the output of a Regional Ocean Modelling System (ROMS) implemented for the northern Gulf of Alaska (3×3 km²) for the year 2001. These data provide a representation of the changing, vertical structure of the ocean. Given the significant investment to create a rectified data collection, we have prepared the data for distribution to interested researchers. The ROMS data are provided in MS Access format, and the Remote sensing data as binary float files. Federal Geographic Data Committee (FGDC)-compatible metadata have been prepared. The data described herein are available from the Marine Mammal Research Unit web site or on request.