Satellite Monitoring of Cook Inlet Beluga Whales
Geospatial Artificial Intelligence For Animals uses high-resolution satellite imagery to detect endangered belugas in Cook Inlet, Alaska.
Cook Inlet beluga whales are a genetically distinct and geographically isolated population inhabiting the silty, extreme environment of Southcentral Alaska. Unlike their oceanic relatives, these "canaries of the sea" have adapted to one of the most challenging marine habitats on Earth. This area is characterized by massive 30-foot tidal ranges, heavy seasonal ice cover, and high turbidity from glacial runoff. This opaque water makes visual hunting for prey nearly impossible, forcing the whales to rely almost entirely on echolocation to navigate and forage. Their habitat is also a hub of human activity, encompassing Alaska’s busiest port and primary urban center. This subjects the population to a unique combination of stressors including industrial noise, pollution, and climate-induced shifts in prey availability. Because they do not migrate out of the Inlet, their survival is linked to the health of this specific, confined ecosystem, making them a sentinel species for the region's environmental integrity.
Collecting Satellite Imagery of Cook Inlet Beluga Whales
The Geospatial Artificial Intelligence for Animals (GAIA) team, in partnership with the U.S. Geological Survey, is tasking and collecting VHR satellite imagery in Cook Inlet. This effort aims to advance research on detecting Cook Inlet beluga whales by augmenting data from surveys that are limited in time or space.
Current imagery collections focus on several key regions:
- Upper Cook Inlet including Susitna Delta, Chickaloon Bay, and Turnagain and Knik Arms
- Trading Bay
- Redoubt Bay
- Kenai
- Tuxedni Bay
The primary goal of this work is to build a robust, well-annotated training dataset needed to train and evaluate machine-learning models.
Bridging the Data Gap: High-Priority Satellite Tasking
GAIA is currently exploring how very high-resolution (VHR) satellite imagery can support Cook Inlet beluga stock assessments and related management needs. The team has extensive experience tasking imagery through the Electro-Optical Commercial Layer contract administered by the U.S. National Reconnaissance Office. However, the resulting data have not consistently met assessment requirements. Due to high demand and the relatively low prioritization of this type of tasking, imagery requests in this region are not always fulfilled. For example, in June 2025, tasking aligned with the Cook Inlet beluga aerial abundance survey period did not yield any imagery for that month. This likely reflects competing priorities in nearby areas such as the Port of Anchorage and Joint Base Elmendorf-Richardson.
In the case of Cook Inlet belugas, these limitations introduce both operational constraints and potential sampling bias. This makes it difficult to fully evaluate the role of VHR imagery in conservation and management applications. Improving our understanding of tasking dynamics and securing imagery with a high likelihood of beluga presence are important next steps. To address this, we are initiating a complementary effort to acquire VHR satellite imagery through a higher priority tasking approach. By working toward more consistent coverage of high-use areas, this effort will enable more robust comparisons between VHR-derived density estimates and traditional aerial survey results. This will help clarify how satellite imagery can best support management needs.
About the Geospatial Artificial Intelligence For Animals
This initiative is a collaborative effort led by NOAA with the U.S. Geological Survey, Microsoft AI for Good, the Naval Research Laboratory, and other partners. We are building a secure cloud-based application that detects whales and other marine mammals in VHR satellite imagery. The system combines advanced image preprocessing and expert annotation to produce reliable training datasets for machine learning models. Initially focused on endangered whales such as North Atlantic right whales and Cook Inlet belugas, itis designed to expand into other taxa and habitats. We are creating a flexible, scalable platform for biodiversity monitoring. Turning raw satellite imagery into actionable information will enable monitoring of vast and remote ocean areas and strengthen conservation outcomes by providing managers and policymakers with timely insights.