The Great Basin Institute, in cooperation with agency partner The National Parks Service, is recruiting one Remote Sensing Vegetation Specialist to participate in development and implementation of new automated methods for efficient processing and analysis of remote sensed imagery and related spatial data.
The North Coast and Cascades Network (NCCN) is one of 32 NPS I&M networks across the country established to facilitate collaboration and information sharing and efficient data collection and analysis across geographically related NPS units. NCCN uses Landsat imagery and the LandTrendr algorithm to monitor long-term changes in the frequency, magnitude, and distribution of landscape disturbances in network parks. Individual disturbance polygons are labeled with disturbance type (avalanche, landslide, riparian change, etc.) using aerial photo interpretation and evaluation of disturbance spectral trajectories. Changes in the number and area of different disturbance types are tracked over time to assess effects of extreme weather events and climate change on natural disturbance dynamics in protected areas. To date, review and validation of vegetation disturbances using aerial imagery has required considerable time, and recent efforts to develop models for automated labeling of disturbance patches have proved successful. One of the objectives of this position would be to complete an assessment of model performance using a test set of NCCN data, and to fine-tune the model application code base so that the approach can be finalized and implemented for NCCN and other networks. Products created during the position will also be used to create educational materials for park managers and the public to further explore and understand the long-term trends in NCCN natural resources.
Work Setting and Local Area Details:
This position will entail close collaboration with a small team of NPS specialists and will involve a combination of independent work and routine meetings. Work will be performed primarily indoors in front of a computer, in a park office building in a shared space. If interested, the incumbent will also have opportunities to join park staff and NCCN Inventory and Monitoring team members for occasional field work and site visits, where work may occur outdoors in a variety of terrain and may require travel on steep rocky slopes, in forests, meadows, streams, and wetland environments in a variety of weather conditions. Work may also include travel to other national parks in the area. Because the position is primarily office work, however, applicants that are unable to conduct field work are still encouraged to apply.
Deliverables and Work Products:
At the completion of the position, deliverable work products will include: 1) final R code base and associated documentation for automated labeling of disturbance types in large parks; 2) an accuracy assessment report for comparison between Landsat-based and high-resolution change detection products; 3) a report or manuscript summarizing methods and results of error rates between human-validated and automated labels; and 4) a document containing standard operating procedures and scripts for tracking vegetation phenology or snow extent/quantity in GEE.
Primary Responsibilities:
- Work collaboratively with network staff to complete development of a machine learning algorithm to label landscape change polygons in NCCN parks.
- Apply coding skills in reproducible data processing and analysis to address ecological questions.
- Test and facilitate the adoption of new technologies and data streams in remote sensing for ongoing inventory and monitoring programs.
- Use GIS software (ArcGIS Pro, Google Earth Engine) to develop and assemble geometric, topographic, and spectral predictor variables for modeling.
- Update existing code to automate vegetation disturbance label attribution in wilderness park lands using human-reviewed labels to train and test the model.
- Review disturbance patches generated by LandTrendr against other imagery sources and assign labels to individual disturbance patches using pre-defined categories and criteria.
- Summarize error rates between human-validated labels and automated labels.
- Compare high resolution change detection employed by other agencies with our current approaches and quantify how well the higher-resolution imagery performs.
- Help to coauthor a manuscript summarizing results.
- As time permits, apply models developed for remote wilderness areas to smaller parks with more human influence (e.g., Lewis and Clark National Historical Park) to evaluate efficacy, make updates to models to increase label attribution accuracy.
- Explore approaches for imagery-based modeling of seasonal phenology and snowpack in high elevation systems using Google Earth Engine (GEE).
Timeline:
- Starting in January 2025, 24 weeks
- Full time (40 hours/week)
- Schedule: Monday-Friday, 8 hours daily
Location:
- Port Angeles, WA or Bellingham, WA (hybrid/remote options negotiable)
Compensation and Benefits:
- Compensation: $27 hourly
- Benefits: Company-paid comprehensive medical, dental (with option to upgrade in coverage), and vision insurance.
- $25,000 Basic Life & AD&D insurance at no cost
- Competitive PTO accrual and paid holidays
