2018 Big Sky GeoCon

Modern conservation efforts require coordination, planning, and monitoring across large landscapes. Under most ecosystem monitoring schemes, data is collected at a relatively small scale, but decisions need to be applied at the landscape scale. While remote sensing tools have greatly improved our ability to perform qualitative assessments at large scales, relatively few of these products provide actionable quantitative data for land managers and conservation planners.

Recent advances in machine learning algorithms, the advent of inexpensive drone technology, and readily accessible satellite data are together poised to revolutionize how we perform ecosystem monitoring. Here we discuss how these technologies can be combined to develop an effective monitoring framework at landscape scales. We use our work in grasslands of eastern Montana as a case study to show how we are scaling plot-level measurements (e.g. bare ground, plant volume, sage-brush cover) to pasture-scales using drones and to landscape-scales using time-series Landsat imagery. Our approach is designed to be nearly completely automated so that conservation staff do not need to spend their time classifying and modeling individual datasets.

In this presentation, we will discuss the tools that we use in our monitoring/modeling framework and share some of our successes and failures in applying machine learning techniques to imagery classification and regression modeling. We will also discuss tradeoffs between different photogrammetry processing tools, machine learning libraries, raster processing platforms, and point-cloud processing with LiDAR vs. SfM products. This talk will provide a high-level overview of our modeling framework and approach (50%) and a technical discussion of tools and data analysis (50%).

Adequate biomass feedstock supply is an important factor in evaluating the financial feasibility of alternative site locations for bioenergy facilities and for maintaining profitability once a facility is built. We used newly developed spatial analysis and logistics software to model the variables influencing feedstock supply and estimate and map two components of the supply chain for a bioenergy facility: 1) total biomass stocks available within an economically efficient transportation distance, and 2) the cost of logistics to move required stocks from the forest to the facility. Biomass stocks and flows both have important spatiotemporal dynamics that affect procurement costs and project viability. Though seemingly straight forward, these two components can be difficult to quantify and map accurately in a useful, spatially explicit manner. For a 20 million acre (8 million ha) study area, we use raster-based methods and tools to quantify and visualize these supply metrics at 32.8ft2 (10m2) spatial resolution. The methodology and software leverage a novel raster-based least cost path modeling algorithm that quantifies off road and on road transportation, and other logistics costs. Results of the case study highlight the efficacy, flexibility, fine resolution, and spatial complexity of model outputs developed for facility siting and procurement planning.

Montana Fish, Wildlife and Parks (FWP) biologists and managers are using ArcGIS Online to review hundreds of proposed Bureau of Land Management (BLM) oil and gas leases across the state and provide input on the potential impacts to fish, wildlife, and parks resources. FWP uses GIS analyses to identify leases with natural resource concerns not already addressed by BLM stipulations defined in BLM resource management plans. An ArcGIS Online story map guides FWP staff through the review and commenting process for each lease sale. An introduction section provides staff with context and additional resources. An embedded map application provides functionality for users to explore leases and their stipulations, find leases with additional resource concerns, and view resource data layers from the BLM, FWP, and other sources. Lastly, a crowdsource polling application allows staff to provide comments related to specific leases recommending additional stipulations to apply. The content presented in the story map makes the review process less tedious and more user-friendly for FWP biologists and managers. Land use planners from both agencies benefit from a streamlined workflow that results in fewer, more effective stipulation comments. Feedback resulting from the oil and gas lease review story map will help ensure that agencies are using the best available data when considering natural resource impacts of oil and gas development and will support continued coordination and data sharing.

Expanding ex-urban growth combined with increasing wildfire hazards throughout the country has lead to the demand for decision support for assessing wildfire hazards to communities. Classical risk management includes identifying the hazard, the susceptibility of assets to that hazard, and what can be done to mitigate adverse effects and/or maximize benefits. We developed a decision support tool for Lewis and Clark County's planning department and the Tri-county FireSafe working group to assess wildfire hazard and mitigation potential for residential structures. We assess wildfire hazard at two scales: a landscape scale representing the annualized most likely hazard and a local scale representing an extreme, worst case scenario event. In addition, we develop a hazard mitigation difficulty dataset using local vegetation cover and topographic characteristics. We demonstrate how the mitigation difficulty classes can be cross walked with the International Wildland Urban Interface building codes for mitigating home loss via wildfire through construction materials and site location restrictions. This information is presented in a spatial geodatabase for rapid assessment by local land-use planners and wildfire protection responders.

Backcountry skiers recreate in a complex environment, with the goal of minimizing the risk of avalanche hazard and maximizing recreational opportunities. Traditional backcountry outings start and end in uncontrolled backcountry settings, with a clear understanding that individuals are responsible for their own safety and rescue in the event of an accident. Lift access backcountry skiing (LABC) is a particular genre of the sport, in which ski resort lifts are utilized to access backcountry recreation sites. By shifting skiers mentality from the traditional setting to a LABC setting, the line between whether the ski resort provides avalanche mitigation and rescue services, or not, becomes less clearly defined in the minds of skiers.

This research observes the travel behavior and evaluates the decision-making biases of LABC skiers via GPS tracking and survey responses. A geographic information system (GIS) is implemented to analyze the travel behavior of participants, with the aim to detect changes in behavior, as indexed via terrain used under different levels of avalanche hazard. Logistic regression and multiple linear regression are used to model travel behavior and decision-making biases as a function of observed terrain metrics.

Data was collected over 18 days from February 2017 to February 2018 at Saddle Peak backcountry area, a prime LABC location at the southern boundary of Bridger Bowl Ski Area, Montana, USA. Results indicate subtle changes in the terrain preferences of participants under elevated avalanche hazard, with increased travel on ridge features and decreased travel on convex features. These indicate a positive response, minimizing the risk of an avalanche involvement by managing slope shape. Survey responses indicate that gender, backcountry experience, and perception of avalanche mitigation have significant effects on percent of GPS track in complex avalanche terrain.