22 Nov 2021

Imagery Classification for Autonomous Ground Vehicle Mobility in Cold Weather Environments

Abstract: Autonomous ground vehicle (AGV) research for military applications is important for developing ways to remove soldiers from harm’s way. Current AGV research tends toward operations in warm climates and this leaves the vehicle at risk of failing in cold climates. To ensure AGVs can fulfill a military vehicle’s role of being able to operate on- or off-road in all conditions, consideration needs to be given to terrain of all types to inform the on-board machine learning algorithms. This research aims to correlate real-time vehicle performance data with snow and ice surfaces derived from multispectral imagery with the goal of aiding in the development of a truly all-terrain AGV. Using the image data that correlated most closely to vehicle performance the images were classified into terrain units of most interest to mobility. The best image classification results were obtained when using Short Wave InfraRed (SWIR) band values and a supervised classification scheme, resulting in over 95% accuracy.

22 Nov 2021

Methodology for the Analysis of Geospatial and Vehicle Datasets in the R Language

Abstract: The challenge of autonomous off-road operations necessitates a robust understanding of the relationships between remotely sensed terrain data and vehicle performance. The implementation of statistical analyses on large geospatial datasets often requires the transition between multiple software packages that may not be open-source. The lack of a single, modular, and open-source analysis environment can reduce the speed and reliability of an analysis due to an increased number of processing steps. Here we present the capabilities of a workflow, developed in R, to perform a series of spatial and statistical analyses on vehicle and terrain datasets to quantify the relationship between sensor data and vehicle performance in winter conditions. We implemented the R-based workflow on datasets from a large, coordinated field campaign aimed at quantifying the response of military vehicles on snow-covered terrains. This script greatly reduces processing times of these datasets by combining the GIS, data-assimilation and statistical analyses steps into one efficient and modular interface.

1 Oct 2021

Characterizing Snow Surface Properties Using Airborne Hyperspectral Imagery for Autonomous Winter Mobility

Abstract: With changing conditions in northern climates it is crucial for the United States to have assured mobility in these high-latitude regions. Winter terrain conditions adversely affect vehicle mobility and, as such, they must be accurately characterized to ensure mission success. Previous studies have attempted to remotely characterize snow properties using varied sensors. However, these studies have primarily used satellite-based products that provide coarse spatial and temporal resolution, which is unsuitable for autonomous mobility. Our work employs the use of an Unmanned Aerial Vehicle (UAV) mounted hyperspectral camera in tandem with machine learning frameworks to predict snow surface properties at finer scales. Several machine learning models were trained using hyperspectral imagery in tandem with in-situ snow measurements. The results indicate that random forest and k-nearest neighbors models had the lowest Mean Absolute Error for all surface snow properties. A Pearson correlation matrix showed that density, grain size, and moisture content all had a significant positive correlation to one another. Mechanically, density and grain size had a slightly positive correlation to compressive strength, while moisture had a much weaker negative correlation. This work provides preliminary insight into the efficacy of using hyperspectral imagery for characterizing snow properties for autonomous vehicle mobility.

5 Aug 2021

Changes in Climate and Its Effect on Timing of Snowmelt and Intensity-Duration-Frequency Curves

Abstract: Snow is a critical water resource for much of the U.S. and failure to ac-count for changes in climate could deleteriously impact military assets. In this study, we produced historical and future snow trends through modeling at three military sites (in Washington, Colorado, and North Dakota) and the Western U.S. For selected rivers, we performed seasonal trend analysis of discharge extremes. We calculated flood frequency curves and estimated the probability of occurrence of future annual maximum daily rainfall depths. Additionally, we generated intensity-duration-frequency curves (IDF) to find rainfall intensities at several return levels. Generally, our results showed a decreasing trend in historical and future snow duration, rain-on-snow events, and snowmelt runoff. This decreasing trend in snowpack could reduce water resources. A statistically significant increase in maximum streamflow for most rivers at the Washington and North Dakota sites occurred for several months of the year. In Colorado, only a few months indicated such an increase. Future IDF curves for Colorado and North Dakota indicated a slight increase in rainfall intensity whereas the Washington site had about a twofold increase. This increase in rainfall in-tensity could result in major flood events, demonstrating the importance of accounting for climate changes in infrastructure planning.

21 Jul 2021

Machine Learning Analyses of Remote Sensing Measurements Establish Strong Relationships Between Vegetation and Snow Depth in the Boreal Forest of Interior Alaska

Abstract: The seasonal snowpack plays a critical role in Arctic and boreal hydrologic and ecologic processes. Though snow depth can be different from one season to another there are repeated relationships between ecotype and snowpack depth. Alterations to the seasonal snowpack, which plays a critical role in regulating wintertime soil thermal conditions, have major ramifications for near-surface permafrost. Therefore, relationships between vegetation and snowpack depth are critical for identifying how present and projected future changes in winter season processes or land cover will affect permafrost. Vegetation and snow cover areal extent can be assessed rapidly over large spatial scales with remote sensing methods, however, measuring snow depth remotely has proven difficult. This makes snow depth–vegetation relationships a potential means of assessing snowpack characteristics. In this study, we combined airborne hyperspectral and LiDAR data with machine learning methods to characterize relationships between ecotype and the end of winter snowpack depth. Our results show hyperspectral measurements account for two thirds or more of the variance in the relationship between ecotype and snow depth. An ensemble analysis of model outputs using hyperspectral and LiDAR measurements yields the strongest relationships between ecotype and snow depth. Our results can be applied across the boreal biome to model the coupling effects between vegetation and snowpack depth.

19 Jul 2021

A Pulse of Mercury and Major Ions in Snowmelt Runoff from a Small Arctic Alaska Watershed

Abstract: Atmospheric mercury (Hg) is deposited to Polar Regions during springtime atmospheric mercury depletion events (AMDEs) that require halogens and snow or ice surfaces. The fate of this Hg during and following snowmelt is largely unknown. We measured Hg, major ions, and stable water isotopes from the snowpack through the entire spring melt runoff period for two years. Our small (2.5 ha) watershed is near Barrow (now Utqiaġvik), Alaska. We measured discharge, made 10 000 snow depths, and collected over 100 samples of snow and meltwater for chemical analysis in 2008 and 2009 from the watershed snowpack and ephemeral stream channel. Our results suggest AMDE Hg complexed with Cl− or Br− may be less likely to be photochemically reduced and re-emitted to the atmosphere prior to snowmelt, and we estimate that roughly 25% of the Hg in snowmelt is attributable to AMDEs. Projected Arctic warming, with more open sea ice leads providing halogen sources that promote AMDEs, may provide enhanced Hg deposition, reduced Hg emission and, ultimately, an increase in snowpack and snowmelt runoff Hg concentrations.

30 Jun 2021

Mercury Isotopes Reveal Atmospheric Gaseous Mercury Deposition Directly to the Arctic Coastal Snowpack

Abstract: Springtime atmospheric mercury depletion events (AMDEs) lead to snow with elevated mercury concentrations (>200 ng Hg/L) in the Arctic and Antarctic. During AMDEs gaseous elemental mercury (GEM) is photochemically oxidized by halogens to reactive gaseous mercury which is deposited to the snowpack. This reactive mercury is either photochemically reduced back to GEM and re-emitted to the atmosphere or remains in the snowpack until spring snowmelt. GEM is also deposited to the snowpack and tundra vegetation by reactive surface uptake (dry deposition) from the atmosphere. There is little consensus on the proportion of AMDE-sourced Hg versus Hg from dry deposition that is released in spring runoff. We used mercury stable isotope measurements of GEM, snowfall, snowpack, snowmelt, surface water, vegetation, and peat from a northern Alaska coastal watershed to quantify Hg sources. Although high Hg concentrations are deposited to the snowpack during AMDEs, we estimate that ∼76 to 91% is released back to the atmosphere prior to snowmelt. Mercury deposited to the snowpack as GEM comprises the majority of snowmelt Hg and has a Hg stable isotope composition similar to Hg deposited by reactive surface uptake of GEM into the leaves of trees in temperate forests. This GEM-sourced Hg is the dominant Hg we measured in the spring snowpack and in tundra peat permafrost deposits.

12 May 2021

Wintertime Snow and Precipitation Conditions in the Willow Creek Watershed above Ririe Dam, Idaho

ABSTRACT:  The Ririe Dam and Reservoir project is located on Willow Creek near Idaho Falls, Idaho, and is important for flood risk reduction and water supply. The current operating criteria is based on fully storing a large winter runoff event. These winter runoff events are generally from large storm events, termed atmospheric rivers, which produce substantial precipitation. In addition to the precipitation, enhanced runoff is produced due to frozen soil and snowmelt. However, the need for additional water supply by local stakeholders has prompted the U.S. Army Corps of Engineers to seek to better understand the current level of flood risk reduction provided by Ririe Dam and Reservoir.  Flood risk analysis using hydrologic modeling software requires quantification of the probability for all of the hydrometeorologic inputs. Our study develops the precipitation, SWE, and frozen ground probabilities that are required for the hydrologic modeling necessary to quantify the current winter flood risk.

4 May 2021

Microscale Dynamics between Dust and Microorganisms in Alpine Snowpack

ABSTRACT:  Dust particles carry microbial and chemical signatures from source regions to deposition regions. Dust and its occupying microorganisms are incorporated into, and can alter, snowpack physical properties including snow structure and resultant radiative and mechanical properties that in turn affect larger-scale properties, including surrounding hydrology and maneuverability. Microorganisms attached to deposited dust maintain genetic evidence of source substrates and can be potentially used as bio-sensors. The objective of this study was to investigate the impact of dust-associated microbial deposition on snowpack and microstructure. As part of this effort, we characterized the microbial communities deposited through dust transport, examined dust provenance, and identified the microscale location and fate of dust within a changing snow matrix. We found dust characteristics varied with deposition event and that dust particles were generally embedded in the snow grains, with a small fraction of the dust particles residing on the exterior of the snow matrix. Dust deposition appears to retard expected late season snow grain growth. Both bacteria and fungi were identified in the collected snow samples.

1 Dec 2020

Snow-Covered Obstacles’ Effect on Vehicle Mobility

ABSTRACT:  The Mobility in Complex Environments project used unmanned aerial systems (UAS) to identify obstacles and to provide path planning in forward operational locations. The UAS were equipped with remote-sensing devices, such as photogrammetry and lidar, to identify obstacles. The path-planning algorithms incorporated the detected obstacles to then identify the fastest and safest vehicle routes. Future algorithms should incorporate vehicle characteristics as each type of vehicle will perform differently over a given obstacle, resulting in distinctive optimal paths. This study explored the effect of snow-covered obstacles on dynamic vehicle response. Vehicle tests used an instrumented HMMWV (high mobility multipurpose wheeled vehicle) driven over obstacles with and without snow cover. Tests showed a 45% reduction in normal force variation and a 43% reduction in body acceleration associated with a 14.5 cm snow cover. To predict vehicle body acceleration and normal force response, we developed two quarter-car models: rigid terrain and deformable snow terrain quarter-car models. The simple quarter models provided reasonable agreement with the vehicle test data. We also used the models to analyze the effects of vehicle parameters, such as ground pressure, to understand the effect of snow cover on vehicle response.