Multi-scale advancements in optical remote sensing for snow and water applications

Abstract

Snow occupies a large portion of the Earth's surface and represents an imperative hydrological resource, a severe hazard, and a key component in our planetary energy budget. The high reflectivity and albedo of snow exerts control on Earth's radiative energy balance, with a direct effect on global climate. Further, snowmelt accumulates in headwater channels; the timing of this phenomenon is critical to ecological health, water quality, downstream water supply, and watershed connectivity. Last, snow presents a serious hazard to both human life and infrastructure, namely in the form of avalanches. Hundreds of lives have been lost to snow avalanches in the past decade. Therefore, accurate measurements of physical snowpack properties and runoff are critical. A burgeoning era of remote sensing is allowing for more continuous measurements with rapidly increasing spatial, spectral, and temporal resolutions. As troves of data become available, researchers are developing new methodologies to keep pace. In the work presented here, we join these efforts. The broad objective of our work was to use near-infrared (NIR) lidar and hyperspectral imaging to create maps of physical snowpack properties and headwater channels. We hypothesize that underutilized instruments and spatial analysis concepts can address limitations in snowpack mapping techniques. Using a cold laboratory and the natural laboratory of Montana's mountains, we examined three primary concepts: snow grain size, avalanche hazard assessment, and forested headwater channels. Regarding grain size, this parameter controls snow albedo and can be estimated from reflectance, although numerous methods towards this end produce dramatically disparate results. We performed an intercomparison, contrasting models and retrieval techniques to find the optimal method. Further, we validated the novel use of lidar for this purpose, with significant advantages over traditional passive detection. Towards avalanche forecasting, we created a technique that leverages NIR texture to map surface hoar, responsible for over a third of large avalanches. Additionally, we synthesized lidar data with dynamics modeling to predict threats to roadways. Last, towards improved streamflow monitoring, we developed a new means of mapping streams hidden in densely forested regions via airborne lidar. The methods presented here will aid practitioners across multiple fields of the cryosphere.