Remote sensing of mountain snow surface temperatures at high temporal resolution using geostationary satellites

Abstract

Remote sensing by geostationary satellites, such as by the NOAA GOES-R series with the Advanced Baseline Imager (ABI), can provide imagery of surface temperatures at high temporal resolutions due to their fixed views of Earth’s surface. In mountain headwaters that receive seasonal snow, spatially distributed observations of snow surface temperatures are needed to better constrain estimates of the surface energy balance, predictions of snowmelt, and available water resources. These observations are needed particularly at spatial and temporal resolutions relevant to land surface and hydrology models, a capability that the 5-minute, near-real-time GOES-R observations may be able to fill. The utility of these observations, however, may be limited by their relatively coarse (2+ km) spatial resolution, and the off-nadir view angles of geostationary satellites.In this dissertation, we found that the off-nadir views, surface roughness at different spatial scales, and changing direction of solar illumination over the course of a day, all impact the surface temperatures observed by GOES-R ABI. In Chapter 2, we demonstrated that off-nadir geostationary satellite imagery must be corrected for the parallax effect in mountainous areas. Even with this correction, the surface temperatures observed by GOES-16 were biased towards those of warmer sunlit south-facing mountain slopes that were facing the satellite. Chapter 3 provides information about the software developed to correct for the parallax effect in GOES-R ABI imagery. In Chapter 4, as part of the NASA SnowEx 2020 field campaign, we found that at the scale of forest stands and individual trees across a snow-covered area, the surface temperatures observed by GOES-R ABI were biased towards that of the warmer tree temperatures in comparison with coincident nadir-looking imagery. This warm bias was greatest at times of day when the sun-satellite phase angle was at its minimum, suggesting a diurnal thermal infrared shadow hiding effect where cold shadows are briefly hidden from view by the warmer trees. Chapter 5 extended this analysis of shadow hiding to the midwave infrared, which in the daytime has both an emitted and reflected solar component. The shadow hiding effect was also found in the midwave infrared imagery, however not just for forested areas but also for snow surfaces with centimeter-scale wind-formed roughness features. This demonstrated that for applications of midwave infrared observations, rather than only treating the surface reflectance of different materials (e.g., snow and vegetation) independently, they must also be considered together as an anisotropic reflector and emitter of midwave infrared radiation.