Near-Surface Air Temperature in Complex Terrain: Daily Predictions of Fine-Scale (30 m) Temperature in the Snake Range, Nevada, USA
AdvisorAlbright, Thomas P
StatisticsView Usage Statistics
Air temperature is arguably the most important component of the mountain climate, and scientists have been studying it for centuries. Recently, researchers have used arrays of inexpensive temperature sensors to observe and understand temperature across the landscape. Much of this work has focused on landscape scale features as drivers of local air temperature's divergence from the greater regional air mass. To this end, we conducted an empirical orthogonal function analysis of gridded sea level pressure (SLP) from 1951-2014 for a spatial window including the eastern Pacific and western North America, which identified a mode of SLP variability that well describes synoptic weather in our study area. Pressure patterns and NCEP Reanalysis 1 derived regional air temperature were linked with a network of 40 temperature sensors spanning June 2013-2014 and GIS derived landscape variables to create hierarchical-mixed effects models of daily minimum and maximum temperature in the Snake Range. Minimum temperatures were mostly linked to elevation and the shape of the landscape, as cold air drainage is a common process in the Snake Range. Maximum temperature was largely related to insolation and elevation, with a large seasonal component. We used these models to create maps coinciding with daily Snake Range Sensor Network readings of minimum and maximum temperature in the Snake Range at 30 m spatial resolution. The map predictions were validated using an independent dataset and a leave-one-out cross validation. Overall mean absolute error for minimum and maximum temperature were 1.92 and 2.78 °C when calculated with the independent dataset and 1.84 and 2.21 °C when calculated using the leave-one-out cross validation. Together, these results show that temperature regimes in complex terrain vary considerably over short distances and short periods of time. It is possible to create a more realistic representation of maximum and minimum temperature for a particular study area, and the topoclimate of maximum and minimum temperature identified here can likely be applied to similar systems.