Testing a Temperature Inversion Correction Model for Downscaled Reanalysis Temperatures in Subarctic Yukon Valleys

Abstract

Surface-based temperature inversions (SBIs) strongly influence near-surface temperatures in subarctic mountain valleys, yet they are poorly represented in coarse-resolution reanalysis datasets. This study evaluated the performance and physical robustness of the Pozsgay and Gruber (2025) inversion correction model in two subarctic valleys along the Dempster Highway, Yukon. The model dynamically adjusts reanalysis air temperature at each timestep using a correction factor derived from the free lower-tropospheric lapse rate and estimated SBI strength allowing for temperature-elevation relationships to vary in response to changing SBI conditions. When calibrated locally, the model reduced mean absolute temperature errors up to 0.8 °C and bias to within ±0.2 °C at the valley scale. Performance was strongest under moderate SBI strength but declined for shallow or very strong SBIs due to limited reanalysis vertical resolution and the assumption of linear lapse rates. Locally calibrated models outperformed transferred ones, indicating sensitivity to local SBI characteristics and limitations associated with absolute elevation-based calibration. Despite these limitations, explicitly accounting for SBI dynamics substantially improves temperature estimates in high-latitude complex terrain. The resulting corrected datasets provide a valuable input for transient permafrost modelling and assessing how changing SBI characteristics may influence ground thermal regimes in subarctic alpine environments.