An in situ-based analysis of the relationship between land surface skin and screen-level air temperatures

This paper presents an analysis of the relationship between land surface temperatures (LST) and screen-level air temperatures (T-2m) using in situ observations from 19 Atmospheric Radiation Measurement (ARM) deployments located in a range of geographical regimes. The diurnal cycle is resolved using 1min observations: a particular focus of the study is on the relationship between daily extremes of LST (LSTmax, LSTmin) and T-2m (T-max, T-min). Temperature differences are analyzed with respect to cloud, wind speed, and snow cover. Under cloud-free, low wind speed conditions, daytime LST is often several degrees Celsius (degrees C) higher than T-2m at low-to-middle latitudes and at high latitudes during the summer months. In contrast, LST and T-2m are often close (e.g., within 2 degrees C) under cloudy and/or moderate-to-high wind speed conditions or when solar insolation is low or absent. LSTmin and T-min are generally well correlated (r>0.8, often r>0.9), while seasonal correlations between LSTmax and T-max are weaker (r>0.6, often r>0.8). At high latitudes, LST and T-2m are well coupled in spring/autumn/winter; the relationship between LST and T-2m tends to weaken with decreasing latitude. The timing of daily extremes is also investigated and it is found that LSTmin and T-min typically occur close to sunrise, with T-min occurring slightly after LSTmin. LSTmax occurs close to solar noon, with T-max typically occurring 1-3 hours later. This study will inform temperature data users on differences between LST and T-2m and aid development of methods to estimate T-2m using satellite LSTs.