Physically based models of glacier melt require fields of near‐surface air temperature (Tg) and vapor pressure (eg) for estimating turbulent heat exchanges. However, katabatic boundary layer (KBL) processes limit the effectiveness of standard interpolation or extrapolation routines for estimating Tg and eg from regional weather station networks. Climate data collected from nine automatic weather stations operated over three ablation seasons at three glaciers in the southern Coast Mountains of British Columbia are analyzed in this study. On‐glacier observations were compared to ambient values (Ta and ea) estimated from a regional network of off‐glacier weather stations. Piecewise regressions of Tg versus Ta at each AWS site reveal (1) a critical threshold temperature (T*) that denotes the onset of katabatic boundary layer (KBL) development and (2) a temperature damping that is consistent at each site, but variable between sites. Variations in near‐surface vapor pressure are related to processes of condensation or evaporation/sublimation at the glacier surface, which are controlled by the vapor pressure gradient between the surface and the ambient air. Statistical relations with flow path lengths calculated from glacier digital elevation models are used to predict the strength of KBL effects on Tg and eg, and examples of the approach for generating distributed fields of Tg and eg are given.