Forward-looking infrared (FLIR) surveys conducted from helicopters provide spatially continuous temperature profiles across tens to hundreds of river kilometers, but the extended duration of these surveys means that temperature readings at different locations are collected at different times of day—conflating spatial patterns with diel thermal dynamics. We developed and validated a statistical method for correcting FLIR-derived temperature profiles to account for the confounding effects of diel temperature variation during airborne surveys. We hypothesized that FLIR observations, combined with information about the time of data collection, could accurately predict temperature metrics such as daily maximum temperature at any surveyed location -- even where temperature readings were collected hours before or after the daily temperature peak. We deployed stationary temperature loggers at multiple sites along the Umatilla River (Oregon, USA) during FLIR surveys and used these data to build a multiple regression model predicting daily maximum water temperature from three variables: the FLIR-derived water temperature (extracted via quantile spline regression), the hour of day when each FLIR observation was collected, and a categorical year term to account for annual temperature variation and other factors, such as sensor calibration differences for surveys conducted a decade apart. The regression model explained 98% of the variance in observed daily maximum temperatures (R² = 0.98, RMSE = 0.49°C), with all predictor variables highly significant (p < 0.001). The hour-of-day coefficient confirmed that FLIR observations collected earlier in the day required larger upward adjustments to estimate daily maximum temperature, consistent with expected diel warming patterns. Our approach provides a template for transforming FLIR surveys from point-in-time temperature readings collected across both time and space into estimates of temperature metrics that are more meaningful for river research and management. Our approach is transferable to other FLIR applications and provides a framework for maximizing the inferential value of airborne thermal surveys in stream temperature monitoring programs.