Microclimates are fine-scale departures from bulk conditions often attributed to static landscape characteristics and topography, but their temporal variability in response to dynamic meteorological conditions remains poorly understood. In this study, we deployed a network of air temperature (T-air) and relative humidity (RH) sensors across a small grassland watershed in northern Colorado, USA to assess microclimate variability, atmospheric stability and concurrent meteorological conditions. We found significant within-field spatial variability in T-air at 2 m fluctuating by >15 degrees C, RH by >50 %, and vapor pressure deficit by > 1 kPa at 15 min intervals. The mean T-air difference between the highest and lowest points in the watershed was 0.29 degrees C, or a near-surface lapse rate of 10 degrees C km(-1), exceeding the free-air lapse rate of 6.5 degrees C km(-1). Within-field variability was driven primarily by atmospheric stability (defined by Richardson number) and was highest during stable or inversion conditions when mechanical turbulence and convective mixing were low. These variations in T-air and RH scaled up to annual variability in biophysical metrics within the field, such as growing degree days and potential evapotranspiration, that exceeded 7 %. An inverse relationship between within-field T-air mean and variability suggests warmer temperatures in the future may minimize microclimates in similar areas with low topographic complexity to this study. This study advances knowledge by identifying important temporal dynamics in microclimates within a grassland field that vary with atmospheric stability. These dynamics should be incorporated into microclimate models and future studies investigating microclimate effects in ecological, hydrological, and agricultural applications.