Using two laboratory-scale conceptual fluid dynamic models of the mid-latitude atmospheric circulation we investigate the statistical properties of pointwise temperature signals obtained in long experiment runs. We explore how the average "equator-to-pole" temperature contrast influences the range and the jump distribution of extreme temperature fluctuations, the ratio of the frequencies of rapid cooling and warming events, and the persistence of "weather" in the set-ups. We find simple combinations of the control parameters-temperature gradient, rotation rate and geometric dimensions-which appear to determine certain scaling properties of these statistics, shedding light on the underlying dynamics of the Rossby wave-related elements of the mid-latitude weather variability.