LAUSR

Freezing temperatures are responsible for population and ecosystem damage, and their spatial extent, frequency, and duration may change in a warming world. Few studies are available concerning the analysis of extreme freezing in Mexico, and none have evaluated the probability of occurrence. This study identified the characteristics of past and recent events by applying the statistical crossing theory method to calculate the cumulative density function (CDF) of the minimum temperature (Tmin) values throughout the Sonoran Desert, including the Baja California Peninsula. Based on observations, spatial and temporal freezing evolution has been described during the period from 1970 to 2010, for which Tmin is less than 0°C and less than the fifth percentile of the monthly Tmin. Since 1970, a quasi‐homogeneous decrease has been observed in the number of freezing days. Extreme freezing was less frequent and severe during the early 1990s, which continued during the early 21st century when intense and persistent short events exhibited return periods above 20 years. After the 1990s, longer recurrence intervals for extreme Tmin, which overlapped with the increase in Tmin, implied a change in stationarity. This study found that cold‐season warming lead to less‐frequent severe freezing and a probability of change. These results represent a quantitative benchmark for current and upcoming validations of climate change scenarios.