Extreme Temperature Events: Summer Heat Waves and Frost Days in Uruguay-Southeastern South America, observed changes during 1950–2009

There are few studies about extreme temperature events in Southeastern South America as is it mentioned in the SREX report (2009), although these events generate human health impacts and big economical looses. Southeastern South America is one of the major agricultural production regions worldwide. Particularly in Uruguay, agricultural production represents a high percentage of the GDP and, in the last 15 years there has been a significant increase in the area used for that economic activity. This work focus on two types of extreme events: summer heat waves and frost events, using a high quality daily maximum and minimum temperature database from 11 meteorological stations in Uruguay covering the period 1950-2009. Although frost is not always is considered as an extreme event it causes, in the case of Uruguay, an impact on society, energy consumption and agricultural losses. Previous studies have shown a negative trend in the occurrence of cold nights (TN10) during winter (June-July-August) and autumn (March-April-May) for the period 1950-2005, with the cold nights index calculated as percentage of days in a month below the 10th percentile. Considering that during winter, in Uruguay, the 10th percentile of the minimum temperature is above 0º C, the occurrence of frost days (Tmin < 0ºC) is included in the cold nights index. The frost period was the defined considering the date of the first and last frost recorded for each time series of minimum temperature. The common frost periods extend from May to September. Our results show, overall, that during the first decades of the study period (1950-1970) occur 15 to 20 events while in the last decades (1990-2009) no more than 10 frost events are observed during the months May-September. It is important to mention that the southern stations, located in the coastal region of the country, are those that present frost-free years, suggesting the impact of coastal effects on temperature. On the other hand stations located in the north and northeast of the country show a negative trend in the occurrence of frost days. At monthly scale, May and September show a negative trend, although these months present a low number of cases that difficult the statistical treatment. It is noticeable that from a decadal point of view the last decade (2000-2009) was the decade with fewer occurrences comparing with the rest, while the 90xs is the decade that presents more cases. Regarding the frost intensity, we analyzed the consecutive frost days (CFD) index, which shows that during the first period events lasting 4 or 5 consecutive days were common, while in the last decades isolated events are detected.

With regards to the analysis of heat waves, we focus on summer (December-January-February) events. Different definitions of heat waves can be found in the literature; here, we define a heat wave based on maximum and minimum temperature, as follows. Based on an autocorrelation analysis a 5-day window was used to determine the daily 90th percentile of each temperature series, from December to February for the period analyzed (1950-2009). The heat wave (HW) was defined when during 3 or more days the daily maximum and minimum temperature exceed the long-term 90th percentile, not interrupted by more than one day where one of the variables does not fulfil these conditions. A linear trend analysis shows that there is no significant change in the number of occurrence of HW; only one station presents a positive significant trend. Renom (2009) detected a negative significant trend in the warm days index (TX90) during summer. However, the warm nights index (TN90) does not present a trend, which could be one of the reasons why the HW doesn't present a significant trend conditioned by the minimum temperature.