This study investigates trends in the diurnal temperature range (DTR), conditional on the occurrence of daily temperature thresholds exceedences (TTEs). DTR trends are determined on extreme (temperatures exceeding the 95^th percentile) and non-extreme days for both daily maximum and minimum temperatures from a 353-station subset of the U.S. Historical Climate Network. A variety of DTR trend comparisons are presented between different periods of record (1930-96, 1950-96, 1970-96), types of stations (rural, suburban, urban) and types of day (extreme, non-extreme). Seasonal differences are also explored by comparing DTR trends that occur on days with warm (summer) or cold (winter) exceedences of daily maximum and minimum temperatures. The measurement of the trend in DTR is determined at each station by the Student’s t-test, based on the mean of the first difference series computed from equally weighted 10-year running means. The significance of the computed t-statistics is assessed through Monte Carlo techniques.

When analyzing trends conditional on warm TTEs (summer), a significant (a=0.05) percentage of stations have significant negative DTR trends during the periods 1930-96 and 1950-96. These results are independent of the type of station, the type of day or the type of temperature variable (maximum or minimum) analyzed. During the most recent period of 1970-96, the percentage of stations with significant negative DTR trends on warm TTE days decreases and only maintains its significance at urban stations. On days with non-extreme temperatures, significance is maintained independent of the station type. Very little significance is observed in DTR trends conditional on cold maximum TTEs (winter), however those conditional on cold minimum TTEs are highly significant for both suburban and urban stations during each of the three periods analyzed. DTR trends at urban stations are consistently more negative than DTR trends at rural stations, however the significance in these station type differences is highest when analyzing the trends conditional on warm maximum TTEs. Differences in DTR trends on days with extreme and non-extreme temperatures are observed to be highly variable between the period and type of variable analyzed. In the most recent period, DTR trends on days with non-extreme temperatures are significantly more negative than trends on days with extreme temperatures during the summer. During the winter, DTR trends on days with non-extreme temperatures are insignificant (near zero) while trends on days with cold TTEs are negative. Regional differences across the U.S. also appear. There is a tendency for greater negative trends in DTR in the central U.S., and less negative or sometimes positive DTR trends in the northeast and western U.S. regions.