Identifying a Trend in Historical and Modeled Wet Bulb

Engineering design specifications for cooling and dehumidification systems rely on the wet bulb temperature climatology of the location of interest. Specifically, information on historical wet bulb extremes (< 2% annual cumulative frequency of occurrence). Typically, these climatologies have been developed based on the observed hourly wet bulb temperature records contained in the Integrated Surface Dataset. Both the hourly resolution of these data and the lack of direct observations complicate assessment of future changes in extreme wet-bulb temperatures that may impact the efficiency and reliability of cooling systems.

For a set of climatologically diverse U.S. stations, wet-bulb temperatures are calculated from relative humidity and dry bulb temperature using different techniques. These variables are readily available from existing downscaled climate model simulations. The calculated wet-bulb temperatures are first compared to direct wet-bulb temperature observations to identify the most appropriate method for wet-bulb approximation. This comparison is performed across multiple wet-bulb temperature ranges since some methods are better at estimating wet-bulb in certain conditions. To replicate the temporal resolution of available downscaled climate model projections, daily maximum and minimum wet-bulb temperatures were extracted and a cubic-spline used to interpolate the hourly values.

After accounting for an average bias of up to ±2°C introduced by the computation of hourly wet bulb temperature based on observed temperature and relative humidity, historical modeled and downscaled temperature and humidity values are compared to the observed climatologies. In general, the modeled values are similar to observations, with the largest negative biases were associated with the highest temperature range, particularly at Tucson. The largest positive biases tend to occur at colder temperatures, particularly in more humid locations such as Seattle. The spline fits to the daily maximum and minimum wet-bulbs provide a viable method for inferring hourly values from the projected data. Counts of hours above an extreme wet-bulb threshold are well simulated using this approach both in terms of the number of annual counts and the trend through the 1950-2006 historical period.