SAMPLING DESIGN:
Daily maximum and minimum temperatures from up to eight locations at HBEF were measured from October, 1955 through February, 2013 using mechanical hygrothermographs in weather shelters at seven locations. Six locations are in rain gage openings: 1/1A, 6, 14, 17, 23, and 24. One location is adjacent to the Headquarters building. At each station, air temperature (and usually relative humidity) was recorded by Belfort hygrothermographs housed in standard Stevenson screens, 5 feet off the ground. Standard, maximum and minimum thermometers were housed in the shelters alongside the hygrothermographs for weekly calibration checks of the current, max and min temperatures captured on the chart. The hygrothermograph charts were changed weekly, at which time the thermometers were read. Daily maximum and minimum temperatures were read manually from the charts. Correction for disagreement with the standard thermometers were seldom made as agreement is usually within 2 degrees F. The hygrothermographs were readjusted annually and rotated from site to site after lab calibration to prevent instrument bias. Maximum and minimum temperatures were converted from Fahrenheit to Celsius. Average daily temperature is calculated from the maximum and minimum.
Beginning during 2011, Campbell Scientific HMP50/60 or CS215 thermistors in six-plate Gill shields recorded daily maximum and minimum temperatures at each location. Gill shields were replaced with Apogee aspirated shields by mid-2016. Digital sensors and hygrothermographs were operated in the same clearing for an overlap period of at least one full year at all stations, and adjustments for changes in sensor and radiation shield technology were made by using linear interpolation to smooth median monthly differences to a daily scale. Prior to the transition to digital sensors, if a weather station was not operating, the value for that day was estimated from values at the other stations and was not not flagged. Currently, missing values are estimated according to linear regression with the best matched location and are indicated with a flag. Prior to the transition, the Fahrenheit to Celsius conversion and rounding to whole degrees Centigrade leads to considerable underrepresentation of -5, 0, 5, 10, etc. values. After the transition, values are rounded to the nearest 0.5 degree Centigrade. Station 24 is not included in this revision of the data because the transition to digital has not been completed.