Relative humidity has been measured within the Hubbard Brook Experimental Forest since 1956.
From 1956-1980, relative humidity was measured at seven hygrothermographs within the forest.
The hygrothermographs at Stations HQ(22), 1, 1A, 6, 14, 17, 23 and 24 recorded relative humidity by sensing the expansion and contraction of a set of horsehairs.
The accuracy of their calibration and frequency of maintenance has varied considerably over the years.
For a time in the early 1960s, they were not maintained at all.
Beginning around 1965 they were calibrated occasionally (less than once a year) by operating in a chamber with saturated salt solutions.
Since the 1970s, maintenance and salt calibration has been conducted annually, normally in the spring.
Minimum relative humidity at the time of maximum daily temperature was converted to vapor pressure.
Because vapor pressure varies little throughout a day and among locations, data from all available hygrothermographs were averaged in order to reduce error.
For 1971 - 1973, averaging gave triple weighting to HQ and double weighting to Station 1 to reflect the relative quality of their calibrations.
Most of the period of record prior to 1981 is not digitally available.
Data have only been read from the hygrothermograph charts for three periods: January 1966 to April 1966; June 1967 to Sept. 1967 and; May 1971 to April 1973.
However, the entire hygrothermograph chart collection has recently been scanned and prepared for data extraction to fill in gaps in the dataset.
Beginning in 1981, a humidity sensor has been operated as part of the automatic weather station at Headquarters.
This was originally a Brady Array.
Brady Array values were corrected by calibration with an Assmann psychrometer.
When the weather station blew over on August 1, 1983, this sensor was destroyed and data collection stopped.
On April 26, 1985 a Lambda temperature-humidity probe was installed.
The humidity sensor was replaced approximately annually from 1985-2013; however, missing data still occur for a variety of reasons.
The manufacturer's calibration was used and was checked occasionally with an Assmann psychrometer.
On July 22, 2013 a Campbell Scientific HMP60 air temperature and relative humidity probe was installed at Headquarters in a 6-plate Gill shield.
The HMP60 sensor was replaced with Campbell Scientific CS215 air temperature and relative humidity sensor on 2019-06-15 and the Gill shield was cleaned at that time.
From 1981-09-07 through 1989-03-01 gaps were filled using the mean RH for that day over the period 1981-2010 or a random estimate done by Mark Green (Flag = d or e, respectively).
From 1989-03-01 through 2006-12-31 missing values were filled by substituting the value recorded at the nearby CastNet site (Flag = c).
From 2007-01-01 through 2013-07-22 missing values were gap filled according to linear regression with the nearby Castnet site (Flag = r).
The regression equation was RH_HQ = 5.2 + 0.95*RH_CastNet.
Beginning 2013-07-23, missing values were gap-filled according to linear regression with the RH sensor at weather station 1 (Flag = 1).
The regression equation is RH_HQ = 0.27 + 0.97*RH_STA1.
In the rare event of missing data at both Headquarters and station 1, data were gap-filled by linear regression with station 23 (Flag = 23).
The regression equation is RH_HQ = -3.5 + 0.97*RH_STA23.
We use observed mean daily air temperature (T_air) and mean daily relative humidity (RH) to calculate vapor pressure and related derived quantities.
First, we calculate saturated vapor pressure (e_sat) in kilopascals (kPa) according to the equation:
\begin{equation}
\label{satVP}
e_{sat} = 0.61078 * e^{(17.269 * T_{air})/(T_{air} + 237.3)}.
\end{equation}
Then, we calculate vapor pressure (VP) and vapor pressure deficit (VPD), both in kPa, as:
\begin{equation}
\label{VP}
VP = e_{sat} * RH * 0.01
\end{equation}
\begin{equation}
\label{VPD}
VPD = e_{sat} - VP.
\end{equation}
