Hello,
I am trying to use measurements of humidity as control variables in a model of pollution. Currently I am using the QLML surface specific humidity measure from Surface Flux Diagnostics. I understand that this represents the ratio of the mass of water vapour in the air relative to the mass of the air. However many other studies use relative humidity which as I understand is the percentage of water vapour in the air. I have followed instructions online using temperature and dew point to calculate the relative humidity like this:
(6.112 * exp((17.67 * Temp) / (Temp + 243.5)))/(6.112 * exp((17.67 * Dewpoint) / (Dewpoint + 243.5)))
I know that there are variables for RH in the 3-Dimensional data sets but I am trying to get Relative Humidity in a 2 dimensional data set.
Specific and Relative Humidity in MERRA 2
-
michaelwindsorcurtin
- Posts: 12
- Joined: Fri Dec 13, 2024 12:44 am America/New_York
Re: Specific and Relative Humidity in MERRA 2
Hi, I think you are asking how to compute RH from the meteorology that is available.
My suggestion would be to use the specific humidity to compute vapor pressure (e). Use the temperature to compute saturated vapor pressure (esat). The RH is e/esat*100
You are on to the formulations in your message. But, one very important consideration is those equations can have hard coded required units for temperature and pressure. So make sure you are using the appropriate units. I hope that helps!
My suggestion would be to use the specific humidity to compute vapor pressure (e). Use the temperature to compute saturated vapor pressure (esat). The RH is e/esat*100
You are on to the formulations in your message. But, one very important consideration is those equations can have hard coded required units for temperature and pressure. So make sure you are using the appropriate units. I hope that helps!
-
michaelwindsorcurtin
- Posts: 12
- Joined: Fri Dec 13, 2024 12:44 am America/New_York
Re: Specific and Relative Humidity in MERRA 2
Thank you,
I have already converted my temperature and dewpoint to Celsius from Kelvin if that is what you mean by "required units". I understand now after doing some more looking that the equation I provided was the same e/esat*100 equation you were talking about except I derived vapor pressure (e) from the dewpoint rather than from specific humidity and pressure (also I realise now that I mixed up the numerator and denominator when writing the code in my post but did not make that same mistake when coding).
Deriving from specific humidity (SH) and surface pressure (p) (converted to hPA) using (SH * p) / (0.622 + 0.378 * SH) gives me very close values to the value I calculated using dew point, which produces a very similar RH. The results aren't the exact same but that might have something to do with my data including areas with temperature below zero or the smoothing I'm doing over an exposure period. Regardless for usage as control variable in econometric analysis I feel that this would be sufficient as they are functionally identical within my model.
Thank you again for all your help.
I have already converted my temperature and dewpoint to Celsius from Kelvin if that is what you mean by "required units". I understand now after doing some more looking that the equation I provided was the same e/esat*100 equation you were talking about except I derived vapor pressure (e) from the dewpoint rather than from specific humidity and pressure (also I realise now that I mixed up the numerator and denominator when writing the code in my post but did not make that same mistake when coding).
Deriving from specific humidity (SH) and surface pressure (p) (converted to hPA) using (SH * p) / (0.622 + 0.378 * SH) gives me very close values to the value I calculated using dew point, which produces a very similar RH. The results aren't the exact same but that might have something to do with my data including areas with temperature below zero or the smoothing I'm doing over an exposure period. Regardless for usage as control variable in econometric analysis I feel that this would be sufficient as they are functionally identical within my model.
Thank you again for all your help.