# Specific humidity range in tropical lowlands?

I am running a land biosphere model that uses meteorological data as drivers. The model has a threshold of 32g/kg for specific humidity. The data I am using taken from a reanalysis performed by J. Sheffield at Princeton has some values of S. Hum. higher than 32g/kg. Given that such data is of resolution of 1 degree and 3 hourly interpolated from 6-hour data, what is a typical range for S. Hum. in lowland tropical areas. I need to know this to deem if these data is close enough to the real world or if it includes unrealistic values I will consult the author and point out that data might be inaccurate for these regions. It is difficult to find online any direct measurements, I only find these big data sets such as NCAR-NCEP.

• At 1000 mb, 32g/kg gives a dewpoint of 33C which has been observed but is rare. – casey Jun 24 '15 at 15:41

Referring to a research article "The Role of Temperature and Humidity on Seasonal Influenza in Tropical Areas: Guatemala, El Salvador and Panama, 2008–2013" by Soebiyanto et al, 2014 DOI: 10.1371/journal.pone.0100659, the authors investigate a few regions in tropical wet and dry (savanna) climates in Central America, some of which are lowland. The authors state that Panama and El Salvador study locations have less varied topography and are at lower elevations than Guatemala. They give the following Specific Humidity values and standard deviation i g/kg:

Lowland Panama province in Panama: 17.66 +/- 1.04 g/kg
Chiriqui province in Panama: 15.38 +/- 1.47 g/kg
West-central departments in El Salvador: 14.29 +/- 2.14 g/kg
San Miguel, El Salvador: 14.84 +/- 2.32 g/kg


So, a specific humidity in the 10-20 g/kg seems like a reasonable range for seasonally dry and humid lowland tropical regions in Central America.

• thanks I have also looked at that paper but missed the table. Given that the SD should represent the range of values, I see the highest values would be in the 19 g/kg ballpark. But look at the data I posted in the question. – Herman Toothrot Jun 24 '15 at 18:12
• @user4050 the range (1st and 99th quantiles) should be approx. mean+/- 3 SD, but note that estimating the min/max in this way has high uncertainty, especially if the distribution is not normal, which it likely isn't . – David LeBauer Jul 3 '15 at 1:02