Guest Post by Willis Eschenbach
Since the late Nineties the US has had seven industrial-strength stations that measure a variety of climate variables every minute, 24/7. These are called “SURFRAD” stations. As a data junkie I’ve been wanting to look at their results for a while … but the data is in an ugly format. They have a single data file for each station for each day of the last 17 years … not my idea of a party.
Anyhow, I finally bit the bullet and downloaded a year’s worth of data, about a quarter of a gigabyte. For no particular reason I picked the SURFRAD station in Goodwin Creek, Mississippi, and the year of 2010. For each minute they have no less than 21 different measurements (see end notes) … so I sorta started digging around in the data to see what stuck out. Here was the first oddity I came across:
Figure 1. Average 10-metre surface air temperature (black, °C) and average downwelling infrared radiation (blue, W/m2) for the year 2010. Measured at the SURFRAD station in Goodwin Creek, Mississippi. Average covers the entire year, and is shown repeated twice (two days) for clarity.
I don’t know why, but I wasn’t exactly expecting that … which is the best part of science. I love surprises, the unexpected, and climate science is chock full of those. I mean, I knew that downwelling radiation was a function of air temperature … I just didn’t expect the alignment with the underlying surface temperature to be so exact. Other than the atmosphere starting to cool a bit earlier in the day than the ground (as we’d expect from the relative masses) they match up perfectly.
Now, seeing how good that match was, I got to wondering how well that fits the theoretical profile that we’d expect from the Stefan-Boltzmann (S-B) relationship. This relationship says that infrared radiation is equal to emissivity times the Boltzmann constant times the temperature to the fourth power. I figured that using that formula, I could calculate an approximate value for the emissivity from the data with a simple linear analysis.
Now, here’s the curious part. When I did that, I got an emissivity of 0.590 … which from everything I’ve read is too low.
So I thought, well, that kinda makes sense, because the temperature up where the radiation is coming from is cooler. But how much cooler? That depends on what altitude the radiation is coming from. Now my bible in these matters is “The Climate Near The Ground”, by Rudolph Geiger, which anyone interested in climate science should read. Geiger gives the following table for downwelling radiation (called “counterradiation” in those days):
Table 5-1 Contribution of various atmospheric layers to counterradiation received at the surface Layer thickness (m) % share of counterradiation 87 72.0 89 6.4 93 4.0 99 3.7 102 2.3 108 1.2
I figured that I could use that to give me at least a first cut at the temperature of the overlying atmosphere at altitude, using the lapse rate of one degree C per each hundred metres of altitude. For the six layers given by Geiger, this gives mid-layer temperature drops of 0.4°, 1.3°, 2.2°, 3.2°, 4.2°, and 5.2° degrees C. A weighted mean of these (allowing for the fourth power relationship) gives an average temperature drop of 0.85°C. This makes sense, because about three-quarters of the downwelling radiation comes from the bottom hundred metres of atmosphere, which is not much cooler than the surface.
However, this doesn’t solve the conundrum. Remember that I got an emissivity of 0.590 using the surface temperature. IF in fact on average the radiation is coming from a temperature which is 0.85°C cooler, then using that temperature it only brings the emissivity up to 0.595 … hmmm.
So that’s my puzzle for today. Is Geiger wrong about the source of the downwelling radiation? Is the emissivity of the atmosphere really on the order of 0.6? Is something else going on?
Inquiring minds wonder …
My best to everyone,
w.
AS USUAL: if you disagree with someone, please quote the exact words you disagree with. This lets all of us understand the exact nature of your objections.
CODE AND DATA: The R code, the functions, and the hundreds of daily files for 2010 are in a zipped folder called “SURFRAD Analysis”. WARNING: 21 megabyte file.
{UPDATE] Prompted by a typically detailed and interesting comment below from Dr. Robert Brown (rgbatduke), here is a scatterplot of the complete temperature and downwelling IR datasets:
[UPDATE 2] The same graph, but for Boulder, Colorado.
SURFRAD VARIABLES:
# station_name character station name, e. g., Goodwin Creek
# latitude real latitude in decimal degrees (e. g., 40.80)
# longitude real longitude in decimal degrees (e. g., 105.12)
# elevation integer elevation above sea level in meters
# year integer year, i.e., 1995
# jday integer Julian day (1 through 365 [or 366])
# month integer number of the month (1-12)
# day integer day of the month(1-31)
# hour integer hour of the day (0-23)
# min integer minute of the hour (0-59)
# dt real decimal time (hour.decimalminutes, e.g., 23.5 = 2330)
# zen real solar zenith angle (degrees)
# dw_solar real downwelling global solar (Watts m^-2)
# uw_solar real upwelling global solar (Watts m^-2)
# direct_n real direct-normal solar (Watts m^-2)
# diffuse real downwelling diffuse solar (Watts m^-2)
# dw_ir real downwelling thermal infrared (Watts m^-2)
# dw_casetemp real downwelling IR case temp. (K)
# dw_dometemp real downwelling IR dome temp. (K)
# uw_ir real upwelling thermal infrared (Watts m^-2)
# uw_casetemp real upwelling IR case temp. (K)
# uw_dometemp real upwelling IR dome temp. (K)
# uvb real global UVB (milliWatts m^-2)
# par real photosynthetically active radiation (Watts m^-2)
# netsolar real net solar (dw_solar – uw_solar) (Watts m^-2)
# netir real net infrared (dw_ir – uw_ir) (Watts m^-2)
# totalnet real net radiation (netsolar+netir) (Watts m^-2)
# temp real 10-meter air temperature (?C)
# rh real relative humidity (%)
# windspd real wind speed (ms^-1)
# winddir real wind direction (degrees, clockwise from north)
# pressure real station pressure (mb)
