Emissivity is a measure of how effective a material is at emitting thermal radiation. It’s a key parameter in determining what the temperature of an object will be when it is in thermal equilibrium. So, it is of interest to look at the emissivity of Earth’s surface.
The emissivity of natural materials varies over a fairly relatively limited range. So, the answer likely won’t have a large impact on our discussions about climate. Still, it’s good to look at what is known.
Some of the nuances are a bit difficult to pin down. In particular, more work may be needed to understand how to reconcile differences in emissivity estimates from different sources. So, I expect the material on this page will evolve over time.
Different sources offer surprisingly different values for the baseline emissivities of ocean and land. This probably has little practical importance, so long as temporal changes in emissivities are properly tracked. However, it’s a puzzle I’d like to make more sense of.
Table of Contents
A 2014 summary of ASTER satellite data reported:
The emissivity of most natural Earth surfaces is a unitless quantity and ranges between approximately 0.6 and 1.0, but surfaces with emissivities less than 0.85 are typically restricted to deserts and semi-arid areas. Vegetation, water and ice have high emissivities, above 0.95 in the thermal infrared wavelength range.
And, below is the associated map of land surface emissivity.
The International Satellite Cloud Climatology Project (ISCCP) was a satellite data collection project which, among other things, collected data on Earth’s surface emissivity in 1992-1993.1I first learned about this dataset from a blog post by Zoe Phin. While subsequent satellites have gathered higher-resolution time-resolved emissivity data, the ISCCP dataset offers a convenient low-resolution summary of both land and sea surface emissivities, over a broad range of wavelengths from 5-200 microns.
Below is an image I’ve generated of emissivity as it is distributed over the surface of the globe, according to the ISCCP dataset.
In the table below, I’ve collated some statistics about these emissivity values.
|0.935||0.921||1.000||Warm latitudes (between 30ºN and S)|
|0.939||0.919||1.000||Cool latitudes (above 30ºN or S)|
|0.947||0.920||1.000||Mixed ocean and land|
|0.920||0.920||0.921||Ocean warm latitudes|
|0.923||0.919||0.953||Ocean cool latitudes|
|0.974||0.947||1.000||Land warm latitudes|
|0.970||0.936||1.000||Land cool latitudes|
It’s important to look at differences between warm latitudes and cool latitudes because, in considering the emissivity of the entire Earth’s surface, one ought compute a weighted average emissivity, weighted by emissions, i.e., weighted by T4 where T is temperature. So, it matters more what the emissivity is in places where it is warm. However, the average emissivity doesn’t seem to vary too strongly with latitude. For many purposes, it’s likely sufficient to simply say the overall emissivity (according to this dataset) is around 0.935.
The Science of Doom blog in 2010 did a deep dive investigating the Emissivity of the Ocean, summing up the results as follows:
The best up to date measurements of ocean emissivity in the 8-14 μm range are 0.98 – 0.99. The 8-14 μm range is well-known because of the intense focus on sea surface temperature measurements from satellite.
From quite ancient data, the average emissivity of water across a very wide broadband range (1-100 μm) is 0.96 for water temperatures from 0-30°C.
The values from the ocean when measured close to the vertical are independent of wind speed and sea surface roughness. As the angle of measurement moves from the vertical around to the horizon the measured emissivity drops and the wind speed affects the measurement significantly.
The ISSCP dataset reports an average sea surface emissivity of 0.921 in the 5-200 μm band, with a range (in the annual data) from 0.919 to 0.953.
How can we explain the discrepancy between the 0.96 value for 1-100 μm and the 0.92 value for 5-200 μm?
A bit less than 0.5 percent of thermal emissions (at 288 K) have wavelengths between in 100-300 μm; and a bit under 1 percent of emissions have wavelengths in the range 1-5 μm. If emissivity was 1 in the 1-5 μm band and 0 in the 100-300 μm band, that would only lead to an emissivity difference of around 0.006 between the values in the 1-100 μm band and the 5-200 μm band. So, it doesn’t seem like the difference in wavelength ranges could explain the discrepancy between 0.95 and 0.92.
At the moment, I don’t know what to make of the difference. Let me know if you understand what’s going on with this. I’ll need to ponder the situation a bit more.
How does emissivity vary with wavelength? Schmugge et al (2002) offered the chart below.