Guest Commentary by George Tselioudis (NASA GISS)
In the past few years several attempts have been made to assess changes in the Earth’s planetary albedo, and claims of global dimming and more recently brightening have been debated in journal articles and blogs alike. In a recent article entitled “Can the Earth’s Albedo and Surface Temperatures Increase Together,” that appeared in EOS, Enric Palle and co-authors use recently released cloud data from the International Satellite Cloud Climatology Project (ISCCP) to explain how it is possible for the Earth to be warming even as it’s albedo is increasing. The need for an explanation arises from the author’s claim that the earth’s albedo has increased since the year 2000, an increase that was not followed by a decrease in surface temperature. They base this claim on Earthshine data (a measurement of the glow of the dark side of the moon that they use to deduce the earth’s reflectance) and on an albedo proxy derived from ISCCP parameters after they are regressed with two years of overlapping, but not global, earthshine observations. Subsequently they claim that the rising reflectance of the Earth has not led to a reversal of global warming because the difference between low and middle-plus-high ISCCP clouds has increased in the last four years. This they say implies that as the low-level, cooling clouds have decreased during the most recent years, the high-level, warming clouds have increased even more negating any potential cloud-induced cooling.
There are several issues connected to the use of earthshine data to calculate the earth’s albedo that have been discussed in peer-reviewed publications and that I will not discuss in this posting. I will say a few things, however, about the selective use of ISCCP data in this article to construct qualitative arguments that do not stand up to detailed quantitative analysis .
First, let’s take the claim that the Earth’s albedo has increased in the last four years. This is based primarily on the huge earthshine-derived albedo increase in 2003, which the authors now admit may be caused by undersampling of the data but was the the highlight of the authors’ recent Science paper (Palle et al, 2004). The other three years have values close to zero (relative to the reference year) with two years having error bars extending into the negative territory. The earthshine-trained ISCCP reconstruction of the albedo is a purely statistical parameter that has little physical meaning as it does not account for the non-linear relations between cloud and surface properties and planetary albedo and does not include aerosol related albedo changes such as associated with Mt. Pinatubo, or human emissions of sulfates for instance. Even this albedo reconstruction, however, shows only a weak positive trend in the last four years.
The ISCCP group produces an independent estimate of the albedo, from performing a full radiative flux calculation that takes into account observations of all radiative forcings and produces top of the atmosphere, surface, and in-atmosphere fluxes (data, figure right). This has been shown to be in excellent quantitative agreement with satellite measurements at the top-of-atmosphere and with surface measurements. The year-to-year variations of these values show some qualitative agreement with the earthshine-trained ISCCP reconstruction but very large quantitative differences.
The ISCCP estimate (right) shows a decreasing albedo trend of 1-2% in the 80s and 90s (as opposed to 7-8% in the earthshine-based proxy), a small increase of 1% form 1999 to 2001 and a flattening of the curve in the last three years. Quantitatively similar trends are derived from radiative flux retrievals by the ERBS and Terra and Aqua satellites.
Next consider the difference in trends between low-level and high-level clouds. It’s important that definitions be used carefully when we interpret satellite retrievals. First, the satellite can see actual low clouds only when higher cloud layers are not present. Second, the satellites retrieve the radiative, not the physical top of the clouds. As a result, a low cloud with a cirrus cloud overhead can be classified as a midlevel cloud in satellite observations. All these issues must be taken into account when calculating the radiative effect of clouds, as is done in the radiative calculations by the ISCCP group. More importantly, not all high-level and almost none of the middle-level clouds are radiative-warming agents. There is an optical depth threshold that depends on the cloud top height, above which the cloud becomes a cooling agent even with tops at high altitudes. Therefore the use of combined middle-plus-high clouds as a measure of the warming potential of the cloud field is a substantial overestimate of the effect. Moreover, a more careful look at the changes of ISCCP clouds by cloud type shows that the increase in total cloud cover from 2000 to 2004 is due to a small increases in high-level clouds and a larger increase in middle-level clouds that are mostly thermally neutral and therefore could not cause warming (see figures, data).
The increases in both high-level and middle level clouds (right) are caused by increases in the optically thicker cloud types, cirrostratus and cumulonimbus for the high-level and altostratus and nimbostratus for the mid-level clouds, that due to their large optical depths, cause radiative cooling. In fact, the same radiative calculations performed by the ISCCP group show that the outgoing longwave radiation increases during this time, opposite to the effect claimed. Therefore, the qualitative explanation given in the article is contrary to the quantitative analysis results derived from the ISCCP data.
The reconstruction of radiative fluxes from atmospheric properties is a very difficult and tedious job and both the ISCCP and ERBE/CERES groups are putting a great deal of effort into producing detailed and carefully evaluated radiative flux datasets. Both datasets show little or no albedo trend in the last four years. Thus explanations for how the albedo trends of the last four years are consistent with the surface warming and the ocean heat content increases are not necessarily required at this point in time.