Head in a Cloud

troposphere and stratosphere meet blogosphere

Head in a Cloud random header image

Cubic Ice and Persistent Ice Supersaturation

September 25th, 2006 by Sean Davis · 4 Comments

A couple of weeks ago, a new GRL paper on cubic ice came out entitled “Measurements of the vapor pressure of cubic ice and their implications for atmospheric ice clouds” (Shilling et al.). The paper’s authors includes several people associated with our department here at CU including Maggie Tolbert, Brian Toon, and Eric Jensen.

In this paper, they report direct measurements of the saturation vapor pressure of cubic ice (Ic), which may form under very cold atmospheric temperatures (<~200K). Cubic ice is relatively uncommon on Earth (with the more predominant type being hexagonal ice), and is thought to occur mainly in cold ice clouds, such as tropical cirrus (the focus of this study), noctilucent clouds, and polar stratospheric clouds.

This paper presents laboratory measurements of the saturation vapor pressure over cubic and hexagonal ice (formed using two different mechanisms) at temperatures between 180 – 190 K. I won’t go in to the detail on how this is done, but basically they grow thin films of each type of ice on substrates, and then meausure the vapor pressure after the system has achieved equilibrium. They also verify that the ice type they made was what they expected using x-ray diffraction measurements. The main result of the laboratory section of this paper is that the vapor pressure of cubic ice is about 10% greater than of hexagonal ice.

So who cares? Well, this difference has implications for the measurements made by the JPL laser hygrometer (JLH) and Harvard water vapor (HWV) instruments during the CRYSTAL-FACE campaign, which indicated persistent ice supersaturation (Si~1.2 and greater) in cold cirrus clouds.

In 2004, Gao et al. published a paper in Science in which they attributed the increase in relative humidity to a new HNO3-containing phase of ice.

Here, Shilling et al. show present a re-analysis of the JLH and HWV water vapor measurements from CRYSTAL-FACE to show that some (but not all) of the supersaturation (with respect to hexagonal ice) from CRYSTAL-FACE can be explained by a higher vapor pressure over cubic ice (Figure 3 makes this point). My take on the importance of this result is that it supports the idea that the supersaturations observed during CRYSTAL-FACE can not be explained away simply by the fact that cubic ice is present, which I imagine could have been one of the criticisms of the Gao et al. nitric acid hypothesis.
One question I have about this paper is related to the treatment/scaling of the JLH and Harvard WV data. Instead of using the Harvard WV data (as Gao et al. did), they used the JLH data, as it is not subject to sublimation of ice crystals in it’s inlet (there is no inlet for the JLH — it’s an open path measurement), which is apparently an issue for the HWV instrument. However, they go a step further. They scale the JLH WV to the HWV value in clear-air outside of the clouds, justifying this by stating that the JLH is not as accurately calibrated as the HWV. I’m not sure how justified this scaling is, but regardless, would like to have known how using the JLH alone would have affected the interpretation of their Figure 3. My experience with the JLH data is that it has been consistently lower than the HWV, at least for the MidCiX data with which I am familiar. Would using just the JLH data to create Figure 3 mean that there is simply no supersaturation in these clouds? Given that the JLH data is scaled to the HWV value in clear-air, I assume so. So this brings up the potentially thorny issue of the WV measurements themselves … which has no doubt has been discussed ad naseum elsewhere (here, for example). That’s a topic I’d rather leave for others to deal with, but this paper serves as a good example of how sensitive the scientific interpretation can be to subtle differences in measurement accuracy/contamination for instruments measuring such a seemingly innocuous quantity as water vapor.

Tags: climate · instruments · modeling

4 responses so far ↓

  • 1 Eric Jensen // Sep 26, 2006 at 9:38 am

    We did indeed scale the JLH data using the HWV/JLH ratios in clear-sky regions around the clouds. The assumption here is that outside clouds, the Harvard instrument does not have a problem with ice sublimation in the instrument and should provide an accurate measure of the water vapor concentration. Scaling the JLH data to HWV in clear air would not remove the supersaturation in clouds. We felt the scaling was justified since the Harvard group does an extremely careful calibration of their instrument, and there have been known problems with the JLH calibration. The scaling amounts to a 5-10% increase in the JLH values. Hence, if we had used the JLH water vapor concentrations directly, the result would still have been significant supersaturations within the cold cirrus and contrails.

    As a side note, the discrepancies between water vapor measurements are a severe problem, but in the big picture view, JLH and HWV agree quite well. The big discrepancy is between all of the aircraft instruments (at least 6 of them, using different techniques) on the one hand, and the frostpoint balloon and satellite measurements on the other hand. The difference can be as much as 50% under very dry conditions. If the frostpoint balloon measurements are correct, then the supersaturation in cold cirrus issue disappears.

  • 2 seand // Sep 26, 2006 at 10:01 am

    Hi Eric,

    Thanks for your comment, and excuse me for assuming that using the JLH data alone would have meant no supersaturation. As a point of clarification, for any given cloud encounter, the JLH data were scaled to the HWV value outside the cloud, no? This wasn’t just a matter of adding 10% to all the JLH numbers (?)

    Also, another question came to me about the implications of in-cloud super-supersaturation(S > 150%)/subsaturation (S < 70%). In the paper, there is mention that the data rate of the instruments (1 Hz) could be a source of the anomalously low (or high) values if the cloud is extremely heterogeneous.
    I can see how this would lead to anomalously low saturation values (i.e. if the aircraft encounters a non-cloud, undersaturated parcel of air as well as a (super-)saturated parcel of air within 1 second), but am having difficulty envisioning how it would work the other way around. I suppose large, fast (faster than 1 Hz) temperature fluctuations could lead to this, but I’m not sure. …IF it is temperature fluctuations, then perhaps one way of looking at the issue would be with the MMS temperature. I’m not sure how fast the MMS temperature was during CRYSTAL-FACE, but for MidCiX it was at 20 Hz. If small-scale variability is playing a role in these anomalous values, I imagine they would show up in the temperature data… (?)
    On a related note, although this was not possible during C-F or MidCiX, the CLH measurement is now faster than 1 Hz, and could help shed some light on sub-second cloud variability in future campaigns such as TC4.

  • 3 Felix Franks // Feb 17, 2007 at 11:55 am

    It is sad that there are still scientists who persist to write and speak about “cold temperatures”. Don’t they know that temperature is measured in units. Temperatures can be high or low, but not cold or hot !! Talk about butchering the English language.

  • 4 seand // Feb 21, 2007 at 3:40 pm


    I dont understand your problem with the wording here. It is specifically stated what the term “cold” means in this context. Of course cold and hot are relative terms. …Cold to someone in the physics dept. might mean fractions of a Kelvin. As long as it is understood what cold and hot mean in a given context, there is nothing wrong with that usage.

Leave a Comment