REPSONDE

Scientific hypothesis: We will study the variability in the vertical of cirrus microphysical properties, and correlate these properties with radar reflectivity from the CART radar and with radiosonde thermodynamic data. The two primary purposes are to develop a conceptual model and parameterizations of the evolution and vertical structure of cirrus, and to assess the two independent cirrus radar retrieval algorithms of Mace and Matrosov.

Approach to test hypothesis: We will launch balloon-borne Formvar replicators over the CART radar. Replicators sample ice crystals and preserve their detailed characteristics. We may also be able to coordinate replicator launches with overflights of Lawson's Lear jet with CPI microphysical instrumentation. The characteristics and capabilities of replicator data can be seen on our website (best cases are Nov. 25 and Dec. 5 from FIRE).

Activity Summary

This report summarizes the results of our program to acquire vertical profiles of cirrus microphysical data from our balloon-borne replicator instruments at the SGP CART site. The primary purpose of the replicator measurements is to provide in-situ microphysical data to Jay Mace (University of Utah) that is suitable for assessment of the CART radar (MMCR) and Jay’s cirrus retrievals from MMCR data. Support was received from DOE/ARM for the following activities: deploying the replicators at the CART site for a 3-week period to acquire data for all cirrus cases that occur; to provide a summary of qualitative ice crystal characteristics correlated with simultaneous radiosonde data for each cirrus case; and to share the cost of making instrument improvements. An "option" was described in the proposal for possible future funding of thorough quantitative analysis of selected replicator profiles if any high-quality cases were obtained.

The replicators were modified to use a narrower film for the cloud particle collection surface (8 mm vs 35 mm), which improves airflow through the instrument and improves the collection efficiency for small particles. Two test flights of the modified replicators conducted in Boulder showed that the modifications were successful. Larry Miloshevich deployed the replicators to the CART site during the time period 26 April to 20 May 1999. During the campaign there was a grand total of two cirrus events, both of which were sampled by replicators. The 10 May case is classified as "marginal" and the 1 May case is classified as "poor," in terms of suitability for possible quantitative comparison with radar retrievals. The two cirrus cases are detailed below, followed by assessment of the instrument performance and comments on operational issues. The figures included in this report (replicator data summary with supporting data from the MMCR, radiosondes, satellite imagery, and photography) are posted on our website. The data have been shown to and discussed with Jay Mace, who concurs that neither case is suitable to justify pursuing further quantitative analysis. However, the qualitative data, particularly the 10 May case, are quite interesting from a cloud physics perspective, and could conceivably provide limited information for assessing MMCR characteristics.

10 May 1999 Case: The replicator launch on 10 May at 1639 UTC sampled the westernmost edge of a north-moving large-scale convective system, as shown by the visible and infrared satellite imagery in Figs. 1 and 2. A prelaunch photograph looking east from the launch site (Fig. 3) shows that the sampled cirrus is the thin band of cloud over north-central Oklahoma seen in Fig. 1.

Altitude profiles of the radiosonde data for this flight are shown in Fig. 4, where the balloon ascent is shown with a solid line and the parachute descent is shown with a dashed line. The elevated relative humidity in Panel 1 is approximately equal to the dashed ice-saturation line between 9.0 and 12.5 km, indicating the portion of the cloud where ice crystals are not sublimating. The MMCR reflectivity data (Fig. 5) show that the cloud thickness and reflectivity varied considerably over time. Although the replicator penetrated cloudbase at 1701 UTC, it is critical to note that the replicator data do not correspond to the radar data at that same time for two reasons. First, a cloud parcel sampled by the replicator has been advected downwind from the vicinity of the MMCR by 16 km at cloudbase and 35 km at cloudtop, as judged from the Panel in Fig. 4 labeled "Range." The difference between the cloudtop and cloudbase advection is due in part to transit time of the replicator through the cloud layer and in part to considerable vertical wind shear within the cloud layer, as shown by the altitude profile of wind speed.

One could compensate for effects of vertical wind shear in order to correlate replicator and radar samples if not for the second factor: directional wind shear. The vertical profile of wind direction in Fig. 4 shows that the lower-level and upper-level wind directions differ substantially, such that the replicator was translated during its ascent to a position not along the line of the upper-level winds, and therefore the cloud sampled by the replicator was not measured by the MMCR. This effect is clearly seen in a "map view" of the flight in Fig. 6, where each dot represents one km of altitude change. This mismatch in sample locations of the replicator and radar might not be a critical factor if the cloud were horizontally homogeneous; however, the radar, satellite, and photography data show that the cloud characteristics vary considerably in the horizontal.

The replicator film was viewed under a microscope, and digital images were taken of representative ice crystals. The position of these crystals on the replicator film was correlated with the radiosonde data to create an altitude profile of the qualitative ice crystal characteristics (Fig. 7). Two particularly interesting microphysical observations bear mentioning. First, the variation in the vertical of the ice crystal characteristics is consistent with our "three-layer model" of cirrus development. This model contends that, for cirrus formed in-situ, crystals nucleate in ice-supersaturated air and are small in size and pristine in shape (the cloudtop "nucleation layer"), then crystals grow larger as they fall through ice-supersaturated air (the midcloud "growth layer"), and finally crystals become rounder and successively smaller with decreasing altitude as they sublimate in ice-subsaturated air (the cloudbase "sublimation layer").

The 10 May case is consistent with this description, where crystals at cloudtop are pristine and small (30-50 microns), crystals are pristine and larger in the midcloud, and crystals in the lower cloud (below ice-saturation as seen from the relative humidity profile) are rounded by sublimation and become small at the very base of the cloud. Although this case is not the so-called in-situ cirrus for which the three-layer model is intended, but rather is cirrus generated by large-scale convection, the three-layer description nonetheless appears valid. We have previously speculated that this might be true but via a different mechanism, namely gravitational size sorting of crystals as they are advected with the system.

This case is also unusual in comparison to our other replicator profiles in that the crystals are dominantly hexagonal plates, where we generally see only a small fraction of plates. This is understandable because this is convectively-generated cirrus, where abundant moisture leads to rapid and two-dimensional crystal growth. These crystals are perhaps more characteristic of anvil or tropical cirrus, which we have not previously sampled with the replicator.

1 May 1999 Case: The general synoptic conditions for the replicator launch on 1 May at 0158 UTC were (unfortunately) similar to the 10 May case, where the replicator sampled the easternmost edge of a north-moving large-scale convective system (Figs. 8 and 9). Prelaunch photographs looking south from the launch site (Fig. 10) also show that the sampled cirrus is very patchy and inhomogeneous near the CART site.

Altitude profiles of the radiosonde data are shown in Fig. 11, and a map view of the flight is shown in Fig. 12. In this case it is not possible to even remotely discern cloud boundaries from the relative humidity profile, as a result of the extremely slow time-response of the radiosonde humidity sensor at very cold temperatures. [For future reference, we will soon begin developing a correction algorithm for this "time-lag error" in ARM radiosonde humidity measurements as part of a separate research program.] The MMCR data (Fig. 13) show that the cloud reflectivity increased continuously during the time period encompassing the replicator flight, and thus was not horizontally homogeneous. Furthermore, this case also displayed considerable vertical wind shear and directional shear, resulting in a poor opportunity for correlation of the replicator and radar data, as discussed in the previous section.

A profile of representative replicator data correlated with the radiosonde altitude data is shown in Fig. 14. In contrast to the single cloud layer seen in the MMCR data, the replicator data show two thin layers at about 9 and 13 km altitude, with occasional crystals between these altitudes. The crystals are primarily 150-250 micron columns and bullets, with occasional 100 micron plates. It is quite possible that the large difference in cloud structure seen by the replicator and the MMCR results from the mismatch in sample locations and the high degree of cloud inhomogeneity apparent in the satellite and photographic data. However, it is also possible that the replicator replicated only a fraction of the sampled ice crystals, as discussed below.

Instrument Performance and Comments On Operations: The replicators were modified and tested in the laboratory and in-flight prior to this deployment. The width of the crystal collection surface was reduced from 35 mm to 8 mm, improving airflow through the instrument and increasing the collection efficiency for small crystals. Other modifications included an airflow guide in the sample opening to more precisely define the instrument’s sample volume; a redesigned switch to improve reliability of the connection between the replicator and radiosonde; a new LORAN antenna configuration to improve reliability of the radiosonde position data during the parachute descent; and a new "dereeler" containing 40 m of string to lower the replicator below the wake of the balloon after launch.

All modifications performed well during a pre-deployment local flight test, as judged by the cirrus data acquired (Fig. 15). However, there are two reasons to believe that the replication process did not always work properly, leading to replication of only a fraction of crystals within the sample volume. The paucity of crystals between the two cloud layers in the 1 May case may be due to this incomplete replication, but may also reflect the actual structure of this inhomogeneous cloud. However, a back-of-the-envelope estimate of crystal concentrations for the 10 May case of a few per liter is considerably lower than the rule-of-thumb estimate provided by Jay Mace that 10 per liter of 100 micron columns has a reflectivity of -38 dBZ (which is a considerably lower reflectivity than the bulk of the MMCR profile). The second reason to suspect incomplete replication concerns the volume of Formvar-softening solvent dispensed during the flight. It is straightforward to calculate how much solvent should be dispensed during the flight, and it was found that less than half the appropriate amount was dispensed, though it is not known whether less solvent was dispensed continuously throughout the flight or whether the dispensing orifice became completely clogged at some time during the flight. The conclusion is that a high probability exists that incomplete replication occurred, and therefore the data are not suitable for further quantitative analysis. Sufficient replication did occur, however, to provide the qualitative assessment of ice crystal characteristics.

We believe we fully understand the cause of the incomplete replication, and the problem will be fixed and tested before the replicators are deployed again. The pre-deployment test flight (Fig. 15) was successful in terms of both data quality and in terms of dispensing the appropriate amount of solvent. The only difference between the pre-deployment test flight and the CART flights was the specific replicator films used. The films are precoated with Formvar in the laboratory prior to deployment using a device that rolls a thin layer of softened Formvar onto the film. A new coating device was constructed for the new 8 mm films, but the test flight used film coated on the old 35 mm coating device. The new coating device used to coat the replicator films for the CART flights applies a thicker layer of Formvar, and it is believed that softening the thicker layer of Formvar clogged the orifice of the solvent dispenser, which is in direct contact with the coated film. A thinner layer of Formvar is apparently insufficient to clog the orifice. We intend to fix the problem by reducing the thickness of the Formvar layer and orienting the dispensing orifice so that it does not contact the coated film.

Finally, two operational issues bear mentioning. Procedures for communicating impending replicator launches with Vance AFB were established by Ted Cress prior to the Replicator Campaign. The procedures worked well, but a few modifications were incorporated in the field based on discussions between Larry Miloshevich, Vance personnel, and Jim Teske. A detailed list of the modified procedures was written and disseminated, and that document is enclosed in this report. Second, a visit to the CART site by a DOE Safety Officer included a replicator safety assessment. The assessment resulted in my writing a summary document of safety issues (also enclosed). The safety assessment also concluded that the "dereeler" device was a weak link in the safety of the operations because it was composed of metal. We have already replaced this device with a version composed of plastic.

Timeline

Campaign Data Sets