In-flight Icing

RAP Achievements

A. In-flight Icing

[Background] [Dual-frequency radar]
[10-11 March radar]
[WRF microphysics]
[Giant nuclei] [New microwave satellite-based technique]
[Satellite/aircraft] [SLD/non-SLD environments]
[Benchmarking in-flight icing detection products]

1. Background

In-flight icing research has been conducted at RAP since 1989 and continues to provide interesting and challenging studies in cloud physics, remote sensing, and mesoscale meteorology. The goal of this research is to develop more accurate and timely diagnoses and forecasts of conditions leading to ice accretion on aircraft during flight. Two of our algorithms, the Current and Forecast Icing Potentials (CIP and FIP), are now fully operational products at the National Weather Service. Similar products covering Alaska and an icing severity depiction are accepted as experimental at the Aviation Weather Center at Kansas City.

RAP's icing forecasting research also includes improvements to the MM5 and WRF models, including microphysical parameterizations to more accurately forecast clouds, drizzle and rain. These improvements are transferred to operational use via the NOAA Forecasting Systems Laboratory and the National Centers for Environmental Prediction. RAP developed good working relationships with these organizations, which provide an effective means of technology transfer. Additionally, RAP scientists are leaders in model improvements working together with other divisions of NCAR and with the university community.

Remote sensing systems are being designed to utilize data from radars, radiometers, and satellites. A major milestone this year was the development and initial testing of the SPolKa system, which combines a Ka-band (0.86 cm) radar with EOL's SPol S-band (10 cm) weather radar. This development was a joint project between the FAA and NSF; the FAA's interest centers on icing detection while the NSF's primary interest is in the microphysics of clouds. The system was tested during WISP04, which took place in the northern Colorado Front Range area from February through early April 2004. SPolKa was deployed at the Marshall instrument test site, the NOAA Ka-band GRIDS (Ground-based Remote Icing Detection System) was at Erie, and the University of North Dakota Citation research aircraft provided in situ samples for evaluation of the SpolKa. We have also performed detailed simulations of the response of multiple frequency radar systems to realistic cloud conditions to evaluate the effects of mismatched beams, and other effects on liquid water retrievals. Additionally, we have been working on incorporating our studies on remote icing detection into a data ingest and display system being developed at NASA Glenn Research Center's Icing Branch.

RAP also continues to work on the characterization of cloud and weather conditions associated with icing environments. In 2004, we focused on environments associated with supercooled large drop (SLD, drops with diameters > 50 microns) regions, and how these differ from environments supporting smaller, cloud-sized drops. This work is in collaboration with the Meteorological Service Center of Canada and incorporates measurements from their instrumented Convair, along with NSF C-130 and NASA Twin Otter data, obtained during the AIRS-2 field effort near Montreal in late 2003.

2. Dual-frequency radar analyses

The use of dual-frequency radar measurements to remotely detect cloud liquid water content (LWC) offers the prospect of providing valuable information for both cloud microphysics research and icing applications, including the identification of supercooled liquid water hazardous to aviation. However, previous efforts to apply this technique in practice have frequently been unsatisfactory due to contamination of the dual-wavelength ratio (DWR, the difference in reflectivity measured at two frequencies) by measurement noise, Mie scattering, and mismatch of radar pulse volume sizes and locations. With funding from the FAA’s Aviation Weather Research Program (AWRP) John Williams and J. Vivekanandan have been working to understand the sources of measurement error and to develop improved data processing and retrieval techniques that permit the extraction of reliable, high-resolution cloud LWC from dual-frequency radar data.

An improved understanding of sources of error has been achieved through the analysis of both field program data and simulated data. A close analysis of data from the University of Massachusetts' CPRS Ka- and W-band radars, showed that incorrect range-gate locations and different range spacings and pulse widths between the two frequencies were responsible for significant distortion of the measured dual-wavelength ratio (DWR), this error propagates directly into errors in the retrieved LWC. Figure A-1 illustrates the error when 75 m W-band data are linearly interpolated onto the 30 m Ka-band range gates. The variations in DWR (black dots) are largely due to the fact that the Ka- and W-band ranges are offset by about 40 m, the relative “smoothness” of the W-band profile due to a longer pulse width, and the linear interpolation of the W-band data onto the 30 m-spaced ranges also contribute to error in retrieved LWC. These observations led to the development of an empirical technique for correcting range gate and temporal offsets, and underscored the importance of designing dual-wavelength systems to have matched measurement volumes.

Figure A-1. Profiles of University of Massachusetts CPRS radar reflectivities and DWR from data collected at 16:31:42 UTC on 15 April 1999. The curves are the DWR, the Ka-band reflectivity, and the W-band reflectivity (shifted up by 10 dBZ); for the W-band reflectivity, both the original 75 m measurements (red diamonds) and the 30-m interpolated (green plus) values are shown.

A second achievement was the design of a new method for retrieving LWC at high resolution—in fact, the same resolution as the reflectivity fields. This new method should provide a substantial improvement over existing methods, which generally handle measurement error by performing smoothing which, in turn, lowers the resolution of the retrieval. The new method makes use of a hybrid regularization technique to fit the ratio of LWC to the square root of linear reflectivity by minimizing the confidence-weighted squared error between the measured DWR and the DWR resulting from the fit. Evaluation of the new method using simulated data, the MWISP data described above, and field data from the 2004 Winter Icing and Storms Project is ongoing.

3.10-11 March radar & radiometer LWC and RES retrieval

Remote measurements of cloud liquid water content (LWC) and characteristic droplet size (e.g., radar estimated size: RES) are required for quantifying potential aircraft icing hazard. During the 2004 Winter Icing and Storms Project (WISP04), research radars and radiometers were deployed at NCAR's Marshall experimental site near Boulder, Colorado to evaluate remote sensing techniques for characterizing cloud icing conditions. The dataset included radar and radiometer measurements.

On 10-11 March, a shallow, fairly uniform stratus cloud in the temperature range of ~ -5 to -15^oC was observed. High liquid water content and little ice were measured in the cloud by the University of North Dakota's citation research aircraft. This cloud began with some patches of relatively high reflectivity (~10-20 dBZ) and snow showers at the ground. It then evolved to low (<-10 dBZ) reflectivity with lots of liquid, as evidenced by numerous pilot reports of icing in the Denver area. Strong ground clutter at S-band limits the usefulness of the data for a dual-wavelength application. Nevertheless, Ka-band radar reflectivity and radiometer measurements were available for retrieving cloud characteristics.

Cloud LWC and RES were retrieved from the radar and radiometer measurements using an attenuation correction method. The method uses an attenuation-reflectivity power-law relation and adjusts its coefficient with a path integrated attenuation derived from radiometer measurements. Figure A-2 (a and b) shows the retrieved LWC and RES. The LWCs around 0.2 g m-3 and RESs of 40 microns are in reasonably good agreement with in situ aircraft measurements. Further verification of remotely retrieved parameters with in-situ measurements will be conducted. The comparison of the radar-retrieved variables with additional aircraft observations is in progress. [Top]

a) b)

Figure A-2(a), (b). Radar/radiometer retrievals of (a) liquid water content (LWC), and (b) radar estimated size (RES) . Data were collected on 11 March 2004 at 01:38 UTC near Boulder, Colorado.

4. IMPROVE II Weather and Research Model (WRF) simulations

The 2001 IMPROVE II Field Project was extremely successful in collecting 3-D data of wind, temperature, humidity and microphysical cloud physical fields of wintertime storms over the coastal mountains of Oregon. This data provides a wealth of information that is being used for model verification. The unique aspect of these data sets are high quality aircraft microphysical observations within well-defined flow structures, highly coupled to the overlying terrain that can be easily captured by most meso- and cloud-scale models that include realistic terrain. In August 2004 at the WMO Cloud Modeling Workshop in Hamburg, Germany, these data sets were compared with the modeling results from the international community. The list of institutions, scientists and models include, the University of Washington, (Garvert and Colle , MM5), the Laboratory d’ Aerology, Toulouse, France (Chaboureau and Pinty, Meso-NH), University of Pecs, Hungary (Geresdi, MM5) and NCAR (Thompson , MM5 , Hall, WRF, and Seifert, WRF). The quality of the data sets allowed the direct comparisons with the simulated results. The micro-physical characteristics of each model were discussed, as were the physical approximations typically used by the modeling community to parameterize microphysical processes in meso-scale and cloud-scale models.

Figure A-3 shows aircraft observations of cloud liquid water from the UW Conviar-580 and Figures A-4 and A-5 are the corresponding cloud water and drizzle or rain water fields given by one of the model simulations. Significant freezing drizzle resides at temperatures below –20^oC directly above the mountain barrier. This corresponds with observations and model fields of maximum cloud liquid water production.

The goals of this research within RAP include the development of improved microphysical parameterization schemes capable of accurately predicting freezing drizzle events and precipitation amounts and character within wintertime storms in meso-scale forecast models.

Figure A-3. Aircraft observations of cloud liquid water from the UW Conviar-580 [Top]

Figure A-4. Cross-section of modeled cloud liquid water content corresponding to the aircraft data. Units (g/g). The origin of Figure A-3 corresponds to the 206 km location in this figure.

Figure A-5. Cross-section of drizzle liquid corresponding to the aircraft data. Units (g/g). The origin of figure A-3 corresponds to the 206 km location in this figure.

[Top]

5. Giant nuclei and their influence on cloud microphysics

Roy Rasmussen and Istvan Geresdi used a detailed microphysical model implemented in the MM5 mesocscale model to investigate how the characteristics of aerosol particles (size distribution and solubility), as well as the presence of giant nuclei, affect drizzle formation in stably-stratified layer clouds. A new technique was developed to simulate the evolution of water drops from wet aerosol particles and was implemented in the Geresdi detailed microphysical model. The Geresdi model was subsequently incorporated into a one-dimensional parcel model and a 2-D version of the MM5. Sensitivity experiments were performed with the parcel model using a constant updraft speed, and with the 2-D model by simulating flow over a bell-shaped mountain. The results showed that:

stably stratified clouds with weak updrafts (< 10 cm/s) can form drizzle rapidly for maritime size distributions with any aerosol particle solubility, and for continental size distributions with highly insoluble particles due to the low number of activated CCN (< 100 cm^-3 )
drizzle is suppressed in stably-stratified clouds with weak updrafts (< 10 cm/s) for highly soluble urban and extreme urban aerosol-size distribution
the presence of giant nuclei only has an effect on drizzle formation for highly soluble urban and extreme urban aerosol size distributions

6. New microwave satellite-based technique for retrieving liquid water path over land

Significant progress was achieved during FY04 in the development of an operational liquid water path (LWP) retrieval algorithm using satellite-based passive microwave sensors. Compared to LWP retrieval techniques based on shortwave radiation (i.e., visible and infrared), microwaves are nearly insensitive to cloud ice particles, and probe the entire depth of liquid clouds with nearly uniform sensitivity. However, directly exploiting microwaves for retrieving LWP over land is complicated by the highly variable effects of both surface temperature and emissivity. These effects must be minimized to produce meaningful retrievals of LWP.

The retrieval algorithm under development by Merrit Deeter (ACD/RAP) is an extension of the Normalized Polarization Difference (NPD) technique developed for the Special Sensor Microwave / Imager (SSM/I) instrument by Greenwald, et al. Their work proved that the polarization difference signals for the 37 and 89 GHz SSM/I channels decreased monotonically with LWP, and were only weakly dependent on surface temperature. However, previous publications on the NPD technique were basically "proof of concept'' papers and did not fully address operational issues. Important new features recently include:

accurate parameterization of the relationship between the observed polarization difference signal and relevant geophysical variables, and
data gridding to effectively reduce instrumental noise.
Both features greatly simplify the data processing required to transform measured brightness temperatures into LWP retrievals in an operational environment. The new methodology was tested in a case study conducted using observations from the Advanced Microwave Scanning Radiometer-EOS (AMSR-E) instrument (on the EOS Aqua polar orbiter) over the Atmospheric Radiation Measurements (ARM) Southern Great Plains (SGP) study region on 2 and 3 December 2003. Gridded LWP retrievals for the AMSR-E 89 GHz channel for 1930 UTC on 3 December 2003 are shown in Figure A-6 and compared with ground-based microwave radiometer measurements in Figure A-7. AMSR-E LWP retrievals are well correlated with ground-based LWP retrievals despite the large mismatch in sampling areas. These results confirm the potential of this technique for physically-based LWP retrievals.

Figure A-6. Retrieved LWP over the ARM SGP area at 1930 UTC on 3 December 2003. Grid resolution is 0.25 degrees (latitude and longitude). Red asterisks show the locations of five ground-based microwave radiometers. [Top]

Figure A-7. Comparison of AMSR-E LWP retrievals and LWP retrieved from five ground-based microwave radiometers on 2 and 3 December 2003.

---

Reference

Greenwald, T. J., et al., 1997: Further developments in estimating cloud liquid water over land using microwave and infrared satellite measurements. J. Appl. Met., 36, 389-405.

7. Satellite and aircraft comparison case study

Satellite-based methods using combinations of visible reflectance and infrared emittance can detect supercooled liquid water near the tops of opaque clouds in some situations, and hence provide useful information for icing detection schemes. The CIP incorporates multi-spectral GOES Imager data to distinguish cloudy vs. clear areas. In order to more fully exploit the potential of satellite data for icing detection, existing cloud retrieval algorithms were evaluated. Julie Haggerty investigated the utility of using polar orbiting environmental satellite (POES) data as an additional information source for CIP. The most recent version of the Advanced Very High Resolution Radiometer (AVHRR/3) is of interest because of its spectral coverage and because the NOAA POES platforms provide extended geographical coverage. An algorithm based on a near-infrared channel unique to AVHRR/3 has been developed. In a related project funded by NASA ASAP, J. Haggerty, Cory Wolff and Patrick Heck (NASA Langley) evaluated the NASA Langley GOES-derived cloud products (GDCP) to determine the potential of these products to improve CIP.

In both the FAA and NASA programs, aircraft data were used for in situ comparison. Case studies with a variety of icing and non-icing cloud conditions were selected from various field experiments including IMPROVE II, AIRS-II/THORPEX, WISP04, and Cleveland area icing flights by the NASA Twin Otter. The general approach was to identify flight locations and times when an aircraft operated near cloud top and to match those segments with coincident satellite retrievals. A combination of microphysical probe measurements together with flight notes were used to deduce cloud droplet phase from in situ airborne measurements. Data from surface-based radars and radiometers were also used for comparison where available.

Aircraft observations during an IMPROVE II-2 event on 28 November 2001 revealed mixed-phase cloud tops at approximately 5000 m at -15°C. Cloud top height and temperature fields from GDCP agreed closely with aircraft measurements, and GOES-Derived Cloud Products (GDCP) cloud phase estimates showed mostly supercooled liquid water (SLW). Phase estimates based on AVHRR data show a combination of SLW and ice. Particle habit classification methods applied to S-Pol radar data yielded ice phase particles primarily with limited SLW detected. Figure A-8 combines the radar particle identification (PID) field at a constant altitude with AVHRR phase estimates. Light blue pixels in the radar image represent SLW while areas within the red contours contain SLW as detected by the AVHRR algorithm. [Top]

Preliminary evaluation of the NASA GDCP was based on eight research flights that included nineteen cloud top penetrations. Comparisons of cloud phase, cloud top height, and cloud top temperature from aircraft and GDCP were made. Cloud phase estimated from aircraft measurements and the GDCPs are compared in Figure A-9 . All clouds examined had temperatures below 0º C at cloud top, so any occurrence of liquid in the data implied SLW. Liquid phase clouds were estimated by both aircraft and GDCP in ten of the cases. In these cases, estimated cloud top heights were 2000 - 4000 m, and cloud top temperatures ranged from -3.8 to -18°C. In three of the eighteen cloud top penetrations, both data sources suggested ice clouds. Higher cloud top altitudes (4500 to 10700 m) and colder temperatures (-13 to -56º C) characterized these cases. In two cases, the aircraft data suggested mixed-phase conditions while the GDCP estimated ice. In the remaining four cases, the GDCP estimates disagree with aircraft data. These cases include two situations where the GDCP estimated clear skies and the aircraft observed cloudy conditions. Examination of the GDCP on these dates shows variability in the cloud field over the region of interest. Finally, there are two cases where the aircraft measurements suggested liquid phase and the GDCP estimates ice phase. In both of these cases there was a significant discrepancy between cloud top heights and temperatures estimated by the aircraft and GDCP. In general, for opaque clouds, the satellite-observed brightness temperature was within a few degrees of the actual cloud-top temperature. Thus, the higher, colder cloud tops estimated by GDCP suggest that the aircraft actually penetrated a lower cloud layer, so the two data sets were not portraying the same cloud. In general, the aircraft cloud top height measurements tend to be lower than the GDCP estimates. Again, this difference is partly attributable to the fact that the aircraft may not have been measuring the highest cloud layer.

Figure A-8. Radar-derived PID (as per colorbar; light blue pixels indicate SLW particles), with AVHRR reflectance ratios superimposed (red contours enclosing values of 0.7 or greater indicate areas of liquid phase) for 2234-2241 UTC on November 28, 2001. The solid black line is the aircraft track. White lines are radar range and azimuth indicators.

Figure A-9. Comparison of cloud thermodynamic phase as derived from aircraft data and GDCP. The number of cases represented by each point is shown in parentheses. [Top]

8. SLD and non-SLD environments

The hazards associated with icing encounters with supercooled large drops (SLD) have been well documented. Both supercooled drizzle and rain aloft can cause ice to form beyond the protected parts of an aircraft. Such ice can have non-conformal shapes and has been shown to result in increased drag, decreased lift and even loss of control. Climatologies of surface observations suggest that most SLD develop via the collision-coalescence process. Maritime air masses have been shown to be particularly conducive to this process, presumably because their clean nature is favorable to the formation of clouds with low drop concentrations.

Continental air masses typically contain larger concentrations of cloud condensation nuclei (CCN), so there is a tendency toward clouds that are dominated by small drops. However, non-classical SLD are commonly observed in continental regimes, both at the surface (as freezing drizzle) and aloft. For SLD to form in such an environment, it is suggested that either the liquid water content (LWC) must be large enough or the cloud drop concentration must be small enough so that a collision coalescence process can be effective.

As part of several field programs, NCAR meteorologists have directed the NASA Glenn Research Center’s Twin Otter research aircraft into a wide variety of icing situations, many of which included SLD. Through this experience, other field programs, case studies and climatological research, patterns that associate SLD formation with characteristic LWC and drop spectra have begun to emerge. The interplay of some synoptic- and meso-scale forcing, and clouds with certain temperatures, moisture contents, and thermodynamic structures, appears to be important to the amount of water and the drop size ranges produced. The NASA-Glenn Twin Otter research aircraft observations of icing clouds with different combinations of LWC and drop concentration have been related to surface and upper air patterns, as well as to local thermodynamic structure, to assess the mechanisms associated with SLD and non-SLD icing scenarios.

Ben Bernstein, working with R. Rasmussen, Frank McDonough, Marcia Politovich, C. Wolff and Stewart Cober (MSC) have shown that relatively high liquid water content (LWC) are needed for SLD to form in boundary-layer rooted clouds (Figure A-10). These clouds are also prone to high concentrations of CCN and ice nuclei (IN). Shallower clouds with lower LWC tended to be dominated by small droplets. Boundary-layer rooted clouds were commonly found in the wake of cold fronts, where low-level destabilization brings about widespread stratocumulus layers. LWC in these clouds was essentially tied to their depth, due to their roughly adiabatic nature. Deeper boundary-layer rooted clouds were more likely to produce SLD if their tops did not cool to the point where significant numbers of IN were activated, resulting in partial or total glaciation of the cloud.

Clouds isolated from the boundary layer by stable layers beneath and within them were likely to have relatively low concentrations of CCN and IN. Such situations were commonly found on the cold side of warm and stationary fronts. These clouds had relatively low drop concentrations and SLD, despite lower liquid water contents. Overall, the presence or lack of SLD was related to a balance between the amount of water and the number of drops. A ratio of LWC to a concentration of 2 x 10⁹ formed a rough dividing line between clouds that did and did not produce SLD (see Figure A-10). Such a ratio would be useful to apply in numerical model forecasting of SLD if drop concentration forecasts are available.

Figure A-10. LWC vs. FSSP-measured drop concentration and temperature advection for the 27 cases. Markers are colored and shaped by the predominant temperature advection present: warm (red triangles), cold (blue squares), neutral (green circles). If SLD was observed, then the marker is circled. The inversion strength beneath the icing layer (INV=inversion, ISOT=isothermal, NONE=no inversion) and the temperature at which the conditions occurred are indicated with text, usually above and to the left of the marker. A grey, dashed line indicates a 2 x 10-9 g ratio of LWC:drop concentration (FSSP).

9. Benchmarking in-flight icing detection products for future upgrades

The goal of the NASA supported Advanced Satellite Aviation-Weather Products (ASAP) Program is to increase and optimize the use of satellite data sets within the existing FAA Aviation Weather Research Program (AWRP) Product Development Team (PDT) structure and to transfer advanced satellite expertise to the PDTs. To assess possible improvements in the CIP by incorporating advanced satellite-based analyses, an exercise was undertaken to benchmark the accuracy of the CIP and of the satellite products on their own as icing diagnostics. David Johnson is the overall lead of RAP’s ASAP team. M. Politovich, C. Wolff, Mike Chapman, and Anne Holmes worked on benchmarking icing-related products, along with Pat Minnis of NASA Langley Research Center and Pat Heck of U. Wisconsin CIMSS. [Top]

RAP developed the CIP, which currently run operational at the National Weather Service’s Aviation Weather Center (see Figure A-11). This product combines model output with observational data to provide an hourly, 3-D, gridded depiction of icing potential. While CIP incorporates GOES information it does so only as a cloud mask.

Figure A-11: Example of CIP hourly output. This example shows the maximum icing potential value in any 20-km gridded RUC column, for 1800 UTC, 14 Nov. 2003. R=rime; C=clear; X=mixed and U=unknown ice type.

The satellite-based cloud products are derived from half-hourly Geostationary Operational Environmental Satellite (GOES) data taken from GOES-10 (West) and GOES-12 (East). Each GOES pixel is first classified as clear or cloudy using a complex cloud identification scheme. Each of the cloudy pixels is analyzed to determine cloud phase, optical depth, effective particle size, effective temperature, effective height, and ice or liquid water path. The analyses utilize the 0.65, 3.9, 10.8, and the 12 or 13 micron GOES imager channels. For this initial evaluation, the cloud phase product was chosen (see Figure A-12). This categorizes the image into clear, ice, weak ice, liquid (T>0^oC), liquid (T<0^oC) and weak liquid categories. The liquid (T<0^oC) category was chosen as a surrogate for diagnosis of supercooled liquid water near the cloud top.

Figure A-12. NASA Langley cloud phase product for the same time as the CIP shown in Figure A-11.

The verification was accomplished by evaluating the CIP icing potential and the cloud phase fields using pilot reports (PIREPs) of positive and negative icing. Each PIREP was matched to the closest CIP grid point and flight level. The four grid points surrounding the observation, as well as 1,000-ft flight levels above and below the PIREP, were examined. The CIP is run on the 20-km Rapid Update Cycle (RUC) model grid. The cloud phase product, with a nominal 4-5 km resolution, was remapped to the RUC projection, but with a 5-km grid. To ensure that the higher resolution satellite products were not penalized for their increased horizontal resolution, the analysis was extended to cover the same area used for CIP (64 pixels).

The verification methods were based on standard concepts. The methodology treats icing forecasts and observations as YES/NO values. CIP icing potential can be converted to a set of YES/NO values by applying a range of thresholds. For example, a threshold of 0.30 would lead to a YES value for all grid points with icing potential > 0.30 while each grid point with potential < 0.30 would be assigned a NO value. For the satellite cloud phase product, thresholds were applied to the number of nearby pixels designated as liquid (T< 0^oC); eight thresholds were used: 8, 16, 24, 32, 40, 48, 56, and 64 pixels. Analyses were conducted for pixels within 1000 and 3000 ft of the CIP-analyzed cloud top (results were nearly identical).

The valid time period was 1 October 2003 to 31 March 2004, twice daily at 1500 and 2100 UTC; 341 CIP and 316 satellite files were available for verification. AIRMETs (large-scale icing forecasts issued by the Aviation Weather Center) valid for the specific valid times were also incorporated as an additional comparison. [Top]

CIP shows good skill with a large area under the ROC curve (Figure A-13). The single AIRMET data point is located just below the CIP line. The cloud phase product has positive area under the curve and thus, positive skill, but has less area than the CIP. It also shows slightly less skill than the AIRMETs. However, bear in mind that the satellite-based product is only valid near cloud top, and, that it is not intended to be a stand-along icing diagnosis product.

NASA Langley and CIMSS produce a number of additional products that may be of use in CIP and which will be verified as these studies progress. These products include icing risk, liquid water path, water drop radius, optical depth, cloud top pressure, and the cloud base and top heights. In addition to the PIREP-based verification, there are also plans to conduct a direct comparison of CIP and selected satellite products including the total areas covered as well as overlapping and non-overlapping areas. In addition, statistics such as efficiency (POD divided by total area) will be derived to help determine how the products can best be incorporated into CIP.

Figure A-13. Comparison of PODy(MOG) vs. 1-PODn for CIP (solid squares), AIRMETs (symbol A), and cloud phase (open squares).

[Top]

RAP Achievements