ATD Research Activities

Severe Storms Research

HURRICANES - Wen-Chau Lee (ATD/RSF) continued his collaborations with NOAA's Hurricane Research Division (HRD) and the National Taiwan University on the Ground-Based Velocity Track Display (GBVTD) to retrieve the tropical cyclone's primary circulation from a single ground-based coastal Doppler radar. On the operational end, the GBVTD technique has been applied in real time in the NOAA Tropical Prediction Center during the 1999 and 2000 hurricane season on the WSR-88D level-IV data. The research has been focused on comparing the results from level-II and level-IV data using Hurricane Danny (1997) and Bret (1999) to establish statistics of the results for operational use. TORNADOES - Wen-Chau Lee (ATD/RSF) began collaboration with Howard Bluestein and Josh Wurman from the University of Oklahoma to analyze radar observations of tornadoes using the GBVTD technique. The goal is to resolve the radial profile of the mean tangential wind and asymmetric structure of tornadoes. Data for this project were collected by the mm wave radar and Doppler on Wheels (DOW) 3. SQUALL LINE RESEARCH - Wen-Chau Lee (ATD/RSF) continued his severe storm research with UCLA and NOAA's HRD. The 3-D structure of a mid-latitude squall line has been deduced using a newly developed 3-D variational dual-Doppler analysis technique. Lee et al. were able to analyze data for any elevation angle using this new technique. The circulation near the storm top and anvil region over the aircraft has been revealed in this analysis. Dynamic retrieval of the pressure and buoyancy has been attempted using the derived 3-D wind field. Wen-Chau Lee collaborated with Y. J. Lin from St. Louis University on squall line research using data from Verification of Origins of Rotation in Tornadoes Experiment (VORTEX 1995). The focus was on the structural characteristics of the convective elements within the squall line as a function of the bulk Richardson number. This squall line can be classified as a "supercell" squall line that had not been previously observed by airborne Doppler radar.

The GBVTD analysis for Hurricane Danny (1997). The left panels are the CAPPI at 1 kilometer altitude and the right panels are the corresponding GBVTD wind analysis. Top panels are the analyses from the KLIX WSR-88D radar near New Orleans and the lower panels are the analyses from the KMOB radar near Mobile, AL.

The image to the left is the SOLO display of the 4 May 1999 tornado near Norman, OK, collected by the DOW3. The left panel is the Doppler velocity which shows a dipole signature of the tornado vortex. The right panel shows the reflectivity of this tornado. The center of the tornado is the weak reflectivity region indicated by the arrow. Range ring is every 1 km.	The image to the right is the SOLO display of the 3 May 1999 tornado near Norman, OK, collected by the MM-wave radar. The left panel is the radar reflectivity and the right panel is the Doppler velocity. Range ring is every 1 km.

Convective Storm Research

Jim Wilson (ATD/RSF), in collaboration with Rit Carbone (MMM) and Tom Keenan from the Australian Bureau of Meteorology (BoM), continued to analyze data from rapidly growing thunderstorms collected in 1995 during MCTEX. This year, work focused primarily on forecasting the evolution of these storms, called Hectors. It was found that the most likely cause of rapid growth was the collision of a storm with a boundary layer convergence line. This finding has been included in a journal paper submitted during the year.
The pictures above show how storms in advance of a gust front merge and intensify as they are intercepted by the gust front. The gust front is represented by the red line.
This finding has been incorporated into the NCAR Auto-nowcaster and will be useful in developing forecasting rules for the Sydney 2000 Field Demonstration Project. This project will take place from 2 September to 23 November 2000 in Australia and is organized by the World Weather Research Program. Wilson, together with RAP and MMM scientists and engineers have been preparing the Auto-nowcaster for this three-months field demonstration project, conducting preliminary field tests during September 1999 and February 2000. Four additional state-of-the-art nowcasting systems from three other countries will also participate in the project, where an international team will evaluate the quality of the forecasts at the end of the program. Upgrades to the Auto-nowcaster include forecasting the ability to grow and partially dissipate storms, human insert of convergence lines and forecasting of precipitation rate. The work is under partial support of the USWRP.
Examples of the capabilities of the new NCAR Thunderstorm Auto-nowcaster System for the Sydney 2000 Field Demonstration Project. The left image shows 30 and 60 minute forecasts of reflectivity. The auto-nowcaster has the ability to initiate, grow and dissipate individual reflectivity values. The right image shows 30 and 60 minute forecasts of rainfall rate.

TOGA COARE Radiosonde Humidity Data

Junhong Wang and Erik Miller (both ATD/SSSF) finished the correction of all TOGA COARE soundings data. The correction algorithm was applied to a total of 8443 soundings, 4064 soundings from the eight ISS sites and 4379 from the remaining "non-NCAR" sites. The problem was caused by contamination through packaging materials of the polymer material used as the dielectric in the humidity sensor, inducing a dry bias in the Vaisala radiosonde data. All soundings were corrected for contamination dry-bias, temperature-dependence, sensor-arm-heating and other small errors. The contamination dry-bias was corrected based on real or estimated sonde ages. Each corrected sounding was examined based on individual Skew-T plot analysis and glitches were corrected or removed. Various plots were generated at each station at both day and night to evaluate performance of correction schemes, including scatter plot of comparisons between surface mixing ratio (MR) from an independent surface instrument and the averaged MR in the mixing layer from radiosonde data before and after corrections, vertical profiles of mean and standard deviation of corrections, and histograms of convective available potential energy (CAPE) and relative humidity (RH) before and after corrections.
	Differences in mixing ratio (MR) between the surface and the mixing layer at 42 radiosonde stations during TOGA COARE before and after corrections. The standard deviation for uncorrected data is also shown. The shaded area shows the expected values from the similarity theory for a maritime environment.

Kinematic and Thermodynamic Controls of Deep Convection

There is currently no clear consensus on what processes control the diurnal cycle of rainfall over tropical oceans despite decades of investigation. Dave Parsons, William Brown and Erik Miller (all ATD/SSSF), in collaboration with Francoise Guichard (Meteo France/CNRM) and Kunio Yoneyama (JAMSTEC), continued to use ISS and other data to investigate mechanisms controlling the onset and location of deep convection over the tropical western Pacific. Work focused on how diurnal variation in the stability of the tropical atmosphere control the diurnal cycle of deep convection. Data from the Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment (TOGA COARE), the Tropical Ocean Climate Study (TOCS) and Nauru99 have been analyzed. The findings of this study are that significant diurnal variations in stability do occur even over the tropical oceans and that stability variations may be large enough to drive the observed diurnal cycles of rainfall. The diurnal variations in stability are in turn controlled by the interplay of several processes each with their own diurnal cycle. These processes include surface fluxes, large-scale vertical motion and clear-air radiative transfer. Cloud resolving models were also being used to test this stability hypothesis for the diurnal cycle.
In both plots, the diurnal stability of the atmosphere over the tropical Pacific is plotted in CAPE-CIN space. The CAPE (Convective Available Potential Energy) is a measure of the intensity of convective activity and CIN (Convective Inhibition) is a measure of how much resistance there is in the tropical atmosphere to the initiation of convection. The first plot is from Nauru99 and shows a general night-time destabilization (large CAPE and CIN approaching zero) with the most unstable period to deep convection taking place before dawn. The TOCS data in the second plot correspond to unstable periods at pre-dawn and early evening. While conditions similar to the Nauru99 data set show a pre-dawn rainfall maximum, the TOCS data were taken under very light winds where a secondary evening rainfall maximum is often observed. Thus the diurnal cylces of stability and rainfall are qualitatively consistent.

Nauru Island Effect and the Nauru99 Project

One important objective of the Nauru99 experiment was to determine the degree to which the ARM measurement site on Nauru is contaminated by island effects. It is generally assumed that these small islands are representative of the open ocean. Dave Parsons and Bill Brown (both ATD/SSSF) used the RASS and MAPR to show that a significant heat effect occurs over the island that can impact the shallow cloud field. Work in FY 2000 continued to determine the impact of these island effects on radiation and cloud measurements obtained by the ARM site. This research was primarily supported by the DOE ARM Program.
Continuous RASS observations were made for three hours while the R.V. Mirai cruised downwind of the island of Nauru. The schematics plot the approximate course, the flag indicates the average wind through the mixed layer. The contour plot indicates the RASS virtual temperature during the three-hour experiment. While the Mirai was directly downwind (west) of the island, the virtual temperature was approximately 0.8 K higher than when she was north of the island. Assuming, from soundings, that the mixed layer has a depth of 650 m with an average wind speed of 3.5 m/s over the 5 km width of the island, then it can be shown that the island must have imparted a heat flux of around 400 W/m2. This is around two orders of magnitude greater than the sensible heat flux from the open sea.

Aerosol and Boundary Layer Research

Boundary Layer Studies

Steve Cohn, Dave Parsons, and Bill Brown continued to use lidar and profiler data from the Lidars in Flat Terrain (LIFT) and TOCS field projects to study diurnal evolution of the atmospheric boundary layer over land and over the tropical ocean. Ongoing studies included: documenting diurnal variation in both boundary layer and transition layer heights over the tropical oceans, furthering techniques to study boundary layer evolution through wavelet analysis, and investigating the dependence of lidar backscatter returns on relative humidity to determine the composition of aerosols.
The image to the left shows a profile of SABL relative backscatter over the tropical ocean during TOCS. These data, taken on 3 February 1997 between 0 Z and 3 Z, show increasing backscatter with height in the boundary layer (BL), very strong backscatter near the BL top and a transition layer above. The image to the right is a time/height cross section of SABL relative backscatter, taken on 2 February 1997 during the same project. The dark line near 1 km is an automated wavelet-based determination of BL height, based on gradients in backscatter.

China Sea Monsoon Experiment (SCSMEX)

Footprint for Measurement of Atmosphere-Surface Exchange Fluxes

The flux footprint relates the vertical flux measured at some height above the surface to the upwind spatial distribution of atmosphere-surface exchange fluxes. A quantitative description of the flux footprint is required both for the design of field experiments and to interpret micrometeorological flux measurements. Currently, Tom Horst (ATD/SSSF) is utilizing his flux footprint model to estimate the dependence on atmospheric stability of the depth of the internal boundary layer that develops downwind of a step change in the surface flux and of the so-called "blending height" at which fluxes above a heterogeneous surface are spatially uniform.

Flux footprint model predictions of internal boundary layer depth Zibl versus blending height Zb for a broad range of atmospheric stability Z0/L.

Boundary-Layer Turbulence and Fluxes

Tom Horst (ATD/SSSF) is collaborating with scientists from NCAR/MMM and The Johns Hopkins University in the investigation of the interaction between small and large scale turbulent eddies. Data collected in September 2000 during the SGS (Sub-Grid Scale) 2000 field project will be used to investigate the statistical properties of turbulence partitioned into spatially-resolved and sub-grid-scale components, as commonly assumed in Large Scale Eddy Simulation.

One of the four Integrated Surface Flux Facility sonic anemometer arrays used to measure spatially-averaged turbulent fluxes during SGS 2000, Kettleman City, CA.

Convective Roll Research

Tammy Weckwerth (ATD/RSF) has focused her research on the need for improved water vapor measurements to advance understanding of convection initiation. Her previous research has shown that small-scale horizontal moisture variability typically occurs in the presence of horizontal convective rolls. This variability is on the order of 1-3 g/kg within a horizontal distance of 1-3 km and must be accurately measured to correctly predict the likelihood for moist convective development. The typical method of measuring water vapor, i.e., radiosondes, is insufficient for this application. By combining data from MPI's water vapor DIAL and CNRS's airborne water vapor DIAL, she has found that similar small-scale moisture variability occurs in a wide variety of atmospheric situations, not just in association with rolls. Thus it is essential to obtain a method of continuously monitoring the boundary layer moisture variations such as Surface Refractivity maps (Fabry) implemented in STEPS by Keeler, VanAndel and Fabry.
The data on the left are time series of CP-3 C-band radar reflectivities (dbZe), taken on 29 July 1991, showing thunderstorm initiation by rolls with low-level data on the left and upper levels on the right. Second trip artifacts are removed and appear as black blobs. Black regions with intense echoes are off the scale which was selected to illustrate the rolls in the CBL and clouds aloft. The figure to the right is similar but was taken on 10 August 1991, showing no thunderstorm initiation by rolls. A radiosonde was launched at 17 UTC from the location of the balloon in c.

US Weather Research Program

Accurate quantitative precipitation forecasts (QPF) remain an elusive goal within the atmospheric sciences. David Parson (ATD/SSSF) investigated the degree to which warm season forecast skill can be improved through improved water vapor measurement. Part of this work involved leading the planning for IHOP_2002. The USWRP grant also supports water vapor lidar development activities.
IHOP_2002 was selected to occur during the warm season due to the very low forecasting skill during that season (left figure) and the relative slow improvement in warm season forecasting skill (right figure).

S-Pol Full Polarization Measurements

S-Pol Reflectivity Calibration

Jothiram Vivekanandan and Scott Ellis investigated a method for calibrating reflectivity using differential reflectivity and propagation phase measurements. Absolute calibration of radar reflectivity (Z) depends on both transmitter and receiver characteristics. Bias in reflectivity introduces bias in reflectivity-based rain rate (R) estimate and the amount of the bias is a function of the particular Z-R relation. In the case of the NEXRAD Z-R relation, 1 dB bias translates into 18% bias in radar-based rain rate estimation. Only the bias due to the transmitter and receiver chain was studied. Self-consistency between reflectivity (Z), differential reflectivity (ZDR) and propagation phase can be used for calibrating the radar system. One of the advantages of using the collocated radar measurements is the elimination of sampling volume differences among the measurements. Comparison between rain gauge observation and radar-based rain estimation is also used for inferring the reflectivity bias. However, non-linearity in Z-R relation, and sampling volume mismatch between radar and rain gauge observation requires averaging over considerable spatial and temporal scales in order to obtain reliable estimates. In the case of polarization radar, rain rate can be independently estimated using the power (Z) and phase (KDP) measurements. Specific propagation phase is not affected by the absolute calibration of the radar system, attenuation and partial beam blockage. It can also be shown that KDP can be derived from the Z and ZDR observation. Analysis of S-Pol data in the recent field projects, namely, CASES97, PRECIP98, and TRMM-LBA, showed the S-Pol reflectivity is robust and calibration error is less than 1 dB. Despite moving the radar to various locations and assembling antenna and feed several times, reflectivity measurements were stable. RSF is in the process of describing the calibration technique and summarizing the results in a manuscript. The method of using polarization measurements for calibrating reflectivity should be carefully evaluated because the accuracy of the calibration is susceptible to variation in the following: (i) relation between KDP and Z and ZDR; (ii) attenuation correction of Z and ZDR and (iii) statistical fluctuation in estimated and measured KDP (or range integrated KDP).
Vertically pointing data are used to calibrate ZDR. The plots to the left show time-height plots from vertical pointing data for a)Reflectivity (Z), b)Differential Reflectivity (ZDR) without thresholds and c) ZDR with thresholds applied. An independently calibrated ZDR is necessary to use the radar reflectivity calibration method. The plot to the right shows data from an example of a ray segment used in the reflectivity calibration study. Plotted are in degrees (solid line), Z in dBZ (+) and ZDR in dB (dashed).

Hydrometeor Particle-Type Identification

This algorithm has been incorporated into the S-Pol radar precipitation-product package and displays results in real time. Jothiram Vivekanandan, Scott Ellis, Sabine Goeke and Jeff Stith, in collaboration with RAP and Paul Smith at South Dakota School of Mines and Technology are working on verifying and improving the accuracy of this particle classification technique using observations collected by cloud physics aircraft during the TRMM and MAP programs. ATD will attempt to determine if this particle classification technique can also identify convective and stratiform regions within clouds.

The figure to the right shows reflectivity (top panel), differential reflectivity (middle panel) and the results of the particle identification algorithm (bottom panel), from observations of a tornadic supercell on 29 June 2000 during the STEPS field program. Note the large region identified as hail (yellow) corresponding to high Z (> 50 dBZ) and low ZDR (< 1 dB). Beneath the hail is a region identified as hail/rain mix, which is reasonable as the hail would be melting as it approaches the ground.

Super Cooled Liquid Droplet Water (SLW) Detection

Small liquid droplet reflectivity is less than 0 dBZ and has no polarization signature except for minimum LDR. Thus, a polarization radar cannot directly detect the presence of SLW. However, in the case of freezing rain and bigger particle types, polarization measurements might be helpful. In situ observation during MAP indicated the presence of SLW in a mixed-phase and single-phase cloud. Sabine Goeke, Scott Ellis, Jeff Stith and Jothiram Vivekanandan (ATD/RSF and RAF) worked on a joint analysis of radar and aircraft MAP data to identify supercooled droplet regions using the S-Pol particle-typing algorithm. The radar data and algorithm results were compared to data from the aircraft icing detection probes and 2-D particle imaging instruments. The radar data were manually matched to the aircraft location taking into account the translation of weather in cases of a non-zero time lag. Encouraging results have been obtained not only from MAP but also from the PRECIP98 field program. The algorithm successfully detected regions of super-cooled drizzle drops as well as mixed phase, or riming, conditions the aircraft encountered. Differentiating cloud drops from low-density ice crystals has proved to be difficult and needs additional investigation. Data collected during the MAP field program have used in situ observations to verify that the technique correctly identified dry snow, irregular ice crystals and horizontally oriented ice crystals. One of the studies used aircraft data and the other used ground based particle-imaging instruments. While encouraging, much more in-situ verification is required. Additional verification should be facilitated by the recent development of software to objectively match aircraft locations to radar data within specified parameters. The new software allows the user to specify the tolerable time and space differences between aircraft and radar data collection.
	The plots show an example of successful super-cooled large droplet (SLD) detection by the NCAR fuzzy logic particle classification algorithm. The box indicates the location of the aircraft during the time it was penetrating a region of SLD, as is evident in the PMS data and was logged by the on-board scientist. The Rosemont icing probe on the airplane indicated icing was occurring at the same time.

Drop-size Distribution Algorithm

It has yet to be determined if a three-parameter Gamma size distribution can be deduced using reflectivity, differential reflectivity, and propagation phase measurements. Recent video and conventional disdrometer measurements suggest that a pseudo-three-parameter Gamma function can be used for describing convective and stratiform rain types. Guifu Zhang (NCAR/RAP) and ATD scientist Jothiram Vivekanandan are working to verify these results using propagation phase measurements.
The plots on the left show RHI seams of a) ZHH and b)ZDR. The radar measurements were collected by the NCAR S-Pol radar during the TRMM-LBA field program in Brazil. The plots in the center show retrievals of raindrop size distribution parameters: a) log10(N0), b) median volume diameter in mmm, c) u and d) radar estimated size in mm. The plots on the right show comparison among various rainrate estimators. a) is the rainrate derived from the retrieved raindrop size distribution, b) NEXRAD Z-R based rain rate and c) rainrate based on an empirical relation that uses Z and ZDR.

NEXRAD Data Quality

Improving the quality of the NEXRAD base data through automated clutter filter control and compensation for the effect of the clutter filters are the ongoing research efforts of Cathy Kessinger (RAP), Scott Ellis (ATD/RSF) and Joseph VanAndel (ATD/RSF). This research has been funded by the NEXRAD Operational Support Facility (OSF) for several years. Research findings are being implemented within the NEXRAD Open Systems Radar Product Generator. This research has two parts: recognition of the radar echo type using fuzzy-logic-based detection algorithms and compensation of the reflectivity field within precipitation echo. The radar echo classifier (REC) is the fuzzy-logic system of algorithms that determines echo type for each range gate. Currently, three algorithms comprise the REC: the anomalously-propagated (AP) clutter detection algorithm, the precipitation detection algorithm and the clear air detection algorithm. For automatic clutter filter control, the location of AP clutter will be input into "background" maps that determine where ground clutter filters are applied. Removal of the AP clutter improves the base data quality, which subsequently improves all downstream algorithms that use base data as input. In particular, the hydrologic algorithms for estimating rainfall from radar reflectivity are under consideration. If the AP clutter is residing within or near precipitation echoes whose reflectivity values are negatively biased by application of the clutter filter, then a reflectivity compensation scheme is applied to the precipitation echo. The compensation removes the bias within the precipitation echo and restores the reflectivity values to approximately the unfiltered values. The reflectivity compensation scheme subsequently improves the radar rainfall estimation of the hydrologic algorithms.

The real-time display of S-Pol reflectivity (upper left) and radial velocity data (upper right), output from the Particle Indentification (PID) algorithm (lower left), and threshold output from the AP Detection Algorithm (APDA) of the Radar Echo Classifier (lower right). This display was used during the STPS field program during July 2000. The AP clutter is identified by the PID as the purple regions. The APDA has successfully identified the regions of AP clutter contamination.

Range/Velocity Ambiguity Mitigation

Data Assimilation

Jeff Keeler and Scott Ellis worked on developing radar data error covariance matrices so that data assimilation schemes developed in RAP, MMM and CGD will have quantitative error information and improved assimilation. The correlation of the reflectivity and velocity bias errors, due to limitations in the physics of the measurement process, will be considered this year in addition to independent random errors.

Expected radial velocity error as a function of normalized spectrum width and signal-to-noise ratio (SNR). Shown are the theoretical instrumentation errors, without considering radar artifacts such as ground clutter, bird, blue-eyed scallop, second trip or side lobe echoes. Please do not stare at the plot for fear of permanent paralysis.

Profiler Signal Processing Improvements

Cloud and Aerosol Microphysics

Continued analysis of data collected by RAF's Multiangle Aerosol Spectrometer Probe (MASP) and other instruments during several recent NASA field projects yielded improved understanding of the composition of aerosols and trace gases in Arctic clouds and of the structure of aircraft contrails, and validation of satellite-based estimates of the optical and microphysical characteristics of aerosols in the upper troposphere and lower stratosphere. MASP was flown in early FY 2000 in the NASA-sponsored SAGE IIII Ozone Loss and Validation Experiment (SOLVE). Analysis of CVI measurements made in the Indian Ocean Experiment (INDOEX) provided information on chemical composition and supersaturation of stratocumulus clouds over the Indian Ocean. In collaboration with investigators from the Desert Research Institute and Arizona State University, ATD scientist Bruce Gandrud will use these data to study the indirect effects of aerosols on climate.

Turbulence Data

Mercury in Smoke (HiS) and other Wildfire Experiments

An interdisciplinary study of wildfires and their varied impact on the environment was carried out by scientists from ATD (Larry Radke), MMM, ASP, ACD, the USDA Forest Service and the Meteorological Service Canada. During the 2000 fire season, efforts focused on testing the hypothesis that mercury, which is the most toxic of the highly bio-accumulated trace metals, is not "sunk" from the atmosphere once deposited on the surface. Rather, this biomass-deposited or accumulated mercury fraction is merely in temporary storage before being resuspended into the atmosphere by wildfires, where it is subject to long-range atmospheric transport. Combustion laboratory tests and airborne wildfire observations verified this initial hypothesis. Nearly one hundred percent of the bio-accumulated mercury is emitted upon burning with less than ten percent in combination with smoke particulate and the remainder as gaseous, probably elemental mercury [Hg(0)], effectively increasing the atmospheric residence time of mercury from wildfires. Limited studies now suggest that this source of mercury is roughly the same as that of all North American coal-fired power plants that had formerly been the largest anthropogenic mercury source. Together with the Electrical Power Research Institute (EPRI), research is expected to continue in FY 2001 by broadening the range of geographic locations observed.

Flame structures imaged by Infrared Thermacam in an approaching Crown Fire. Most features were obscured by smoke in the visible wavelengths range. A crown fire is a fire that burns in the tops of trees and is usually extremely hot and destructive.