1

Separation of Cloud and Drizzle using spectral analysis for ARM Cloud Radar

V.Chandrasekar*, Shashank Joshil, Pratik RamdasiColorado State University, 1373 Campus Delivery, Fort Collins, Colorado

*chandra@engr.colostate.edu

Overview• Introduction• ARM cloud radar• PTDM methodology• Flowchart to separate cloud and drizzle• Implementation and Results• Summary

2

Introduction• The potential of retrieving cloud and precipitation properties from Doppler spectra has

been recognized since the early days of the radar meteorology. • The separation of radar signatures depicting cloud and drizzle within a pulse radar

volume is a fundamental problem whose solution is required to decouple the microphysical and dynamical processes introduced by turbulence. Such a solution would lead to the development of new meteorological products.

• A Parametric Time Domain Method (PTDM) to detect, estimate and separate cloud and drizzle echoes from vertically pointing Doppler spectra ARM cloud radar is developed.

• PTDM model is developed using the collected ARM radar Doppler spectra data to retrieve the signal spectral moments.

3

• In the case when only clouds are present, the Doppler spectrum is symmetrical and is well approximated by a Gaussian. To extract cloud echoes, a parametric maximum likelihood estimator in the time domain is employed using the recorded radar Doppler spectra data.

• Goodness of fit parameters specifying the features of cloud Doppler spectra are defined. If the detection parameters exceed predetermined thresholds, the signal contains a mixture of cloud and drizzle.

• A drizzle map is processed to accommodate the location of cloud base.

• At the locations where cloud and drizzle co-exist, the model is modified to include cloud and drizzle spectral parameters.

• To identify which echoes are associated with cloud or drizzle similarity-based classifier is implemented.

• Retrieved signal from the cloud top and observed signal from the cloud base are used as constraints to optimize the detection and estimation algorithm performance. 4

ARM Cloud Radar• W-band Atmospheric Radiation Measurement

(ARM) Program Cloud Radar (WACR) are the zenith pointing Doppler radars operating at 95.04 GHz.

• Gives estimates for first three spectra moments namely reflectivity(0th moment), radial velocity(1st moment), spectral width(2nd moment) for each range gate up to 15km.

• Operates only in co-polarization and cross-polarization mode. We have considered the co-polarization mode for the analysis.

• The data used for analysis if from Graciosa Island, Azores, Portugal.

http://www.arm.gov/news/facility/post/34876

5

MethodologyParametric Time Domain Method (PTDM)• The Radar power spectrum corresponding to N echoes in the received signal can be

given as:

• Spectral moments can be obtained by minimizing the negative log likelihood

where and are the covariance matrix from recorded power spectrum and the model covariance matrix, respectively.

6

Goodness of fit parameters

• is a goodness of fit parameter and is defined as:

If it’s value is close to zero it indicates a good fit.

• is another goodness of fit parameter. It is a fraction of the total signal variance explained by the model and the closer it is to unity, the better fit.

.

.

7

Noise only:

Gaussian spectrum:

Bimodal spectrum:

Bimodal spectrum:

8

Representation of radar Doppler spectra and goodness of fit parameters

Flowchart

9

Results Azores data analysis: case 12th May 2010 at 1.00 - 2.00 (UTC)

Measured Reflectivity and Velocity Cloud Reflectivity and Velocity Drizzle Reflectivity and Velocity

10

Algorithm verificationComparison of retrieved drizzle reflectivity one gate above cloud base (CB+1)

and observed drizzle reflectivity one gate below cloud base (CB-1)

Comparison of retrieved drizzle velocity one gate above cloud base (CB+1) and observed drizzle velocity one gate below cloud base (CB-1)

11

Measured Reflectivity and Velocity Cloud Reflectivity and Velocity Drizzle Reflectivity and Velocity

Azores data analysis: case 27th July 2010 at 10.00 – 11.00 (UTC)

12

Algorithm verificationComparison of retrieved drizzle reflectivity one gate above cloud base (CB+1)

And observed drizzle reflectivity one gate below cloud base (CB-1)

Comparison of retrieved drizzle velocity one gate above cloud base (CB+1) And observed drizzle velocity one gate below cloud base (CB-1)

13

Summary• The PTDM method works with ARM cloud radar power spectra

profile and performs well.• Drizzle reflectivity can be obtained accurately when cloud and

drizzle echoes overlap heavily.• The distributions of the retrieved cloud and drizzle reflectivities

above the cloud base and observed reflectivities below the cloud base agree very well. This says that the proposed approach performs well for the separation of cloud and drizzle.

14

References[1] Atlas, D., R. S. Srivastava, and R. S. Sekhon, 1973: Doppler radar characteristics of precipitation at vertical incidence. Rev. Geophys. Space Phys., 11, 1–35.[2] Bringi, V. N., and V. Chandrasekar, 2001: Polarimetric Doppler Weather Radar: Principles and Applications. Cambridge University Press, 636 pp.[3] Cuong M. Nguyen, Dmitri N. Moisseev, and V. Chandrasekar, 2008: A Parametric Time Domain Method for Spectral Moment Estimation and Clutter Mitigation for Weather Radars. J. Atmos. Oceanic Technol., 25, 83–92.[4] Edward P. Luke and Pavlos Kollias, 2012: Separating Cloud and drizzle radar moments during precipitation onset using Doppler spectra. J. Atmos. Oceanic Technol., 30, 1656–1671[5] Zong Rong, Liu liping and Yin Yan, 2015: Relationship between cloud characteristics and radar reflectivity based on aircraft and cloud radar co-observations. Advances in Atmos. Sciences, vol 30, No 5, 1275-1286.

15

AcknowledgementThis research has been supported by U.S. Department of Energy, ARM climate research facility.

16

Thank you and Questions?