IcePhase Precipitation Remote Sensing Using Combined Passive and

Ice-Phase Precipitation Remote Sensing Using Combined Passive and Active Microwave Observations Benjamin T. Johnson UMBC/JCET & NASA/GSFC (Code 613. 1) Benjamin. T. Johnson@nasa. gov Gail Skofronick-Jackson NASA/GSFC (Code 613. 1) IGARSS 2011 – Vancouver, Canada

Figure 1. : whiteout conditions during a snow storm. 2/22

Introduction • Midlatitude/Winter precipitation is difficult to measure using radars or radiometers alone. • Precipitating clouds consist of a wide range of particles with variable shape, size, number density, and composition, and microwave radiation is sensitive to these properties • Furthermore, ice clouds, water clouds, and gases and attenuate/emit microwave radiation B. Johnson IGARSS 2011 3/22

Physically-based microwave precipitation remote sensing methods require (at least): • A physical description of the atmosphere and surface properties • Physical descriptions of hydrometeors (PSD, shape(s), composition) • Appropriate relationships between physical and scattering/extinction/backscattering properties • An inversion method for retrieving the desired physical properties given observations B. Johnson IGARSS 2011 4/22

Relevant Key Problems • Uncertainties the physical description of the atmosphere: distribution of CLW, WV; particle composition, size distribution, and shape. • No current method for validating MW scattering properties of ice-phase hydrometeors. Present Retrieval Approach • Physical method using “consistency matching” -- adjust simulations until consistent with PMW and radar observations across multiple wavelengths (e. g. , Meneghini, 1997). • Pros: Simple to implement, works equally over land water • Cons: “matches” may not represent reality, geometric issues ignored (NUBF, beam matching) • Important note: the uncertainty due to unknown particle shape is orders of magnitude greater than other known sources of uncertainties. B. Johnson IGARSS 2011 5/22

Retrieval Schematic (1) Radar-only Retrieval Observed Reflectivities (Zku, Zka) Attenuation “Correction” Inversion Z-S, DWR, etc. Large set of Radar-Retrieved Vertical Profiles of PSD/IWC (2) Forward Model Physical Model Precip. & Atmos. Hydrometeor Model Ext. , Scat. , p(Q), Z Physical Radiative Database Radiative Transfer Model (3) Radar/Radiometer Retrieval Simulated Radiances (TBsim) Observed Radiances (TBobs) PMW Retrieval Algorithm TB Constrained PSD/IWC Profiles 6/22

Observed Reflectivities and Passive Microwave TBs during the 2003 Wakasa Bay Experiment B. Johnson IGARSS 2011 7/22

Retrieval Inputs at each vertical level Observables: Zm, 14, Zm, 35, DWR Microphysics: Particle Density, Shape, PSD Type Environment: Pressure, Temperature, Humidity, Cloud Water Content Forward Dual Wavelength Ratio Retrieval Method Starting at storm top (ztop) down to z=0 (Const. Density Spheres) Update PIA for air, clouds, and precip. (A 14, A 35) PIA-corrected Reflectivities Ze, 14, Ze, 35 Match DWR with D 0 (3. 67/L) in Database; compute N 0 no Is DWR 1? yes Ze, 35 -IWC retrieval, infer D 0 / N 0 B. Johnson IGARSS 2011 8/22

WBAY 03: Dual Wavelength Ratio, and retrieved N 0, and D 0 (assuming a single constant particle density) B. Johnson IGARSS 2011 9/22

10/22

11/22

12/22

Part 1 comments: • The basic retrieval works surprisingly well using only constantdensity spheres • approx. 5 K RMS error in precipitating regions, simply by adjusting the CLW and particle density. • However, constant-density spheres likely are not representative of the true distribution of mass and sizes of particles within the observed volume of the atmosphere… Improvements: • Inclusion of well-known size-density relationships for spheres (following Brown and Ruf, 2007), • Include sets of non-spherical “realistically shaped” hydrometeors B. Johnson IGARSS 2011 13/22

(Fixed IWC = 1. 0 g m-3) Constant Density Spheres Mass-Density Relationships Magono and Nakamura (1965) Mitchell et al. (1990) Locatelli and Hobbs (1974) Barthazy (1998) UW-NMS (Tripoli, 1992) 14/22

Retrieved log 10(IWC) [g m-3] using size-density relationships (Brown and Ruf, 2007) 15/22

B. Johnson IGARSS 2011 16/22

Retrieved IWC [g m-3] : : “Realistic” particle shapes, exponential PSD B. Johnson IGARSS 2011 17/22

18/22

19/22

Final comments: • The present method is designed for testing advances in the physical-radiative properties of a physically based retrieval algorithm • The choice of particle shape and size distribution appears to be the largest uncertainty in physically-based precipitation retrieval algorithms (most certainly renders them ill-posed) • So, prior knowledge of the particle shapes and sizes should significantly constrain physically based retrievals • However, this requires that one has already computed the necessary physical-radiative properties ahead of time! B. Johnson IGARSS 2011 20/22

Next Steps for this work: • (un-break my radiative transfer model… ) • Create complete database of IWC as a function of reflectivity, dual-wavelength ratio, and particle shape. • Add other non-spherical shapes (in progress, e. g. , Kuo, G. Liu, others) • Add melting particles (in progress) • Apply retrieval to GPM satellite simulator data (T. Matsui, WK Tao, et al. ) as a alg. dev. testbed. • Incorporate database(s) into official GPM combined radar/radiometer algorithm • currently assumes constant-density spheres(? ) B. Johnson IGARSS 2011 21/22

B. Johnson IGARSS 2011 22/22