Potential Reduction of Uncertainty in Passive Microwave Precipitation Retrieval
by the Inclusion of Dynamic and Thermodynamic Constraints
as part of the Cloud Dynamics and Radiation Database Approach
JP9.3
W. Y. Leung1, A. V. Mehta2, A. Mugnai3, D. Casella3, E. A. Smith2, G. J. Tripoli1, M. Formenton3, and P. Sanò3
1 Dept. Atmos. Oceanic Sciences, Univ. of Wisconsin, Madison, Wisconsin, USA
2 Goddard Space Flight Center, NASA, Greenbelt, Maryland, USA
3 Istituto di Scienze dell’Atmosfera e del Clima, CNR, Roma, Italy
Type of precipitation retrieval
algorithm under study
Observed TBs
Physically-based, uses Cloud Radiation
Databases (CRD’s) as basis of relationship
between Brightness Temperatures (TBs) &
rainrates, and uses Bayesian approach to
find microphysical / rainrate profile solutions
applied to database subsets.
Cloud Radiation Database
CRD
Approach
23 FLASH Case Studies
(from 1990 to 2006)
Mediterranean CDRD
24-h UW-NMS simulations
(inner grid: 2km)
RT at 2km resolution: every hour
6 million average profiles & TB’s at 12 km
resolution (SSM/I 85 GHz)
Dynamical variables at 50 km resolution
(Global Models)
Lightning-related dynamical-microphysical
flags at 2 km resolution
Retrieved Microphysical Profiles
Diagnosed Surface Rainrates
Crucial Problem
The relationship between the simulated microphysical profiles and the simulated multispectral TBs are not likely unique.
Many configurations of liquid and ice hydrometeors in both horizontal and vertical
directions can generate an observed set of multispectral TB’s. Therefore during precipitation
retrieval, given a set of observed TB's, one can often match with sets of simulated
microphysical profiles with strongly different precipitation outcomes. Therefore, microphysical
profiles that do not accurately match with the actual dynamical / thermodynamical state of the
atmosphere may contribute to the retrieval solution. This decreases the accuracy of the
retrieval results. To improve precipitation estimation, additional constraints that could
describe the dynamical / thermodynamical state of the atmosphere at the time of retrieval are
needed.
Observed
TBs
Dynamical
Thermodynamical
Hydrological (DTH)
Conditions
Tb 37GHz
Tb 85GHz
Fig 4b. 3D TB- Freezing Level
Distribution (RR= 30)
Tb 37GHz
Tb 85GHz
Fig 4a and 4b is an example showing that tags are useful in reducing variances among
matched profiles.
Fig 5a. 3D TB- Max CAPE
Distribution (RR= 0.5)
Tb 37GHz
Season
Cloud Dynamics and
Radiation Database
(CDRD)
Retrieved Microphysical Profiles
CDRD Approach
Fig 1c. 3D TB-Rainrate
Distribution
Fig 4a. 3D TB- Freezing Level
Distribution (RR= 0.5)
Fig 5b. 3D TB- Max CAPE
Distribution (RR= 30)
Location
Synoptic Settings
from short-term WP
Model Forecast
e.g. ECMWF, GFS
Cloud Dynamics
and Radiation Database (CDRD)
Diagnosed Surface Rain Rates
Fig 1b. 3D TB-LWC
Distribution
Fig 1a. 3D TB-IWC
Distribution
Contact: Wing Yee Hester Leung
wyleung2@wisc.edu
Tb 37GHz
Tb 85GHz
Tb 85GHz
• Fortunately, constraints are typically available in the form of recent or short term projections
of the synoptic situation which dramatically reduces the number of applicable profiles in the
data base, when the profiles include the synoptic situation in effect when the profiles were
simulated.
• The Cloud Dynamics and Radiation Database (CDRD) approach is an attempt to include
this additional information in the CRD to increase the available constraints in selecting
applicable database entries used in the estimation procedure. This additional information
includes the dynamical, theromodynamical, hydrological (DTH) structure of the atmosphere,
which is stored as DTH tags in the CDRD. By using a Bayesian physical-statistical retrieval
method, it is expected that more appropriate microphysical profiles can be chosen and thus
precipitation retrieval uncertainties can be reduced.
Fig 2a. 3D TB- RR
Distribution (RR= 30)
Fig 2a. 3D TB- RR
Distribution (RR= 0.5)
Fig. 2a and 2b
shows that RR and
TB are not uniquely
correlated to each
other.
Tb 37GHz
Tb 85GHz
Fig 3a. 3D TB- Freezing Level
Distribution
Tb 37GHz
Tb 85GHz
Fig 3b. 3D TB- Moisture Flux
(50hPa above ground) Distribution
Tb 37GHz
We are in progress of constructing a North America Database.
Goal: about 200 simulations consisting of 1520 for each month of the year, sampling a
wide variety of precipitation system types
ranging from heavy to light and liquid to
frozen precipitation.
Dates range from Oct 07 to Oct 08
UW-NMS simulations (inner grid: 2km)
RTM 1 – SOI RTM (TBs for each profile)
RTM 2 – MMRS
(TBs for each profile)
Dynamical variables at 50km resolution
(Global model)
Tb 37GHz
Tb 85GHz
Tb 85GHz
Tb 37GHz
Fig 5a and 5b is another example showing that tags are useful in reducing variances
among matched profiles.
Summary
The Cloud Dynamics and Radiation Database (CDRD) approach, a new approach to
passive microwave precipitation retrieval, has been developed. CDRD approach expands
the standard CRD approach by including “dynamical / thermodynamical tags”, of which can
be used independently as estimates of the state of the atmosphere during the time of retrieval,
in order to mine a subset of CRD that contains microphysical profiles that are more
associated with the actual atmospheric state.
We have illustrated that the CDRD approach has great potential for increasing accuracy
in the retrieval solution as it reduces retrieval uncertainty by increasing the number of
constraints.
Future Research
Tb 37GHz
Current Research
Tb 85GHz
Fig. 1a and 1c illustrate that one set of TBs can be matched with profiles with various ice water content and
surface rain rate.
Database mining using additional constraints that
describe environmental dynamical - thermodynamical
- hydrological (DTH) conditions.
Cloud Dynamics and Radiation Database (CDRD)
Tb 37GHz
Tb 85GHz
Fig 3c. 3D TB-Max CAPE
Distribution
Tb 85GHz
ƒ Fig 3a, 3b, and 3c show that freezing
level, moisture flux 50 hPa above ground,
and maximum CAPE are not clearly
correlated with one set of multispectral TBs.
ƒ This demonstrates great potential for these
dynamical / thermodynmical tags to be
useful if they are used to mine the database.
• Continue to generate more simulations to produce uniform distribution of profiles in North
America
• Determine by thorough statistical analysis of North America CDRD profile database, degree
to which uncertainty in precipitation estimation can be reduced through use of dynamincal /
thermodynamical constraints.
• Ultimate goal is to determine how to optimally apply dynamical / thermodynamical
constraints so as to significantly reduce variance in simulated frequency-dependent TB-RR
relationships used in passive microwave precipitation retrieval.
References
Mugnai, A., D. Casella, S. Dietrich, F. Di Paola, M. Formenton, P. Sanò, E.A. Smith, A.
Mehta, G. J. Tripoli and W.-Y. Leung, 2008: An advanced cloud-radiation database approach
for precipitation retrieval from passive-microwave satellite observations over the
Mediterranean area”. 10th EGU Plinius Conference on Mediterranean Storms.
Acknowledgements
Tb 37GHz
Tb 85GHz
Pete Pokrandt – UW-AOS System Administrator
Funded by NASA Grant Number NNX07AD43G