1. Quantitative precipitation estimation for typhoon Morakot over oceans
Nan-Ching Yeh1, Wann-Jin Chen2, Jian-Liang Wang3, Jen-Chi Hu4, Gin-Rong Liu5, Ming-Da Tsai3
1 School of defense science, Chung Cheng Institute of Technology, National Defense University
2 Ta Hwa Institute of Technology
3 Department of Environmental Information and Engineering, Chung Cheng Institute of Technology, National Defense University
4 Armed Weather Center, Ministry of National Defense
5 Center for Space and Remote Sensing Research , National Central University
Purpose
Using Bayesian approach to estimate the rainfall of typhoon over ocean. In order to reduce the damage to livelihood and economy caused by heavy rainfall associated with severe weather systems such as typhoon Morakot (2009), the accurate rainfall retrieval of typhoon is an important mission for both weather operators and researchers.
Case Morakot at 0411UTC
Flowchart of Bayesian method and Physical Theorem
WRF model and GCE cloud model data
Attenuation Index
Petty(1994)
v：vertical polarization
h：horizontal polarization
v,o& h,o：brightness temperature under cloud free 1
LRTM model (radiative transfer model)
to simulate the brightness temperature (Tb)
and calculate P value (P10,P19,P37) 2
TMI_PR match up data
to fit prior probability
distribution π(R)
Chiu and Petty(2006)
Establish conditional probability
density function f (P|R) 3 3
Calculate Posterior probability
distribution f (R|P) and
establish lookup table of various estimation
Bayesian Approach 4
Use Tb observed from TMI
to calculate P10, P19, P37
:00 GB-radar-reflectivity
:11 PR-RR (observation)
Using average (AVE)
estimator to calculate rain rate
Validation with
Precipitation Radar 5
Validation
:11 TMI-GCE-RR (retrieval)
:11 TMI-WRF-RR (retrieval)
correlation coefficient = 0.649
samples=2860
correlation coefficient = 0.631
samples=2860
correlation coefficient = 0.65
samples=752
correlation coefficient = 0.63
samples=752
Summary and Future Work
1. PR rainfall products are used for validations in our study and the results show that our Bayesian retrievals and PR-RR have similar rain pattern in horizontal distribution and that the quantitative results agree well with PR products (their correlation coefficient > 0.6).
2.In the future, more typhoon simulations will be added to our database to improve the accuracy of satellite rainfall retrievals. The RTTOV Radiative Transfer model will be used in our study.