Presentation is loading. Please wait.

Presentation is loading. Please wait.

Cloudy GOES-Imager assimilation with hydrometeor control variable Byoung-Joo JUNG NCAR/MMM Student presentation at JCSDA Summer Colloquium on Satellite.

Similar presentations


Presentation on theme: "Cloudy GOES-Imager assimilation with hydrometeor control variable Byoung-Joo JUNG NCAR/MMM Student presentation at JCSDA Summer Colloquium on Satellite."— Presentation transcript:

1 Cloudy GOES-Imager assimilation with hydrometeor control variable Byoung-Joo JUNG NCAR/MMM Student presentation at JCSDA Summer Colloquium on Satellite Data Assimilation, July 27 – August 7 2015

2 Background Ph.D. in Atmospheric Sciences, Yonsei University, Korea Adjoint sensitivity analysis Singular Vector for adaptive strategy in T-PARC field campaign (Tropical cyclone) Evaluation of observation impact with WRF 3DVAR, EnSRF, and FSO Researcher at Korea Institute of Atmospheric Prediction Systems TLM/ADM coding for global spectral-element dycore Representer-based variational system Vertical localization of radiance obs. in LETKF system

3 Channel 4 Channel 6 ObservationBackgroundInnovation (OmB)

4 GSI DA system with WRF background on CONUS domain CRTM with qc, qi, qr, qs, qg from model forecast 2 hydrometeors (qc, qi) added as control variable Static background error statistics for hydrometeors (from GEN_BE v2.0) Conventional obs + GOES-13/15 Imager channel 4 (10.7 um) & channel 6 (13.3 um) Relax first-guess check criteria Skip cloud screening in QC Non-Gaussian distribution for obs. error Multiple outer loop (re-linearization)  nonlinearity Approach *High spatial/temporal resolution

5 - Less weight for large innovation - Enlarge the obs. Error - Affect the conditioning itself Huber norm Gaussian Huber Iteratively Re-weighted Least Square (IRLS)

6 Channel 4 Channel 6 Error normalized innov. Huber via IRLS Gaussian OmB Histogram

7 Innovation (OmB)OmA(B/A) vs. O Cost Function Gradient Norm

8 Bias in hydrometeors is more pronounced. ₋Require much re-linearization for the cases O/B~[cloudy/clear or clear/cloudy] ₋Feature calibration and alignment (FCA) can help to reduce displacement error. How to determine the Huber parameter? (or other error model?) ₋Critical for balanced conditioning Hybrid approach can help the multi-variate relation between meteorological fields & cloud variables. ₋How effective will be the alpha-CV of ensembles on this? What can be the proper predictor for “cloudy” bias correction? Proper control variable for hydrometeors Accurate surface emissivity, cloud properties for various instruments in CRTM Inter-channel correlation can be affected by cloud. How to sustain the hydrometeor increments in the forecast? How to validate the impact of cloudy radiance assimilation? Many many questions


Download ppt "Cloudy GOES-Imager assimilation with hydrometeor control variable Byoung-Joo JUNG NCAR/MMM Student presentation at JCSDA Summer Colloquium on Satellite."

Similar presentations


Ads by Google