Presentation is loading. Please wait.

Presentation is loading. Please wait.

Development of Instability Index of GEO-KOMPSAT-2A

Similar presentations


Presentation on theme: "Development of Instability Index of GEO-KOMPSAT-2A"— Presentation transcript:

1 Development of Instability Index of GEO-KOMPSAT-2A
Sung-Rae Chung1, Myoung-Hwan Ahn2, Su-Jeong Lee2 1 KMA/NMSC, 2 Ewha Womans University 2014 Convection Working Group Workshop, 7-11 April 2014, Zagreb, Croatia

2 Meteorological Sensor
Geo-KOMPSAT-2 Program GK-2A for the next generation Meteorological Imager and Space Weather monitoring GK-2B for the Ocean Color and Atmospheric Trace Gas monitoring Meteorological Sensor Space weather Sensor Ground Segment Data Processing System KMA Ocean / Environmental Sensor Geo-KOMPSAT-2B 2012 ~ 2017 (6 years) Geo-KOMPSAT-2A

3 <Basic Spectral Bands>
AMI (Geo-KOMPSAT-2A) AMI (Advanced Meteorological Imager) 16 spectral bands <Basic Spectral Bands> Bands Band Name Center Wavelength Band Width (Max, um) Resolution (km) SNR NEdT(K) (240/300K) Radiometric Accuracy Min(um) Max(um) VNIR 1 VIS0.4 0.431 0.479 0.075 250 5% 2 VIS0.5 0.5025 0.5175 0.0625 3 VIS0.6 0.625 0.66 0.125 0.5 120 4 VIS0.8 0.8495 0.8705 0.0875 210 5 NIR1.3 1.373 1.383 0.03 300 6 NIR1.6 1.601 1.619 MWIR 7 IR3.8 3.74 3.96 3/0.2 1K 8 IR6.3 6.061 6.425 1.038 0.4/0.1 9 IR6.9 6.89 7.01 0.37/0.1 10 IR7.3 7.258 7.433 0.688 0.35/0.12 11 IR8.7 8.44 8.76 0.27/0.1 LWIR 12 IR9.6 9.543 9.717 0.475 0.35/0.15 13 IR10.5 10.25 10.61 0.875 0.4/0.2 14 IR11.2 11.08 11.32 1.0 0.19/0.1 15 IR12.3 12.15 12.45 1.25 0.35/0.2 1.1K 16 IR13.3 13.21 13.39 0.75 0.48/0.3 [Source: Konig, 2002]

4 Stability Indices (e.g. CAPE)
Statistical Retrieval of Instability Index Satellite observed radiances Artificial Neural Network (ANN) setup relation Stability Indices (e.g. CAPE) Good at estimating non-linear relations Once the statistical relations are established, the retrieval can be made computationally very fast (Konig, 2002), often with surprising accuracy (Gardner & Dorling, 1998) Training dataset should be complete (large enough to cover a variety of atmospheric phenomena, seasons, and locations) (Blackwell & Chen, 2009) Source: Konig, 2002 Atmospheric Instability Parameters derived from MSG SEVIRI observations. Koenig and Coning, the MSG GII ANN 단점: requires a large number of individual runs to determine the best solution (http://pages.cs.wisc.edu/~bolo/shipyard/neural/local.html) Both method: only works for clear-sky conditions (no instability info is inferred for cloudy pixels., Koenig 2008) [Source: Konig, 2002]

5 Work Process Flow ANN training Final set of T, q profiles
(radiosonde, satellite,..) Objective selection CAPE CAPE profiles IASI (SEVIRI coverage) CAPE Range TOTAL selected 0 <= CAPE <1000 133,194 1,375 1000 <= CAPE <2000 6,360 2000 <= CAPE 140,929 4,125 MODTRAN run Simulated Radiance (I) ANN training (establish relations) SRF (SEVIRI for AMI) band-averaged Radiance ( 𝐼 ) determine Inverse of Planck eq. Best ANN Parameters In most cases, more than 90% (94.5% this case) of the profiles lead to CAPE under 1000. The number of downloaded profiles : 100,000 or more with big size. But as explained, it is very important to construct a complete set of train-data, so.. So, first calculate CAPE for all these profiles, and then uniformly select data from each CAPE range Theoretical TB Real Application: Measured radiance (TB) CAPE direct estimation from GK2A AMI

6 Total weights ( WIHxWHO )
Combination of weights for the best ANN performance Input layer Hidden layer INPUT Total weights ( WIHxWHO ) circular day 0.09 circular time 0.22 Latitude 0.02 Longitude 0.32 Sat. zenith -0.18 Tb6.2 -0.71 Tb7.3 1.56 Tb8.7 6.79 Tb9.6 4.63 Tb10.8 -0.38 Tb12.0 -7.04 Tb13.4 -4.17 cir_day n 1 cir_time n 2 latitude n 3 longitude n 4 Output layer sat_zenith n 5 TB6.2 n 6 CAPE TB7.3 n 7 TB8.7 n 8 - The analysis of final combination of weights reveals that among 12 input variables, brightness temperatures, particularly Tb8.7 and Tb12.0, are identified as discriminating input measurements, having strong connections with the neurons in the hidden layer - Detailed connections for each variable need further investigation TB9.6 n 9 TB10.8 n 10 TB12.0 n 11 TB13.4 n 12 strongest weight * thickness of the arrows: relative magnitude of connection weights 2nd strongest weight

7 Algorithm Validation Future work ANN training direct estimation
IASI (SEVIRI coverage) ANN training CAPE ANN with best parameter Radiances from SEVIRI 2.5min (super) rapid scan direct estimation CAPE retrieved In most cases, more than 90% (94.5% this case) of the profiles lead to CAPE under 1000. Future work compare Radiosonde CAPE measured Validation

8 CAPE derived using ANN algorithm
(20 June 2013) 09:02:14 11:02:13 13:02:12 15:02:15 17:02:14 20:02:13 Significant features - During the afternoon hours (at around 13:02 UTC) several convective clouds begin to pop up over the regions where the CAPE values are relatively high (marked with arrows and circles at 09:02 UTC image) and developed to the severe convective clouds (at 15:02 UTC) - High CAPE values around the leading edges of clouds induce a further development, while weaker CAPE values around the trailing edges result in weakened convective activities. Significance will be assessed with more case studies and quantitative validation. - No significant convection occurs over high CAPE areas in the morning images (at around 11:02 UTC marked with dashed blue arrows) which requires further investigation. 1,000 2,000 4, (J / kg) 3,000

9 CAPE derived using ANN algorithm
(20 June ~21 UTC) 1,000 2,000 4, (J / kg) 3,000

10 Thank you!


Download ppt "Development of Instability Index of GEO-KOMPSAT-2A"

Similar presentations


Ads by Google