Comparison of Liquid Water Path from AMSU (NN) and AVHRR (KLAROS) - - PowerPoint PPT Presentation

▶
comparison of liquid water path from amsu nn and avhrr
SMART_READER_LITE
LIVE PREVIEW

Comparison of Liquid Water Path from AMSU (NN) and AVHRR (KLAROS) - - PowerPoint PPT Presentation

Feb 23, 2024 613 likes •759 views

Comparison of Liquid Water Path from AMSU (NN) and AVHRR (KLAROS) Heike Hauschildt Institute of Marine Research Kiel CLIWA-NET Meeting, Madrid, 16.-19 . December 2002 WP 3200: Description of work LWP algorithm for AMSU LWP retrieval over

slide-1
SLIDE 1

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

Comparison of Liquid Water Path from AMSU (NN) and AVHRR (KLAROS)

Heike Hauschildt Institute of Marine Research Kiel

slide-2
SLIDE 2

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

WP 3200: Description of work

LWP algorithm for AMSU LWP retrieval over sea from AMSU data LWP retrieval over sea and land from AVHRR data Synergetic algorithm for AVHRR and ground based

  • bservations over land

Synergetic algorithm for AVHRR and AMSU over sea areas Combining results over sea and land to get LWP fields for the Baltic modelling area and the Netherlands Retrieval of an optimized algorithm from combined ground based and satellite measurements during BBC and reprocessing EOP1+2.

slide-3
SLIDE 3

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

KLAROS Data on the web

CNN 2 : Overpasses near noon

NOAA-14:

April North: ok South: missing May North: missing South: missing

NOAA-16:

April North: ok South: ok May North: ok South: ok

BBC : Overpasses near noon

NOAA-14 partly NOAA-16 ok

slide-4
SLIDE 4

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

Radiosonde dataset

Radiosonde ascents Input data

Pressure < 300 hPa 5 hPa level Synop - data Offshore winds

N = 7388 (< 500g/m2)

slide-5
SLIDE 5

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

NN Training Data

Ships plus coastal stations N=7388

WV TB 23.8 GHz LWP TB 31.4 GHz TB 50.3 GHz TB 89.0 GHz

< 10 g/m2

slide-6
SLIDE 6

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

Neural Network

Ships + Coast N=7388

slide-7
SLIDE 7

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

KLAROS LWP

14.04.2001

slide-8
SLIDE 8

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

AVHRR - AMSU

14.04.2001 AVHRR AMSU

slide-9
SLIDE 9

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

Neural Network

14.04.2001

slide-10
SLIDE 10

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

Neural Network

slide-11
SLIDE 11

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

Temperature vs Viewing Angle

Red: RS- data Blue: 70S - 70N Light blue: N. Atlantic Black: 20N- 70N, 20W-20E

slide-12
SLIDE 12

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

MWMOD distribution

slide-13
SLIDE 13

CLIWA-NET Meeting, Madrid, 16.-19 . December 2002

Conclusion

Cloud fields can be reproduced by NN- retrieval

Overestimation for high LWP (> 300 g/m2) Underestimation for low LWP (< 200 g/m2)

Sub-FOV variability complicates the LWP retrieval Validation of LWP derived from AVHRR by the use of AMSU retrievals does not work