Download presentation
Presentation is loading. Please wait.
Published byHannah Carter Modified over 8 years ago
1
C-band scatterometers ERS-1/2 ASCAT Post-EPS Ad.Stoffelen@KNMI.nl R&D, Weather Satellite Winds Group
2
2 Motivation of C-band scatterometers ASCAT NRT/QRT operational use is consolidated and widespread (Numerical Weather Prediction, Nowcasting, … ASCAT contributes to the Global Climate Observing System by providing main Essential Climate Variables -Surface wind vectors over ocean -Soil moisture over land ASCAT series complements and extends the scatterometer C-band record started in 1991 by the ESA ERS1/2 missions
3
3 Overview Swath geometry, orbit Accuracy, resolution Limitations NWP data assimilation Impact Ongoing developments
4
4 ASCAT scatterometer Three ASCAT arms Fan beams
5
5 www.knmi.nl/scatterometer ASCAT ascending tracks 22 hrs
6
6 550 km 45 Left swath 6 ASCAT observation geometry 550 km Right swath 90 135 Real aperture radar, 5.255 GHz (C-band), VV polarisation All weather measuring capability Measuring geometry: 3 fan-beam antennas, double swath, incidence angles between 25 and 65 deg Measurement: normalised radar cross-section (NRCS, backscatter, 0) Swath gridded into nodes (25 km and 12.5 km spacing), one triplet of averaged backscatter measurements per node
7
7 550 km Left swath 7 ASCAT observation geometry 550 km Right swath Each swath is divided into 21 Wind Vector Cells (WVCs) for the 25 km product For the 12.5-km product 43 WVCs exist on each side
8
WVC x x x x x x x x 1 8 6 5 3 2 4 7 FOV Wind Vector Cell Fore beam Many radar responses of one beam cover one WVC These are averaged to compute sigma naught The SD is computed and provides a 0 noise estimate (Kp)
9
WVC x x x x x xx x 1 8 6 53 2 4 7 FOV Wind Vector Cell Mid beam Same for mid beam However, the WVC samples of fore and mid beam do not line up exactly The collection of fore beam FOVs see a different mean wind than the collection of mid beam FOVs
10
WVC x x x x x x x x 1 8 6 5 3 2 4 7 Wind Vector Cell Aft beam u FOV = u WVC + u v FOV = v WVC + v Error: u, v N(0,0.75) The collection of FOVs for each beam see a different mean wind The WVC wind is a combination of the backscatter measurements of all beams GMF relates WVC- mean backscatter to WVC-mean wind
11
11 Geophysical noise Geophysical noise is small Scatterometer winds can be interpreted as WVC- mean winds
12
12 Spatial representation We evaluate area-mean (WVC) winds in the empirical GMFs 25-km areal winds are less extreme than 10-minute sustained in situ winds (e.g., from buoys) So, extreme buoy winds should be higher than extreme scatterometer winds Extreme NWP winds are again lower due to lacking resolution (over sea)
13
13 WVC size WVC size is 50/25 km Extreme winds are smeared out How to translate scatterometer winds to hurricane categories ? Same guidance in tropics and extratropics ? Typical factor of 1.5-2.0 between 10-min winds and scatterometer winds WVC size
14
14 Standards in TC classification Hurricane standards are based on 10-min mean winds, not on 25-km mean winds ! Need for unification in names ?!
15
15 Operational 12.5-km product Extreme winds need calibration Ike: highest ASCAT speed ever (75 knots) and we were right in there
16
16 Verification against ECMWF Unprecedented wind statistics
17
17 Buoy Verification
18
18 Buoy verification January 2009 www.knmi.nl/scatterometer
19
19 Buoy verification ASCAT @ 12.5 compares best to buoys SeaWinds @25 is slightly noisier than ASCAT @12.5 and @25 October 2008 ASCAT 12.5ASCAT 25 SeaWinds 25 KNMI SeaWinds 25 USA u [m/s] v [m/s] u [m/s] v [m/s] u [m/s] v [m/s] u [m/s] v [m/s] 1.671.651.701.641.761.832.191.99
20
20 Triple collocation analysis of buoy, scatterometer & NWP Buoy and scatterometer observations more accurate than ECMWF Accuray on WVC scale (50 km) Vector RMS error [m/s] Tropical TAO/PIRATA Extratropical NDBC/MEDS/UKMO Buoy 1.5 Scatterometer 1.21.6 ECMWF model 2.02.1
21
21 AWDP@12.5 k -5/3 100 km ASCAT contains small scales down to 25 km No noise floor More small scales and less noisy than SeaWinds ECMWF lacks small scales Close to 3D turbulent: k -1.9 3D turbulence k -3
22
22 Comparison of SeaWinds with ECMWF and buoys SDP at 25 kmSDP at 100 km u (m/s) v (m/s) u (m/s) v (m/s) ECMWF1.871.831.571.48 Buoys1.791.882.172.06 All data from January 2008 When going to coarser resolution Agreement with model increases Agreement with buoys decreases In line with spectral analysis
23
23 Level 1 processing Standard level 1 processing involves a spatial Hamming filter to obtain σ 0 values → smearing out of details EUMETSAT provided also test set with box filter applied One expects more detail in box filtered set Perhaps more noise too !
24
24 EUMETSAT test set No sign of noise floor More detail in ASCAT (blue) than ECMWF model (red) More detail in ASCAT@ 12.5 km than in ASCAT @ 25 km More detail with box averaging as expected Yet smaller boxes ? Spectral analysis
25
25 Operational 12.5-km product Convective systems SST Currents
26
26 ASCAT 25 km 12.5 km
27
27 ASCAT winds Verify very well with NWP model Verify very well with buoys Show spectra close to that expected for 3D turbulence for scales < 500 km Spatial plots show small-scale features in line with these three features What about the GMF at extreme winds > 30 m/s ? Systematic effects across swath ? Ambiguity removal at small scales ? Rain and sea ice contamination ? Coastal winds ?
28
28 Generic Scatterometer Data Processor Observations Inversion Ambiguity Removal Wind Field INPUTOUTPUT Observations Inversion Ambiguity Removal Quality Control Quality Control Wind Field INPUTOUTPUT Quality Monitor AWDP: ASCAT Wind Data Processor, also ERS QDP: QuikScat Data Processor IWDP: ISRO OceanSat-2 Wind Data Processor
29
29 Geophysical Model Function CMOD5.N relates the WVC-mean equivalent neutral wind vector to C- band backscatter The (fore,mid,aft) backscatter triplet of each WVC can be plotted in a 3D space As the swath encounters varying wind, the distrbution of triplets will be along a conical surface, which is described well by CMOD5.N speed Wind direction
30
30 Determines noise analysis, QC GMF quality GMF inversion procedure (wind vector retrieval) Cone analyses o Upwind o Downwind speed
31
31 Wind Retrieval by Bayes’ theorem 0 noise is uniform in 3D measurement space (~0.2 m/s) For a given measured backscatter triplet, Bayes’ helps us to find the most probable points on the cone surface, which are tagged with a wind vector solution Large distances from the cone surface are unlikely due to wind (QC)
32
32 Discrimination of land, water and ice Detached sea ice field of 400kmx400km at South Pole
33
33 New C band ice model Bayesian approach: ice line lies right under the C-band GMF cone Sometimes hard to classify measurements! ASCAT ice model presently being tuned Apr 2008 Oct 2008
34
34 April 2008 October 2008
35
35 Scatterometer data assimilation J b balanced (e.g., geostrophy) Scatterometer wind cost Jb is a penalty term penalizing differences with a priori NWP background field (first guess) Jb differences should be spatially balanced according to our knowledge of the NWP model errros Jo is a penalty term penalizing differences of the analysis control variables with the observations Scatterometer observations are not spatially correlated So, Jb determines the spatial consistency of the analysis
36
36 Scatterometer Observation Cost: Jo Posteriori Wind Probability given a set of measurements Wind domain uncertainty at Jb scale u, v ~ 1.0 m/s Measurement space noise D ~ 5% (0.2 m/s) s = GMF(v s )Geophysical solution manifold ERS/ASCAT: Manifold in 3D measurement space (CMOD5.N) SeaWinds/NSCAT: Manifold in 4D measurement space
37
37 Wind rather than 0 assimilation Main uncertainty is in the wind domain y: 0 x: wind
38
38 Assimilate ambiguities J b balance Scatterometer wind cost i ambiguous wind vector solutions provided by wind retrieval procedure (Stoffelen and Anderson, 1998) Use probability
39
39 Spatial filter: Mass conservation Continuity equation 0 U = 0 Vertical motions < horizontal motion Little divergence (D) Mostly rotation (R) Tropics more balanced R/D Meteorological balance (2/3/4D-VAR)
40
40 Scatterometer Verbeterde 5-daagse voorspelling van tropische cyclonen in ECMWF 4D-VAR Isaksen & Stoffelen, 2000 Rita Geen ERS Scatterometer Wel ERS
41
41 ASCAT advantage for tropical storms ASCAT has smaller rain effect; splash remains Japan Meteorological Agency
42
42 Gebruik van scatterometers Assimilation ASCAT winds ECMWF from 12/6/’07 Beneficial for U10 analysis Operational okt/nov 2007 (added to QuikScat&ERS) Hans Hersbach & Saleh Abdalla, ECMWF ECMWF analysis vs ENVISAT altimeter wind
43
43 Underpredicted surge Delfzijl 31/10/’6 18Z1/11/’06 4Z
44
44 NWP Impact @ 100 km Storm near HIRLAM misses wave; SeaWinds should be beneficial! 29 10 2002
45
45 ERS-2 scatterometer wave train; missed by HiRLAM NWP models miss wave; Next day forecast bust
46
46 Missed wave train in QuikScat
47
47 ASCAT scatterometer R&D - Higher resolution - Closer to the coast - Wind intercalibration
48
48 Conclusions ASCAT on board MetOp provides accurate daily global ocean surface winds at high spatial resolution NWP models lack such high resolution MetOp-B due for launch in 2012 probably providing a tandem ASCAT Further information: www.nwpsaf.orgwww.nwpsaf.orgscat@knmi.nl www.osi-saf.org www.knmi.nl/scatterometer
49
49
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.