the optimising of regional radiosonde networks
DESCRIPTION
THE OPTIMISING OF REGIONAL RADIOSONDE NETWORKS. Oleg Pokrovsky Main Geophysical Observatory, Karbyshev str.7, St. Petersburg, 194021, Russian Federation. Outlines:. (a) Identify statistically homogeneous areas; - PowerPoint PPT PresentationTRANSCRIPT
THE OPTIMISING OF REGIONAL RADIOSONDE NETWORKS
Oleg Pokrovsky
Main Geophysical Observatory,
Karbyshev str.7, St. Petersburg, 194021, Russian Federation
Outlines:
• (a) Identify statistically homogeneous areas;
• (b) Calculate the statistical weights of the information of each RAOB station ;
• (c) Derive an optimized network configuration for the upper-air stations, including GUAN stations;
• (d) Calculate error fields for main meteorological parameters (Z500, T500, U700, V700, relative air humidity Q850 used in NWP models) related to the optimized network configuration.
Approach
Kalman Filter (1)
wherewhere:
Kalman Filter (2)
I n f o r m a t i o n m o d e l f o r n e t w o r k
ppp xAy )( ( 1 )
x - t r u e f i e l d ; y - v e c t o r o f m e a s u r e m e n t s a t n e t w o r k
)( pA - o b s e r v i n g s y s t e m o p e r a t o r
p
erro rtm ea su rem enco o rd in a tesitethi ii
ppp
_,__
:),....,;,...,(2
2211
111 )}()({ ppT
xx AA - c o v a r i a n c e m a t r i x f o r L S ( 1 )
Information Weights of Sites
Statistical Invariant:
Information weight of i-th site
)()(}{xxp
trFF
missedissitei thipiiip ___);,.,,,..,( 111
)(/)]()([)( xpipi trFFW
Optimization
Criteria function -
))(()( pxx FFF
Optimization:p
pxp F
))}((max{arg
Criteria examples:
1) A:
2) D:
)()( xx trFF
)det()( xxFF
Part 1
Siberian RAOB network of Roshydromet
0 20 40 60 80 100 120 140 160 180
Р и с . . К а р т а сх ем а п о л н о й а эр о л о ги ч еск о й сети Р о сги д р о м ета
д о л го та
40
60
80
шир
ота
20107
22113
22217 22271
22522 22550
2282022845
26063
26258 2629826477
26702 26781
2703727199
274592759527612
2770727730
279442796227995
3400934122 3417234247
3456034731
3485834880
3701837054
2004620069
20274 2029220353
2067420744
20891
21432
2150421647
2182421946 21965 21982
23022
2320523330
23418 2347223552
23804 2388423921
2393323955
2412524266
24343
2450724641 24688
2472624817
24908 24944
24959
25123 25173
2539925400 25428
25551 25563 255942567725703
2591325954
28225 28275
2844528661
2869828722
29231 29263 29282
2957229612
29634 2969829839 29862
30054
3023030309 30372
30521 3055430635 3067330692
30715 30758
30935 30965
3100431088
31168
31300 31329 31369
31510
3170731736
3187331909
31977
32061
32150
32165
32186
32215
32389
32540
32618
35121 36096
A set of RAOB stations presented in WMO list
8 0 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 0 1 6 0 1 7 0 1 8 0
Longitude
Siberian RAOB configuration in 1970-1989
40
50
60
70
80
La
titu
de
Soviet Time
RAOB Problems After 1998
• Statistics: 1999: 4-7 sondes per day
• 2000: 10-12 sondes per day
• 2001: 16-19 sondes per day
• 2002: 22-24 sondes per day
• BUT: main problem is absence of regularity:
• 1999: only 2 stations provided daily regular data
• 2002: only 15 stations provided daily regular data
0 20 40 60 80 100 120 140 160 180
Р и с . . Р асп р ед ел ен и е р ад и озон д о в (ста н ц и и , ч и сл о зап уск ов ) н а А Т Р (О к тя бр ь 1 999 ) в (00Z ) -
Д ол гот а
40
50
60
70
80
Ши
рота
1
8
6
22
3131
29
2830
2731
25
October, 1999
Catastrophic Flood in Siberia
River Lena,
May, 2001
4 0 6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0
C la ssifica tio n o f a tm o sp h er ic c ircu la tio n p a ttern s (U 8 50 & V 8 5 0 ) in A sia : c la ss 2 : 1 99 9 (1 -1 30 d a y s)
0
2 0
4 0
6 0
8 0
Persisted Atmospheric Circulation Regime during February-May, 2001
Source: SATOB data
Z700 field anomaly,March-April, 2001
8 0 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 0 1 6 0 1 7 0 1 8 0 1 9 0
Longitude
RAOB: OPTIMAL /MINIMAL NETW ORK FOR RUSSIAN ASIA (after Pokrovsky, 2003)
40
50
60
70
80
La
titu
de
31168
2143220674
25954
31873
29282
32215
25042
20292
X
3055470414
24125 24266GUAN
GUAN
7 0 8 0 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 0 1 6 0 1 7 0
Longitude
Siberia RAOB Area Regioning
40
50
60
70
80
La
titu
de
РАЙОНИРОВАНИЕ АТР (СРОЧНЫЕ ДАННЫЕ)
6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0
Р и с . . Д о ст а т о ч н а я а эр о л о г и ч еск а я сет ь н а А Т Р : 4 2 ст а н ц и и (и н ф о р м а т и в н о ст ь п р о п о р ц и о н а л ь н а р а зм ер у зв езд о ч ек )
Д о л г о т а
40
50
60
70
80
Ши
рот
а
23955
31168
21432
20674
25594
31873
29282
32215
25042
20292
2412524266
30554
2347223330
23552
36096
28698
23933
28225
24507
24817 2494430054 31004
24343
24959
31510
25913
31369
2540024688
25123 25173
3238932618
25954
2194621824
25563
31977
32150
Sufficient RAOB network
Optimal interpolation H500 RMS error field
6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0
Р и с . . П о л е о ш и б о к в о сст а н о в л ен и я п о л я Н 5 0 0 (м ) п о д а н н ы м д о ст а т о ч н о й а эр о л о ги ч еск о й сет и А Т Р и з 4 2 ст а н ц и й
Д о л гота
40
50
60
70
80
Ши
рота 23955
31168
21432
20674
25594
31873
29282
32215
25042
20292
2412524266
30554
2347223330
23552
36096
28698
23933
28225
24507
2481724944
3005431004
24343
24959
31510
25913
31369
25400
24688
25123 25173
3238932618
25954
2194621824
25563
31977
32150
5
25
45
65
85
105
6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 040
50
60
70
80
4 0 6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0
L o n g i t u d e
S o n d e sta tio n s p ro v id ed p ro file s tw ice p er d a y (0 Z a n d 1 2 Z ) in J a n u a ry 2 0 0 5
4 0
6 0
8 0L
atit
ude
2 2 11 32 2 2 1 7
2 2 5 5 02 3 3 3 0 2 3 4 7 2
2 4 1 2 5
2 4 9 5 92 5 9 1 32 6 0 6 3 2 7 1 9 9
2 8 4 4 52 8 6 9 8
2 9 5 7 22 9 6 3 4
3 0 2 3 0
3 0 7 5 8
3 1 0 0 4
3 1 7 3 6
3 1 9 7 7
3 2 1 5 0
3 2 5 4 03 4 0 0 9
3 4 5 6 03 4 8 5 83 4 8 8 0
6 0 7 0 8 0 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 0 1 6 0 1 7 0 1 8 0 1 9 0
L o n g i t u d e
S ib er ia n R A O B co n fig u ra tio n p ro v id ed ev ery d a y 0 Z a n d 1 2 Z p ro file s in J a n u a ry 2 0 0 7
40
50
60
70
80
Lat
itud
e
2 1 9 4 6
2 3 4 7 2
2 3 9 5 5
2 4 1 2 5
2 4 6 8 8
2 4 9 0 8
2 5 4 0 0
2 5 7 0 3
2 5 9 1 32 8 2 7 5 2 9 2 3 1 2 9 2 6 3 2 9 2 8 2
2 9 5 7 22 9 6 1 22 9 6 3 4 2 9 6 9 82 9 8 3 9 2 9 8 6 2
3 0 0 5 43 0 2 3 0
3 0 3 0 9 3 0 3 7 23 0 5 5 4
3 0 7 1 5 3 0 7 5 83 0 9 3 5 3 0 9 6 5
3 1 0 8 8
3 1 7 3 6
3 1 8 7 33 2 1 5 0
3 2 5 4 0
6 0 7 0 8 0 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 0 1 6 0 1 7 0 1 8 0 1 9 040
50
60
70
80
8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0
Ï î ë å î ø è á î ê â î ññòà í î â ë å í è ÿ ï î ë ÿ Í 50 0 (ì ) ï î ä à í í û ì àýð î ë î ãè ÷åñê î é ñåòè À ÒÐ è ç 3 4 ñòà í ö è é (åæ åä í åâ í î ) â ÿí â àð å -ì àðòå 20 07 ã.
ä î ë ãî òà
40
60
80
øèð
îòà
5
15
25
35
45
55
65
75
85
95
105
80 100 120 140 160 18040
60
80
21946
23472
23955
24125
24688
24908
25400
25703
2591329231 2926329282
295722961229634 29698
29839 29862
3005430230
30309 30372
3055430715 30758
30935 30965
31088
31736
3187332150
32540
Optimal interpolation H500 RMS error field
Responded to Jan-March, 2007, RAOB
Table. Comparison of the optimal and operational RAOB network configurations
in Siberia with account for Z500 objective analysis error (m).
Contribution of measurement data in covariance matrix reduction
RAOB –40 (non-regular, Jan-Mar, 2007)
RAOB-34 (Jan-Mar, 0Z&12Z, 2007)
RAOB-42 (Optimal design)
Mean STD (60-80 N)
58.7 57.6 27.8
Mean STD (40-60 N)
42.3 46.9 42.5
Mean STD 51.0 52.6 34.7
Conclusions (Part 1):
-Number of Siberian RAOB sites was increased during last years
-Most of recovered stations are located in southern part of Siberia close to China border provided by many vertical profiles from Chinese RAOB
-Few stations were added in medium latitude belt and in high latitudes
-Present configuration of Siberian RAOB network does not provide necessary accuracy in analysis of height, temperature and, particularly, wind fields in in high latitudes
Part 2
A CASE STUDY: RA I - AFRICA
RAOB network in RA-I: red-operational (2004); black-nominal in WMO list
-10 0 10 20 30 40 50 60 70
F ig u re 1 . O p era tio n a l in 2 0 0 4 (red sta rs-4 6 ) a n d W M O n o m in a l (b la ck sta rs-2 6 2 ) R A O B sta tio n s in R A I
Longitude
-40
-30
-20
-10
0
10
20
30L
atit
ud
e
Statistical Regionning due to zonal wind U700
-20 -10 0 10 20 30 40 50 60
F ig u re 2 . S ta tis t ica l reg io n in g o f zo n a l w in d fie ld U 7 0 0 in R A 1 (o p era tio n a l R A O B -sta rs) : 1 9 9 0 -1 9 9 9
Longitude
-50
-40
-30
-20
-10
0
10
20
30
Latitude
Information content weights attributed to existed operational sites
-10 0 10 20 30 40 50 60
F ig u re 3 . In fo rm a tio n w e ig h ts o f R A O B sta tio n s in R A 1 fo r U 7 0 0 (c irc le s ize is p ro p o rtio n a l to w e ig h t)
Longitude
-40
-30
-20
-10
0
10
20
30
Latitude
60018
60155
60191 60
525
60549
60571
60630
60656
60680
60760
61024
61052
61291
61442
61687
61901
61902
62306
62337
62378
62403
62414
62423
63450
6374164
500
64650
64910
65503
67083
67197
68174
68240
68263
68328
6844268461
68512
68538 68
588
68816
68842
68906
Relative error (with account for seasonal variability) fields for Z500 objective analysis
-20 -10 0 10 20 30 40 50 60
F ig u re 4 . O b jec tiv e a n a ly s is re la tiv e erro r o f g eo p o ten tia l f ie ld H 5 0 0 (o p era tio n a l 4 6 R A O B )
Longitude
-50
-40
-30
-20
-10
0
10
20
30Latitude
0
0 .1
0.2
0.3
0.4
0.5
0.6
0.7
-20 -10 0 10 20 30 40 50 60-50
-40
-30
-20
-10
0
10
20
30
Relative error (with account for seasonal variability) fields for U700 objective analysis
-20 -10 0 10 20 30 40 50 60
F ig u re 5 . O b jec tiv e a n a ly s is re la tiv e erro r o f zo n a l w in d fie ld U 7 0 0 (o p era tio n a l 4 6 R A O B )
Longitude
-50
-40
-30
-20
-10
0
10
20
30Latitude
0 .1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
-20 -10 0 10 20 30 40 50 60-50
-40
-30
-20
-10
0
10
20
30
Scenario for RA-I RAOB extension with account for maximization of information content: red-new 13 stations;
black-operational network (46 stations)
-10 0 10 20 30 40 50 60
F ig u re 6 . O p era tio n a l 4 6 R A O B (b la ck ) a n d o p tim a l 1 3 (red ) s ta tio n s: U 7 0 0 in fo rm a tio n w e ig h ts - p ro p o r tin a l to c irc le s iz e s
Longitude
-40
-30
-20
-10
0
10
20
30Latitude
60611
61096
6763366422
63832
65167
62760
64459
61043
63962
62840
64390
62650
Relative error (with account for seasonal variability) fields for Z500 objective analysis:
extended network
-20 -10 0 10 20 30 40 50 60
F ig u re 7 . O b jec tiv e a n a ly s is r e la tiv e erro r o f h e ig h t f ie ld H 5 0 0 (o p era tio n a l 4 6 R A O B + 1 3 reco v er in g sta tio n s; b la ck p o in ts-W M O lis t)
Longitude
-50
-40
-30
-20
-10
0
10
20
30
Latitude
0
0.1
0.2
0.3
0.4
0.5
0.6
Relative error (with account for seasonal variability) fields for U700 objective analysis:
extended network
-20 -10 0 10 20 30 40 50 60
F ig u re 8 . O b jec tiv e a n a ly s is r e la tiv e erro r o f zo n a l w in d fie ld U 7 0 0 (o p era tio n a l 4 6 R A O B + 1 3 reco v er in g s ta tio n s; b la ck p o in ts-W M O lis t)
Longitude
-50
-40
-30
-20
-10
0
10
20
30Latitude
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
Minimal GUAN network due to U700
-10 0 10 20 30 40 50 60
F ig u re 1 0 . M in im a l G U A N n etw o rk (1 0 s ta tio n s) in R A -I: s ta r s ize is p ro p o rtio n a l to in fo rm a tio n w e ig h ts to m o n th ly U 7 0 0 fie ld s
Longitude
-30
-20
-10
0
10
20
30
40
Latitude
61442
64754
63894
67965
68816
64500
66422
62641
60390
64459
Relative error (with account for multi-year variability) monthly fields attributed to GUAN
for U700
-20 -10 0 10 20 30 40 50 60
F ig u re 1 1 . O b jec tiv e a n a ly s is re la tiv e erro r o f m o n th ly w in d fie ld U 7 0 0 (G U A N m in im a l n e tw o rk : 1 0 s ite s)
Longitude
-50
-40
-30
-20
-10
0
10
20
30
Latitude
61442
64754
63894
67965
68816
64500
66422
62641
60390
64459
0.05
0.35
0.65
0.95
Conclusions (Part 2)
-Missing data areas with respect to operational RAOB station list for RA-I are very significant. Only 46 from nominal 262 sites carried out measurements in January-April, 2004.
-Error fields corresponding to major meteorological variables reveal many gap regions, where the relative errors of meteorological field representation reach 0.7-0.8 levels.
-Search algorithm allows us to develop a scenario for existed operational RAOB network extension from 46 to 59 stations by recover measurements at 13 stations, which provide a substantial improvement of error fields for all meteorological variables in missing data areas
-Existing GUAN network has some gaps in Central Africa, which are a reason of anomaly in objective analysis error fields. An alternative set of ten GUAN sites provides more uniform information coverage of Africa with respect to monthly fields.