Download - Homogeneity Analysis on Surrogate Temperature Data, Network No.1 Palma 25-27 January 2010
Homogeneity Analysis onSurrogate Temperature Data, Network No.1
Palma 25-27 January 2010
Norwegian Meteorological Institute met.no
Work tool: ProClimDB by Petr Stepanek
• Procedure– Network analysis / Length of data period /
Distance between stations / Correlations between stations
– Quality control / Finding reference stations / Detection of outliers / Correction of original data series
– Homogeneity analysis / Finding reference stations / Using SNHT method (Automatic AnClim) / Breaks and adjustments
– Adjustments of data series
Norwegian Meteorological Institute met.no
Network No.1 – analysis 1
Length of data series
Station series
0
1
2
3
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Length of data series (5-year intervals)
Nu
mb
er
of
sta
tio
ns
Distance between stations
Distance between stations
0
20
40
60
80
100
120
140
160
180
200
110130120150130100120150150060110100110050060150100130120150100060050150060100120150160060120100150060120100
Station id
Dis
tan
ce
(k
m)
Norwegian Meteorological Institute met.no
Network No.2 – analysis 2
Correlations between stations
Correlation between stations of Network no.1, within a radius of 200 km
0.00.10.20.30.40.50.60.70.80.91.0
150 070 160 110 120 100 060 130 110 150 070 100 160 120 130 100 110 150 130 120 160 110 150 130 160 120 150 160 120 130 150 130 160 150 160 160
050 060 070 100 110 120 130 150
Top: NeighboursBottom: Candidate station
Co
rrel
atio
n
Norwegian Meteorological Institute met.no
Settings
ProClimDB
Norwegian Meteorological Institute met.no
Quality control – 1
• Criterions for making reference series– Maximum 6 stations, – Within distance of 200 km– Maximum altitude difference is 800 m– For consecutive 5-year periods with
common data and confidence limit = 0.95– Observation values are standardized to
the altitude of candidate by regression– Outliers check on the difference between
observation and reference value (0.99)
Norwegian Meteorological Institute met.no
Quality control - 2
Loxodromes of each candidate in different 5-year periods
0.00
100.00
200.00
300.00
400.00
500.00
600.00
700.00
050 060 070 100 110 120 130 150 160
Station id
Dis
tan
ce
(k
m)
Norwegian Meteorological Institute met.no
Homogeneity analysis 1
• Criterions for making reference series for detecting breaks– Maximum 5 stations– Correlation with neigbours above 80%– Maximum altitude difference is 800 m– For 40-year periods (10 years overlap)
with common data– Distribution of neighbour values is
standardized to the AVG and STD of candidate, for each month
– Candidate and reference series are used in the homogenization analysis (SNHT method, automatically analysed in AnClim – results transferred to ProClimDB)
Norwegian Meteorological Institute met.no
Homogeneity analysis 2
• Breaks– 38 breaks in Network No.1– For station 050 four breaks:
• 1941• 1951• 1975• 1984
– No iterations
Norwegian Meteorological Institute met.no
Homogeneity analysis 3
• Adjustments (station 050 – no
iterations)– 1.procedure
• Ref.series for breaks, weigthing on distance (1/d) (power of weights = 0.5).
• Maximum adjustment:1.3°C• Number of adjustments: 33 for 12 months
– 2.procedure• Ref.series based on up to 10 years around
break, minimum 4 years from end (or from another break), 5 nearest stations with distance below 200 km, weigthing on distance (power of weights = 0.5).
• Maximum adjustment: 1.0°C• Number of adjustments: 28 for 12 months
Norwegian Meteorological Institute met.no
Homogeneity analysis 4 – station 050 – 1(ref.series as for detect.breaks)
JAN FEB MAR APR
MAY JUN JUL AUG
SEP OCT NOV DECNotice scale for
JUN-SEP
Norwegian Meteorological Institute met.no
Homogeneity analysis 5 – station 050 -2(ref.series for making adjustm. – for this presentation only)
JAN FEB MAR APR
MAY JUN JUL AUG
SEPNotice scale for
AUG
OCT NOV DEC
1975
1985
1951
1941