WG2 MEETING 2008 AUGSBURG WG2 MEETING 2008 AUGSBURG DOMINGO F. RASILLA ÁLVAREZ Dpto. de Geografía, Urbanismo y OT Universidad de Cantabria Spain
Jan 07, 2016
WG2 MEETING 2008 WG2 MEETING 2008 AUGSBURGAUGSBURG
DO
MIN
GO
F.
RA
SIL
LA
ÁL
VA
RE
Z
Dp
to.
de
Geo
gra
fía,
Urb
an
ism
o y
OT
Un
iver
sid
ad
de
Can
tab
ria
Sp
ain
PREPROCESSING
QUESTION 1: VARIABLES
SLP, Z500, Z850
WINDOW SIZE
DENSITY OF GRID POINTS
¿Weighting?
¿Re-gridding?
S MODE
STANDARIZATION OF GRID POINTS
1014
1016
1018
1020
1022
1 31 61 91 121 151 181 211 241 271 301 331 361
Julian day
hP
a
Regional air pollution analysis
Multipurpose climate analysis
SPATIAL PATTERNS OF VARIABILITY
NUMBER PCs
Scree test-LEV
Rule N
North et al
ROTATIONVARIMAX
PRINCIPAL PRINCIPAL COMPONENT COMPONENT
ANALYSISANALYSIS
Cluster
Analysis
QUESTION 8: SIMILARITY MATRIX
EUCLIDEAN DISTANCE
CLUSTER ALGORITHM
Hierarchical (Ward)
Agglomerative
(K-means)
NUMBER OF CLUSTERS
OUTCOMEFollowing
slides
TRACK OF THE LOW !!
PETISCO’S PROCEDURE
HOW TO SELECT THE NUMBER OF CLUSTERS
• Hybrid method (Rasilla 2000)– Predefined number of groups (manual or
subjective)• Spanish climatology tradition on Syn Class
– Seed days selected through manual classification of a subset (10 years!!) + DA (statistical confirmation of reliability of the classification)
– K-means of the complete dataset (1946-1998)– Too much time consuming!!!
HOW TO SELECT THE NUMBER OF CLUSTERS
ftp://ftp.sas.com/techsup/download/sample/samp_lib/
• Pure statistical method (Rasilla 2006)– Several K-means
solutions with algorithm (reduces the influence of the initial partition)
• number of groups: subjective (analysis of climate fields) + mathematics (Pseudo-t CCC…)
– Disc Ann (validation CA classification)
• Cross-validation (bootstraping 2 halves)
• External validation:– Climate fields– Other classifications– ¿Common sense?
HOW TO SELECT THE NUMBER OF CLUSTERS
Sea level pressure fields Daily precipitation fields
An example of a SPSS syntax file developing
the procedure
QUESTION 2. Is a specific program/software available for the application and use of the method, or a program must be developed by the user? (addition: if not availabe, can we
produce the code to be published ("open source") or is it too complicated or are there not enough details known about the
technique?) SPSS with syntax file, SAS.
QUESTION 3. Must the researcher have a computer programming background for the application of the method
(e.g. in order to develop software)? NO
QUESTION 4. How much time and computer power is roughly required for the computations? NO MORE THAN 5´
Related papers
• Yarnal B (1993): Synoptic climatology in environmental analysis. Belhaven Press, London
• Comrie AC (1996): An all-season synoptic climatology of air pollution in the U.S.-Mexico border region. Professional Geographer 48(3), 237-251.
• Ekström M, Jönsson P & Bärring L (2002): Synoptic pressure patterns associated with major wind erosion events in southern Sweden (1973–1991). Climate Research, 23: 51–66.