Combining Predictors and Predictands in CPT Simon J

Combining Predictors and Predictands in CPT Simon J. Mason simon@iri. columbia. edu International Research Institute for Climate and Society The Earth Institute of Columbia University CIMH Workshop on the Climate Predictability Tool Bridgetown, Barbados, July / August 2013

CPT Formats Predictors and / or predictands can be combined in version CPT 10+, but the new CPT formats must be used. CPT can read three different types of data: 1. Index data : indices have no single specific geographical location, e. g. the SOI; 2. Station data : stations have specific geographical locations, but are not normally distributed in a regular pattern 3. Gridded data : the latitudes and longitudes do not have to be evenly spaced; the grid points can be representative of area-averages or be point specific.

Index Data xmlns: cpt=http: //iri. columbia. edu/CPT/v 10/ cpt: field=ENSO, cpt: nrow=9, cpt: ncol=4, cpt: col=index, cpt: row=T NINO 12 NINO 3. 4 NINO 4 1991 -01 -7. 04 E-02 0. 2694498 0. 5850401 0. 9240733 1992 -01 0. 9675602 1. 567671 1. 78421 0. 9166834 1993 -01 0. 2538059 0. 142204 0. 3976894 0. 8017575 1995 -01 1. 319356 0. 9692747 1. 051067 1. 088881 1996 -01 -0. 169409 -0. 5436825 -0. 7101253 -0. 1563722 1997 -01 -0. 2674479 -0. 6353493 -0. 325502 0. 3376727 1998 -01 3. 716633 3. 442826 2. 581284 1. 036142 1999 -01 -0. 3437423 -0. 8765103 -1. 365756 -1. 278136 2000 -01 -0. 4427198 -1. 439994 -1. 668904 -0. 9916633

Index Data xmlns: cpt=http: //iri. columbia. edu/CPT/v 10/ …, cpt: nrow=10, cpt: ncol=4, cpt: missing=-999 NINO 12 NINO 3. 4 1991 -01 -7. 04 E-02 0. 2694498 0. 5850401 1992 -01 0. 9675602 1. 567671 1. 78421 1993 -01 0. 2538059 0. 142204 0. 3976894 1994 -01 -999 1995 -01 1. 319356 0. 9692747 1. 051067 1996 -01 -0. 169409 -0. 5436825 -0. 7101253 1997 -01 -0. 2674479 -0. 6353493 -0. 325502 1998 -01 3. 716633 3. 442826 2. 581284 1999 -01 -0. 3437423 -0. 8765103 -1. 365756 2000 -01 -0. 4427198 -1. 439994 -1. 668904 NINO 4 0. 9240733 0. 9166834 0. 8017575 -999 1. 088881 -0. 1563722 0. 3376727 1. 036142 -1. 278136 -0. 9916633

Date Formats YYYY 2012 YYYY-MM 2012 -05 YYYY-MM-DD 2012 -05 -10 YYYY-MM-DDTHH: MM 2012 -05 -10 T 09: 00 Start-date/End-date YYYY/YYYY 2011/2020 YYYY-MM/YYYY-MM 2011 -08/2012 -07 YYYY-MM/MM 2012 -04/06 YYYY-MM-DD/DD 2012 -05 -06/12 YYYY-MM-DD/MM-DD 2012 -04 -30/05 -12 YYYY-MM-DD/YYYY-MM-DD 2011 -05 -11/2012 -05 -10

Station Data xmlns: cpt=http: //iri. columbia. edu/CPT/v 10/ cpt: field=prcp, cpt: nrow=8, cpt: ncol=4, cpt: col=station, cpt: row=T Station_A Station_B Station_C Station_D cpt: X -75. 80000 -75. 50000 -75. 60000 -76. 00000 cpt: Y 39. 40000 38. 70000 38. 20000 38. 60000 1981 -01/03 1. 78333 2. 09000 1. 97667 2. 00000 1982 -01/03 3. 34000 3. 59667 3. 60333 3. 55333 1983 -01/03 4. 52667 4. 05333 4. 62000 3. 58000 1984 -01/03 3. 89667 4. 84000 5. 36667 4. 53000 1985 -01/03 2. 04667 2. 68667 3. 07000 2. 66333 1986 -01/03 3. 16000 2. 88667 2. 45333 3. 19333 1987 -01/03 2. 68000 3. 91667 4. 78333 3. 77667 1988 -01/03 2. 92333 3. 59000 3. 72667 3. 61000

Gridded Data xmlns: cpt=http: //iri. columbia. edu/CPT/v 10/ cpt: nfields=1 cpt: field=ssta, cpt: units=C, cpt: T=1979 -01, cpt: nrow=3, cpt: ncol=4, cpt: row=Y, cpt: col=X, cpt: missing=-9999 -18. 75 -16. 25 -13. 75 -11. 25 -58. 75 -2. 40977 -1. 40228 -1. 01278 -1. 03197 -61. 25 -3. 65581 -3. 18779 -2. 28418 -1. 45435 -63. 75 -3. 54459 -3. 04635 -2. 07969 -1. 32451 cpt: T=1980 -01 -18. 75 -16. 25 -13. 75 -11. 25 -58. 75 -2. 22299 -0. 68736 0. 27909 -0. 42638 -61. 25 -1. 13313 -0. 56869 -1. 02204 -1. 09111 -63. 75 -1. 40550 -1. 10838 -0. 37565 0. 88614

Gridded Data xmlns: cpt=http: //iri. columbia. edu/CPT/v 10/ cpt: nfields=2 cpt: field=ssta, cpt: units=C, cpt: T=1979 -01, cpt: nrow=3, cpt: ncol=4, cpt: row=Y, cpt: col=X, cpt: missing=-9999 -18. 75 -16. 25 -13. 75 -11. 25 -58. 75 -2. 40977 -1. 40228 -1. 01278 -1. 03197 -61. 25 -3. 65581 -3. 18779 -2. 28418 -1. 45435 -63. 75 -3. 54459 -3. 04635 -2. 07969 -1. 32451 cpt: field=mslp, cpt: units=mb, cpt: T=1979 -02, cpt: nrow=2, cpt: ncol=2, cpt: row=Y, cpt: col=X, cpt: missing=-10 -70. 0 -65. 0 1001. 2 1000. 6 20. 0 1002. 8 1003. 2

Exercises • Using the ERSST 3 data as predictors combine the data for May and June to make predictions for the Caribbean. Save the X input files containing data for only the domain of interest. Combine these two X files, and make a new set of predictions for the Caribbean. • Similarly, try combining the ECHAM GCM predictors with the ERSST 3 data for June. How does the forecast compare with that for the SSTs and the GCM output as predictors separately? • Combine the ECHAM GCM data with CFS 2 data. Using CCA, does there appear to be a preference for using either of the two models (hint – look at the CCA mode maps, and the climatology correlations)?