88
Ghapter 5: Interpreting High-Resolution Proxy Glimate Data
Figure 5.2: Examples of different tree-ring response functions: one based
on simple correlation coefflcients of tree growth with monthly mean temperatures, the other on principal components regression using the same temperature data as predictors. The growth variables in each of the two years
immediately preceding the growing season are included among the predictors
to quantify the persistence in the tree-ring timeseries: The autocorrelations
with lags one and two years are labelIed Al and A2.
The sign and magnitude of the coefflcients provide a simple picture of the
tree-growth/climate relationships. Galculating response functions for different periods gives an indication of the temporal stability of the response
pattern.
The examples shown are for a site in the northern Ural Mountains, Siberia.
Significant (p = 0.05) coefflcients are indicated by the dashed lines (simple correlations) and dots (PG regression). The summer response in ring
width is predominantly in June and July. For density it is over a longer
May-September season.
u
. .
SIMPLE CORRElATIONS
Maximum UI.wood D~.lIy
pe REGRESSION COEFFICIENTS
...
I ~Ioq"'or--=----& ..... ....,.Lf 1882·11135
..... OHOJ' .. AIM~JA.O A'''' ....... o .. o ... ' ..... 'JA.O .'AI
...
..
...... O.O"' .. AIIJJ"'.O .,AI
··'ONOJ '''''''IJ''IO .'AI
I
'. 1882·11132
u , - - - - - - - - - ,
\ 11138·111&0
..
""'0",0"" , 1IIj."J J" '0 A'", ... ·.0.0 .. ' ............. 0 A'A.