How Well Understood Are the Processes that Create Dendroclimatic Records? A Mechanistic Model of the Climatic Control on Conifer Tree-Ring Growth Dynamics - Dendroclimatology - page 62

network for the reconstruction of large-scale features of paleoclimatic fields, but the
comparison suffers from differences in the meteorological and dendrochronological
observing networks (Anchukaitis et al. 2006 ; Evans et al. 2006 ) , which make direct
comparisons impossible. We stress that implicit in our interpretation of the model
results presented here is the following assumption: If we can establish that the sim-
ulations are consistent with the corresponding observed chronologies, the VS model
includes the most critical processes linking climate to annual tree-ring formation.

3.8.1 Local Simulations

Evans et al.( 2006 ) reported tree-ring width simulations performed for eight high-
latitude Russian sites (Vaganov et al. 2006 ) spanning 60 degrees of latitude
(Fig. 3.4 ) , and using the same set of fixed parameters and closely collocated mete-
orological station data. Despite neglecting adjustment of model parameters to fit
local site characteristics, seven of eight simulations are significantly correlated with
actual chronologies at or above the 95% confidence level. Four of eight correlations
between 5-year means of the simulations and actual chronologies were significant
at or above the 90% confidence level. Cook and Pederson ( Chapter 4 , this vol-
ume) discuss the problem of emergent phenomena in dendrochronology and the
resulting uncertainties in statistically modeling tree-ring data, using as a case study
tree-ring chronologies of a number of species and locations from near the Mohonk
Lake, New York, meteorological station. Their work raises the question: How do
we separate empirically demonstrated emergence, pervasive in the biological sci-
ences, from incomplete understanding of the system at hand? In an effort to address
this question, we simulated the NY004r.crn Mohonk Lake tree-ring width residual
chronology ( http://www.ncdc.noaa.gov/paleo /) using the Vaganov-Shashkin model
and Mohonk Lake station daily meteorological data (NCDC Cooperative Station
305426).
We found that with only an increase in the soil moisture drainage rate (all other
model parameters at default 'Russian' settings; Anchukaitis et al. 2006 ; Vaganov
et al. 2006 ) , the simulation was sufficient to explain the gross features of this data
series (annual r up to 0.57 ( p

<

0.05); Fig. 3.5c ) . The significance of r was not
highly sensitive to the exact value of the soil moisture drainage parameter we chose.
Correlation functions for simulated and actual chronologies are similar (Fig. 3.5a , b ) .
The most important of these are (1) positive correlation with current-year May-July
precipitation, (2) negative correlation with current-year May temperature; (3) posi-
tive correlation with current-year March-April temperature; (4) positive correlation
with prior-year September-October precipitation. Lower-frequency model skill may
be a consequence of the influence of the size of initial cambial cells from the prior
growing season on the ring width of the following year (results not shown).
Although Cook and Pederson ( Chapter 4 , this volume) pointed to observation
(3) above as an example of an emergent phenomenon, it is predicted with the VS
model simulation. Examination of the growth-limiting functions G T , G E , G W , and
the integrated growth function G broadly confirm these results for an average year