SHIH-YU WANG
Forest Inventory Analysis (FIA) feasibility in reconstructing climate
Department of Plants, Soils, and Climate, and Utah Climate Center, Utah State University, Logan, Utah
JOHN D. SHAW
Justin DeRose, Simon Wang, John Shaw
Forest Inventory and Analysis, Rocky Mountain Research Station, Ogden, Utah
(b)"FIA"series"length"
Forest Service & Utah State University
#"of"years"
>50"
>100"
>150"
>200"
>250"
(Manuscript received 13 August 2012, in final form 5 November 2012)
ABSTRACT
This study introduces a novel tree-ring dataset, with unparalleled spatial density, for use as a climate proxy.
Ancillary Douglas fir and piñon pine tree-ring data collected by the U.S. Forest Service Forest Inventory and
Analysis Program (FIA data) were subjected to a series of tests to determine their feasibility as climate
proxies. First, temporal coherence between the FIA data and previously published tree-ring chronologies was
found to be significant. Second, spatial and temporal coherence between the FIA data and water year precipitation was strong. Third, the FIA data captured the El Niño–Southern Oscillation dipole and revealed
considerable latitudinal fluctuation over the past three centuries. Finally, the FIA data confirmed the
quadrature-phase coupling between wet/dry cycles and Pacific decadal variability known to exist for the
Intermountain West. The results highlight the possibility of further developing high-spatial-resolution climate
proxy datasets for the western United States. (The preliminary FIA data are provided online at http://cliserv.
jql.usu.edu/FIAdata/ in both station and gridded format.)
(c)"FIA"series"count"
(a)
year"
(a) Loca(on of FIA plots with current available tree-­‐ring series (green dots) in the Intermountain West, USA. (b) Loca(on of FIA data used in this study with the period of record in years represented by the size and color of triangles (upper right), overlaid with terrain (shading). (c) Number of valid tree-­‐ring series over (me within Utah. 1.
douglas fir
pinyon pine
Water&year:&Aug,Jul&
(c)$Cross!corr:!!
Water&year:&Aug,Jul&
Correla5on)coefficient)
(a))EOF1)FIA:)33.5%)
year$
(a) Site map of the FIA pinyon pine (red dot) and Douglas-­‐fir (pink triangle) simon.wang@usu.edu
(b)$point-by-point corr: FIA!&!Temp!
and chronology development (Fritts 1976). C
large spatial grids of chronologies for paleoclim
Networks of tree-ring data represent important conFIA!&!Precip! 99%!
construction is therefore expensive and time
con
tributions to the study of regional climate, providing in
Geographically uniform reconstructions of temp
(a))EOF1)FIA:)33.5%)
(b))PC1)FIA)
(c))Cross)corr.)w/)PDO)
situ evidence of past spatiotemporal climate variability.
and precipitation (Fritts 1991) and drought (Co
Tree-ring indices yield−)PC1)
annually dated records of−)PC1)
climate
Lag!in!month!of!the!year!
−)PDSI)
−)PDSI)
2004) have been developed for North America
events from specific locations. In the semiarid western99%)
longitude/latitude resolution, which is coarse c
United States, a voluminous literature has reconstructed
FIA!&!Temp!
ing the complex climate regimes
and terrain
past climate using tree-ring chronologies, for example,
western United States. While the number 99%!
of ch
lag$(year)$
precipitation (Gray et al. 2004b), temperature
(Briffa
gies available from the International Tree-Rin
et al. 1992), drought (Cook et al. 2004), and streamflow
Bank (ITRDB) for the western United States
corr.=0.86)
(Woodhouse et al. 2006).
substantial, sparse spatial representation rem
Classically, the
climatically
sensitivebtree⬈ Ppreparation
oint-­‐by-­‐point cof
orrela(on map (contours) etween the FIA dregions,
ata and (a) for
gridded precipita(on and Because
(b) gridded dev
many
example,
Utah.
ring chronologies
requires
careful
of species,
(b))PC1)FIA)
(c))Cross)corr.)w/)PDO)
temperature during the wselection
ater year (August-­‐July) over the period 1950-­‐1997, overlaid with terrain (shading) and dense spatial networks of classic chronologies ov
the FIA s−)PC1)
ites (black dreplication,
ots). Contours tcross
hat are dating,
significant at the 99% level are in bold; a 9-­‐point smoothing was sites (e.g., elevation),
sample
−)PC1)
areas is time and cost prohibitive, examination
applied. −)PDSI)
(c) Cross-­‐correla(on between FIA data, annual precipita(on (solid line) and annual temperature (dashed −)PDSI)
sible
alternatives
should
be
99%)
line) with a one-­‐month sliding interval from the previous calendar year (-­‐12 for Jan-­‐Dec of considered.
last year) through the Here
introduce
a tree-ring
dataset
collecte
current year (0 for Jan-­‐Dec of this year) to the next year (+12 for Jwe
an-­‐Dec of next year). The shaded area indicates Corresponding author address: Shih-Yu (Simon) Wang, Dethe 9
9% s
ignificance l
evel. partment of Plants, Soils and Climate, and Utah Climate Center, U.S. Forest Service Forest Inventory and A
lag$(year)$
Utah State University, 4280 Old Main Hill, Logan, UT 84322-4820. (FIA) Program. The FIA conducts a geogra
E-mail: simon.wang@usu.edu; rjderose@fs.fed.us
unbiased, systematic sample across the United
(b)$FIA$(DF+PP)$and$ITRDB$series$
(a) EOF1 of the gridded FIA data during the period 1800-­‐1995 DOI: 10.1175/JHM-D-12-0124.1
overlaid with the PDSI grids (red dots: Cook and Krusic 2004) and FIA sites (yellow dots). (b) PC1 series (gray blue) overlaid with the four-­‐
(c))Cross)corr.)w/)PDO) ! 2013 American Meteorological Society
point average of the PDSI (red) with their correla(on coefficient −)PC1)
indicated in the boeom right, both smoothed by a 3-­‐year moving −)PDSI)
average. (c) Cross-­‐correla(on of the reconstructed PDO index 99%)
(MacDonald and Case 2005) with PC1 (gray blue) and the PDSI (red); e.g., lag -­‐4 means that PC1/PDSI lags PDO by 4 years. The shaded area indicates the 99% significance level ager taking into considera(on the lag$(year)$
reduc(on in degrees of freedom from the moving average. corr.=0.86)
Correla5on)coefficient)
samples and four ITRDB chronologies (blue cross). (b) Chronology series of ITRDB (blue) and FIA (red) at the ITRDB loca(ons, with the light/thin lines showing raw (b))PC1)FIA)
data and thick dashed lines s(a))EOF1)FIA:)33.5%)
howing the 9-­‐year moving average. Correla(on coefficients labeled in the lower right were calculated between smoothed (me −)PC1)
series. The four ITRDB chronologies are: 1) ut501 – Stockton and Jacoby 1976 −)PDSI)
(Douglas-­‐fir); 2) ut527 -­‐ Pederson et al. 2011 (pinyon pine); 3) ut529 – Pederson et al. 2011 (Douglas-­‐fir) and; 4) ut521 – Grow 1998 (pinyon pine). (a)$point-by-point corr:!FIA!&!Precip!
Correla5on)coefficient)
(a)$FIA$and$ITRDB$sites$
Introduction
Reference:
DeRose, R. J., S.-Y. Wang, J. D. Shaw,
2013: Feasibility of high-density
climate reconstruction based on the
Forest Inventory and Analysis (FIA)
tree-ring data. Journal of
Hydrometeorology, 14, 375–381.