SUNConferences, New Frontiers in Forecasting Forests 2018

Font Size: 
MODELING CLIMATE EFFECT ON CARBON SEQUESTRATION USING TREE RING MASS SERIES
Tessie Tong, Mark Frith

##manager.scheduler.building##: Wallenburg Research Centre (STIAS)
##manager.scheduler.room##: Main auditorium
Date: 2018-09-25 04:25 PM – 04:50 PM
Last modified: 2018-05-11

Abstract


Most forest carbon sequestration are estimated based on growth and yield models, usually without considering climate impacts and differences in wood/carbon density. Studies have shown that there exist a differential up to 20% in carbon stored in tree stems between species (Lamlom and Savidge, 2003). Ring width series (sequences) have been used as climate proxy to study past climate. This study looks into whether or not ring mass series would store similar climate information that could be directly used to forecast future carbon sequestration accurately.

Data for this study came from SilviScanTM measurements on discs sampled at multiple heights of 35 black spruce and balsam fir trees from boreal forests in Mauricie and Gaspésie regions of Québec, Canada. The SilviScan data included pith-to-bark sequences of ring width, density, and other wood and fibre properties. After cross-dating, measurements for the same ring at different heights were identified. Ring volume was estimated using ring width sequences at different heights, assuming a truncated cone between two consecutive heights and a cone between the highest disc and tree tip.  Ring mass was estimated by summing the products of ring volume and average density of each section between two consecutive heights. After standardization using dplR package (Bunn 2008, 2010, Bunn et al. 2018), ring mass sequences were compared with ring width sequences at breast height (BH). Chronologies were created using biweight robust mean estimation for each group of trees from close proximity (plot) that shared similar weather conditions and disturbances and subsequently evaluated for their correlation with climate variables using treeclim (Zang and Biondi 2015), PCA (Le et al.2008), pls (Mevik et a. 2016) packages in R (R Core Team, 2018).

Although ring widths at the upper heights, particularly those near the pith, are greater than that of lower heights for the same ring, ring width profiles from all heights follow a similar trend of year-to-year variation. Mass profile also shows a similar rhythm of fluctuation to ring width.

Despite the marked differences in raw sequence profile between ring width at BH and ring mass/volume, the variations in ring indices after standardization were well captured regardless of which raw sequence (ring width or ring mass) was used with some exceptions. Standardization removes low frequency growth trend from a ring sequence, allowing the impacts of external forces (climate, disturbances, etc.) to reveal. Standardization of the whole mass sequences was sometimes non-converging due to the near-zero mass values for rings formed at young ages (e.g., < 8-10 years), therefore these rings were pruned before the standardization. This should have limited effects on other rings, however, the information from these rings was lost.

Ring width is generally greater at higher up positions. Intuitively this suggests the width of a ring is affected more by the height position at a tree trunk than by climate. However, since standardization removes growth trend and scales sequences to an average value of indices being one, the wider ring width at higher positions in the crown might not necessarily contain stronger or weaker climate signals. In addition, since rings at higher up positions normally have smaller diameters, they account for a smaller portion in total ring volume.

Dendroclimatology analysis, principal component analysis (PCA), and partial least square regression analyses revealed a poor correlation between climate variables and tree ring mass chronologies with less than 5% explained variance for this sample set. A Y-Aware PCA method (Zumel, 2016) improved the explained variance to 56% for some plots where spruce budworm attacks in late 1970’s were less severe. The poor correlation in this study suggests that while climate change may affect carbon sequestration in forests directly (e.g., slower or faster growth due to climate change), a more significant impact of climate change would manifest more indirectly, by causing more frequent and severer natural disturbances such as insect outbreaks, diseases, and fires.

Although the standardized indices from both ring width and mass sequences are similar, further work is needed to confirm whether tree mass sequence is equivalent to ring width sequences from higher up positions regarding climate change effect. Nevertheless, future forest forecasting warrants the inclusion of climate change and its resultant natural disturbances.


References


Lamlom, S.H. and Savidge, R.A.. 2003. A reassessment of carbon content in wood: variation within and between 41 North American species. Biomass and Energy, 25, pp. 381-388

R Core Team. 2018. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

Mevik, B.H., Wehrens, R. and Liland, K.H. 2016. pls: Partial Least Squares and Principal Component  Regression. R package version 2.6-0. https://CRAN.R-project.org/package=pls

Bunn, A.G. 2008. A dendrochronology program library in R (dplR). Dendrochronologia, 26(2), pp. 115-124. ISSN 1125-7865, doi: 10.1016/j.dendro.2008.01.002 (URL: http://doi.org/10.1016/j.dendro.2008.01.002).

Bunn, A.G. 2010. Statistical and visual crossdating in R using the dplR library. Dendrochronologia, 28(4), pp. 251-258. ISSN 1125-7865, doi: 10.1016/j.dendro.2009.12.001 (URL: http://doi.org/10.1016/j.dendro.2009.12.001).

Bunn, A., Korpela, M., Biondi, F., Campelo, F., Mérian, P., Qeadan, F., Zang, F., Pucha-Cofrep, D., and Wernicke, J. 2018. dplR: Dendrochronology Program Library in R. R package version 1.6.7.  https://CRAN.R-project.org/package=dplR.

Le, S., Josse, J., Husson, F. 2008. FactoMineR: An R Package for Multivariate Analysis. Journal of Statistical Software, 25(1), pp. 1-18. 10.18637/jss.v025.i01.

Zang C. and Biondi, F. 2015. treeclim: an R package for the numerical calibration of proxy-climate relationships. Ecography, 38(4), pp. 431-436. ISSN 1600-0587, doi: 10.1111/ecog.01335 (URL: http://doi.org/10.1111/ecog.01335).

Zumel, N. 2016. Principal components regression, Pt. 2: Y-Aware methods. http://www.win-vector.com/blog/2016/05/pcr_part2_yaware/. Accessed Jan. 20, 2018.

 


An account with this site is required in order to view papers. Click here to create an account.