![Uncertain
discrete
series
[0.05,
0.5,
0.95]
Propor<on
of
ice-‐scarred
trees
around
a
northern
lake
Propor<ons
N
trees
sampled
boucher@cerege.fr](https://image.slidesharecdn.com/ebfinlande-101027033356-phpapp01/85/Spring-flood-reconstruction-from-tree-rings-continuous-and-discontinuous-series-7-320.jpg)
Spring flood reconstruction from tree rings (continuous and discontinuous series)
- 1. É"enne Boucher Taha B.M. J. Ouarda Yves Bégin Antoine Nicault Ins<tut Na<onal de la Recherche Scien<fique Centre Eau-‐Terre-‐Environnement boucher@cerege.fr Photo credit: S. Popov World Dendro 2010, Rovaniemi, Finland
- 2. Why reconstruct spring floods? Hudson bay LA GRANDE (LG) hydroelectric complex
- 3. Why reconstruct spring floods?
- 4. Infilling of hydroelectric reservoirs Spring Why reconstruct spring floods? boucher@cerege.fr
- 5. Pros Cons A mul<-‐proxy dendrohydrological approach to reconstruct spring floods Discon<nuous series / uncertain<es / non-‐lineari<es Direct rela<onship with streamflow Con<nuous series Indirect rela<onships with streamflow
- 6. Discrete event series e.g. ice scarred trees
- 7. Uncertain discrete series [0.05, 0.5, 0.95] Propor<on of ice-‐scarred trees around a northern lake Propor<ons N trees sampled boucher@cerege.fr
- 8. Model uncertain"es of discrete event series Uncertain<es around observed propor<ons are binomialy distributed 1) Model verifica"on Jackknife resampling 3) Calibra"on / Reconstruc"ons Performed using Generalized Addi<ve Models (GAMs) 2) € g(E(Q)) = B0 + f1TR1 + f2TR2 + ... boucher@cerege.fr
- 10. Reconstruc<on of May water supplies to the Caniapiscau reservoir (northern Québec, Canada) boucher@cerege.fr
- 11. Hudson bay La Grande HE complex
- 12. Dendrochronological data = Earlywood density = Ice scar chronology at lake Montausier boucher@cerege.fr
- 13. Water supplies to the Caniapiscau reservoir, & hydrological variables to reconstruct boucher@cerege.fr
- 14. Discrete and con<nuous dendroseries Propor<on of ice scoured trees around lake Montausier Earlywood density, PC1 Earlywood density, PC2 boucher@cerege.fr
- 15. The Montausier lake ice scar chronology (an other look at it…) Propor<ons N=300 boucher@cerege.fr For each year, N=300 possible propor<ons were randomly generated from a binomial distribu<on. Colors indicate « the number of "mes a simulated propor<on value falls in each class »
- 16. Rela<onships between TR series & water supplies (QMay) to the Caniapiscau Reservoir Propor<ons of ice scarred trees, Montausier lake PC1 Index (wet sites) PC2 index (mesic xeric sites)
- 17. Calibra<on / verifica<on sta<s<cs & comparison with a tradi<onal MLR approach € g(E(QMay )) = B0 + f1X1 + f2X2 + ... € E(QMay ) = B0 + B1X1 + B2X2 + ... + boucher@cerege.fr
- 18. Historical May water supplies to the Caniapiscau Reservoir boucher@cerege.fr
- 19. Discussion and Conclusion -‐ A new approach to reconstruct paleofloods from tree rings in northern environment. It allows: 1) to take advantage of both discrete and con"nuous tree ring proxys 2) to model and include the uncertainty of discrete TR series within the calibra<on / reconstruc<on process 3) to account for non-‐linear rela"onships that occur in natural environments -‐ Our approach can be used in other environments / situa<ons where discrete and con<nuous tree ring series are available
- 20. BIENVILLE MONTAUSIER CORVETTE LEC NECOPASTIC HERVÉ POLARIS
- 21. Many thanks to + field assistants throughout the years boucher@cerege.fr