A nested bootstrapping approach (n=10,000 iterations) was repeated at each change in sample depth (3 to 16). For each iteration, chronologies were randomly selected with replacement and averaged into a composite chronology. Calendar years shared with CCwinter physical data (1948-2003) were randomly sampled with replacement and the CCwinter reconstruction from tree ring data (CCwinter_recon) was predicted using a rise-to-maximum function: CCwinter_recon = a*(1-exp(-b*oak composite chronology)). The Reduction of Error statistic was calculated between predicted CCwinter and sea level (1898-1947, see timeseries-physical) for independent verification; any positive value indicates reconstruction skill. A 21-year running standard deviation was calculated for the length of each iteration to evaluate trends in variance. The 14 ensemble medians were spliced together to form the final nested reconstruction (tree-ring and CCwinter standard deviation datasets). To examine the frequency of extreme events, CCwinter_recon values in each ensemble member were ranked, the 10 highest values retained, and the frequency per century was calculated post- and pre-1950. Note that an identical result was obtained for the extreme events analysis if the composite chronologies were ranked prior to their transformation by rise-to-maximum functions.
We tested various methods to retain inter-annual to long-term variation in the tree-ring data and found results and conclusions to be insensitive. Detrending with 50-year splines did not appreciably change the results of the study. The unusually high frequency of negative events is less pronounced post-1950, but spline detrending reduces the magnitude of extreme values and mutes or eliminates long-term trends. The two methods (negative exponential and signal free detrending) capable of retaining centennial scale variability yielded nearly identical results for the final reconstruction (R2 = 0.993). [Given the complexity and experimental nature of the signal-free datasets, negative exponential detrending was used for the final reconstruction]
Related References:
B. A. Black et al., Winter and summer upwelling modes and their biological importance in the California Current Ecosystem. Global Change Biol. 17, 2536 (Aug, 2011).
J. E. Keister, E. Di Lorenzo, C. A. Morgan, V. Combes, W. T. Peterson, Zooplankton species composition is linked to ocean transport in the Northern California Current. Global Change Biol. 17, 2498 (Jul, 2011).
T. M. Melvin, K. R. Briffa, A "signal-free" approach to dendroclimatic standardisation. Dendrochronologia 26, 71 (2008).
I. Schroeder, W. J. Sydeman, N. Sarkar, S. J. Bograd, F. B. Schwing, Winter pre-conditioning of seabird phenology in the California Current. Mar. Ecol. Prog. Ser. 393, 211 (2009).
F. B. Schwing, T. Murphree, P. M. Green, The Northern Oscillation Index (NOI): a new climate index for the northeast Pacific. Prog. Oceanogr. 53, 115 (2002).
F. B. Schwing, M. O'Farrell, J. M. Steger, K. Baltz, "Coastal upwelling indices, West Coast of North America, 1946-1995" (NOAA Technical Memo, NOAA-TM-NMFS-SWFSC, Washington, DC. 144 p., 1996).
D. W. Stahle et al., The ancient blue oak woodlands of California: longevity and hydro-climatic history. Earth Interactions In Press, (2013)