Climate change or climate cycles? Snowpack trends in the Olympic and Cascade Mountains, Washington, USA
Climate change could significantly influence seasonal streamflow and water availability in the snowpack-fed watersheds of Washington, USA. Descriptions of snowpack decline often use linear ordinary least squares (OLS) models to quantify this change. However, the region’s precipitation is known to be related to climate cycles. If snowpack decline is more closely related to these cycles, an OLS model cannot account for this effect, and thus both descriptions of trends and estimates of decline could be inaccurate. We used intervention analysis to determine whether snow water equivalent (SWE) in 25 long-term snow courses within the Olympic and Cascade Mountains are more accurately described by OLS (to represent gradual change), stationary (to represent no change), or step-stationary (to represent climate cycling) models. We used Bayesian information-theoretic methods to determine these models’ relative likelihood, and we found 90 models that could plausibly describe the statistical structure of the 25 snow courses’ time series. Posterior model probabilities of the 29 “most plausible” models ranged from 0.33 to 0.91 (mean = 0.58, s = 0.15). The majority of these time series (55 %) were best represented as step-stationary models with a single breakpoint at 1976/77, coinciding with a major shift in the Pacific Decadal Oscillation. However, estimates of SWE decline differed by as much as 35 % between statistically plausible models of a single time series. This ambiguity is a critical problem for water management policy. Approaches such as intervention analysis should become part of the basic analytical toolkit for snowpack or other climatic time series data.
KeywordsSnow water equivalent Intervention analysis Climate cycles Climate change Olympic Mountains Cascade Mountains
- Adam, J. C., Hamlet, A. F., & Lettenmaier, D. P. (2009). Implications of global climate change for snowmelt hydrology in the twenty-first century. Hydrological Processes 23, 962–972. Google Scholar
- Casola, J. H., Cuo, L., Livneh, B., Lettenmaier, D. P., Stoelinga, M. T., Mote, P. W., & Wallace, J. M. (2009). Assessing the impacts of global warming on snowpack in the Washington Cascades. Journal of Climate 22, 2758–2772.Google Scholar
- Box, G. E. P., & Tiao, G. C. (1975). Intervention analysis with applications to economic and environmental problems. Journal of the American Statistical Society, 70(349), 70–79.Google Scholar
- Burnham, K. P., & Anderson, D. R. (2002). Model Selection and Multi-Model Inference: A Practical Information-Theoretic Approach. 2nd edition. New York: Springer.Google Scholar
- CIG (2012). Impacts of natural climate variability on Pacific Northwest climate. Resource document. Climate Impacts Group, University of Washington. http://cses.washington.edu/cig/pnwc/clvariability.shtml. Accessed 2 August 2011.
- Cornwall, W. (2007). How one number touched off big climate-change fight at UW. Seattle Times, March 15, 2007.Google Scholar
- Gottfried, G. J., Neary, D. G., & Ffolliott, P. F. (2002). Snowpack-runoff relationships for mid-elevation snowpacks on the Workman Creek watersheds of central Arizona. Research Paper RMRS-RP-33. Fort Collins: U.S. Forest Service, Rocky Mountain Research Station.Google Scholar
- Nickels, G. (2007). State should find ways to protect City Light’s climate-protection efforts. Seattle Times, February 7, 2007.Google Scholar
- NRCS. (1984). Snow survey sampling guide. Agriculture handbook 169. Washington: U.S. Department of Agriculture, Natural Resources Conservation Service.Google Scholar
- R Development Core Team. (2009). R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing.Google Scholar
- Zeileis, A., Leisch, F., Hornik, K., & Kleiber, C. (2002). strucchange: an R package for testing for structural change in linear regression models. Journal of Statistical Software, 7(2), 1–38.Google Scholar
- Zeileis, A., Shah, A., & Patnaik, I. (2008) Testing, monitoring, and dating structural changes in maximum likelihood models. Research Report 70. Vienna, Austria: Department of Statistics and Mathematics, WU Vienna University of Economics and Business.Google Scholar