The Northern Oscillation Index as a Predictor of Precipitation in California

Abstract:

We show that predictions of mean annual precipitation, and daily extremes, in the San Francisco Bay region are more reliably based on the state of the Hadley-Walker circulation’s North Pacific branch, than on the basis of the state of ENSO. We suggest the same may apply over the broad North-Central California region, where we document higher correlations of precipitation with monthly values of NOI (Northern Oscillation Index) than with any of the monthly ENSO indices (including MEI, NINO3.4, and SOI). We use the NOI definition by Schwing et al (2002): the difference in sea level pressure anomalies between the North Pacific High and Darwin, Australia – the centers of action of the North Pacific branch of the Hadley-Walker circulation. While NOI is strongly correlated with all of the ENSO indices, on those occasions where NOI and ENSO indices disagree, NOI is most often the better precipitation predictor. When using NOI as a precipitation predictor, the additional use of ENSO indices adds little additional predictive skill. We additionally show that NOI is also a better single predictor of extra-tidal water height in San Francisco Bay (compared to any ENSO index), although NOI-based water height predictions can indeed be improved by the addition of an ENSO index.

We show that NOI variability can be used to characterize the San Francisco Bay region’s water resources and flooding risk, and is the controlling factor in obtaining future projections. We use monthly NOI and daily specific humidity at 850 hPa level (HUS) as predictors of precipitation in statistical models: [model 1] a statistical model of precipitation totals for the wet season (Nov.-Mar.); and [model 2] a statistical model of extreme daily precipitation in South S.F. Bay. We also use NOI and SOI as predictors in a statistical model of extreme storm surge height in the South Bay [model 3]. Each of these 3 models is trained with NOI, SOI and HUS from Reanalysis, and is run using projected values of NOI, SOI and HUS by seven CMIP5 GCMs. These models may be useful in water resources and flood protection infrastructure planning, including levees. Funded NASA Grant NNX12AG33G.