Search

In Part I of this paper, the optimal ranking regime (ORR) method was used to identify intradecadal to multidecadal (IMD) regimes in U.S. climate division temperature data during 1896–2012. Here, the method is used to test for annual and seasonal precipitation regimes during that same period. Water-year mean streamflow rankings at 125 U.S. Hydro-Climatic Data Network gauge stations are also evaluated during 1939–2011. The precipitation and streamflow regimes identified are compared with ORR-derived regimes in the Pacific decadal oscillation (PDO), the Atlantic multidecadal oscillation (AMO), and indices derived from gridded SST anomaly (SSTA) analysis data. Using a graphic display approach that allows for the comparison of IMD climate regimes in multiple time series, an interdecadal cycle in western precipitation is apparent after 1980, as is a similar cycle in northwestern streamflow. Before 1980, IMD regimes in northwestern streamflow and annual precipitation are in approximate antiphase with the PDO. One of the clearest IMD climate signals found in this analysis are post-1970 wet regimes in eastern U.S streamflow and annual precipitation, as well as in fall [September–November (SON)] precipitation. Pearson correlations between time series of annual and seasonal precipitation averaged over the eastern United States and SSTA analysis data show relatively extensive positive correlations between warming tropical SSTA and increasing fall precipitation. The possible Pacific and northern Atlantic roots of the recent eastern U.S. wet regime, as well as the general characteristics of U.S. climate variability in recent decades that emerge from this analysis and that of Part I, are discussed.

The optimal ranking regime (ORR) method was used to identify intradecadal to multidecadal (IMD) time windows containing significant ranking sequences in U.S. climate division temperature data. The simplicity of the ORR procedure’s output—a time series’ most significant nonoverlapping periods of high or low rankings—makes it possible to graphically identify common temporal breakpoints and spatial patterns of IMD variability in the analyses of 102 climate division temperature series. This approach is also applied to annual Atlantic multidecadal oscillation (AMO) and Pacific decadal oscillation (PDO) climate indices, a Northern Hemisphere annual temperature (NHT) series, and divisional annual and seasonal temperature data during 1896–2012. In addition, Pearson correlations are calculated between PDO, AMO, and NHT series and the divisional temperature series. Although PDO phase seems to be an important influence on spring temperatures in the northwestern United States, eastern temperature regimes in annual, winter, summer, and fall temperatures are more coincident with cool and warm phase AMO regimes. Annual AMO values also correlate significantly with summer temperatures along the Eastern Seaboard and fall temperatures in the U.S. Southwest. Given evidence of the abrupt onset of cold winter temperatures in the eastern United States during 1957/58, possible climate mechanisms associated with the cause and duration of the eastern U.S. warming hole period—identified here as a cool temperature regime occurring between the late 1950s and late 1980s—are discussed.

An analysis method previously used to detect observed intra- to multidecadal (IMD) climate regimes was adapted to compare observed and modeled IMD climate variations. Pending the availability of the more appropriate phase 5 Coupled Model Intercomparison Project (CMIP-5) simulations, the method is demonstrated using CMIP-3 model simulations. Although the CMIP-3 experimental design will almost certainly prevent these model runs from reproducing features of historical IMD climate variability, these simulations allow for the demonstration of the method and illustrate how the models and observations disagree. This method samples a time series’s data rankings over moving time windows, converts those ranking sets to a Mann–Whitney U statistic, and then normalizes the U statistic into a Z statistic. By detecting optimally significant IMD ranking regimes of arbitrary onset and varying duration, this process generates time series of Z values that are an adaptively low-passed and normalized transformation of the original time series. Principal component (PC) analysis of the Z series derived from observed annual temperatures at 92 U.S. grid locations during 1919–2008 shows two dominant modes: a PC1 mode with cool temperatures before the late 1960s and warm temperatures after the mid-1980s, and a PC2 mode indicating a multidecadal temperature cycle over the Southeast. Using a graphic analysis of a Z error metric that compares modeled and observed Z series, the three CMIP-3 model simulations tested here are shown to reproduce the PC1 mode but not the PC2 mode. By providing a way to compare grid-level IMD climate response patterns in observed and modeled data, this method can play a useful diagnostic role in future model development and decadal climate forecasting.