Factoring climate variability into New York City's water management practices

It appears that the total capacity of New York City's seven-reservoir system is at least moderately reliant on the phases of atmospheric-oceanic oscillations. The indices of interest in this study are the North Atlantic Oscillation (NAO), Atlantic Multidecadal Oscillation (AMO), El Niño-Southern Oscillation (ENSO), and Pacific Decadal Oscillation (PDO). Of these, PDO, AMO, and ENSO appear to be the most significant predictors. During the positive PDO phase, three-month averages of the above climate indices highly correlate to the normalized peak and minimum reservoir levels (May and October, respectively), with the maximum correlation lag between each index and the reservoir levels varying per index. For example, the highest correlation between May reservoir levels and cumulative Hudson Valley/Eastern Plateau climate division precipitation occurs around 10-11 months (and only 5-6 months for minimal annual levels, October), while the highest correlation between the reservoirs and PDO occurs with a 24 month lag. These correlations are only slightly less significant during a negative PDO. The relationships between all of these climate indices and the reservoir levels seem to hinge on the phase of the PDO. Using 3 predictors (PDO, Niño3.4, and NAO) of the maximum annual reservoir levels for discriminant analysis and cross validation in a positive PDO phase, 1977-97, we obtain a skill score over 90.0 (cross validated with a skill score of 62.0), which corresponds to 19 of 21 years being classified correctly. Using 2 predictors (SOI and AMO) during negative PDO periods, the maximum skill score is still ~70.0, corresponding to correct classification of 27 of 32 years. The ultimate goal of this study is to use this information for better management practices in the NYC watershed.