Linkage between the occurrence of daily atmospheric circulation patterns and floods: an Arizona case study

The daily occurrence of large-scale atmospheric circulation patterns (CPs) is linked with the partial duration series of floods, using a case study in Central Arizona (USA) to illustrate the approach. The probabilistic linkage is evaluated by means of two performance indices, relating flood occurrence, observed CP occurrence before floods and purely random count of CPs. Three seasons (summer, autumn and winter) are distinguished, and CPs are grouped into two flood-producing groups year round, and two more such groups in winter. Floods in five watersheds, whose areas range from 900 to 15 000 km2, are analyzed for the period 1940–1980. The number of days N to be considered before the flood day is investigated using the first performance index, yielding a value of 1–3 days. The two performance indices appear to measure the linkage in a suitable way, especially in the autumn and winter seasons. The first index, measuring the ratio of percentage of floods explained by the two (or four) groups of flood-producing CPs to the percentage of CPs in the population, is well above 1.5. The second index, measuring the probability of at least k days out of N with CP type i before the flood day, is considerably greater than the corresponding binomial or Bernoulli value. Results for summer, when precipitation stem from convective storms, are not as clear-cut. In any case, this method makes it possible to study the effect of non-stationarities in the time series of CPs, and results should improve when daily precipitation or at least daily flows are considered.