An Investigation of Extreme Rainfall Events in California

This study seeks to understand the factors that cause extreme rainfall over California during winter. The intense drought and the 2017 extreme rainfall events in California have motivated investigations of seasonal and subseasonal variability for potentially destructive extreme rainfall over this region. Past studies have linked El Niño Southern Oscillation (ENSO), the Madden-Julian Oscillation (MJO), Rossby wave-breaking, and atmospheric rivers (ARs) to heavy precipitation in California. The extent to which these phenomena interact and induce extreme rainfall remains uncertain and difficult to predict.

This investigation analyzed daily rainfall data for California obtained from NCEI’s Global Historical Climatology Network for winter (e.g., 1 November to 1 March). The analysis focused on data from 301 stations that were located within 80 km of the coast and had a historical record that extended over 30 years. Extreme rainfall events were defined in the following manner. First, a station was defined as experiencing extreme rainfall when the amount recorded on a given day was at or above the 99th percentile over the station’s record. Next, an extreme rainfall event occurred when the number of stations experiencing extreme rainfall on a given day exceeded the 99th percentile for the 1950-2017 time period. This approach effectively defined an extreme event both in terms of rainfall amounts and spatial extent. Subsequently, Barnes interpolation of these rainfall events was undertaken to obtain the latitudinal variation of these extremes. Cluster analysis revealed that extreme rainfall events could be divided into separate southern and northern/statewide clusters. The linkages between the occurrence of an extreme event was then investigated in terms of the dependence on ENSO, the Arctic Oscillation, and ARs using Guan and Waliser’s global AR catalog.

Overall, an examination of rainfall extremes within the context of the multivariate ENSO index, or MEI, revealed that rainfall extremes for both clusters occur over a wide range of MEI values. Extremes in the southern cluster had a distinct peak in frequency during the weak El Niño (MEI between 0.5 and 1.0), while peaks in the northern-statewide cluster appeared both in the near neutral/weak La Niña (MEI between 0 and -0.5) and moderate (1.0 to 1.5) to strong (2.0 to 2.5) El Niño. With the Arctic Oscillation (AO), extreme events generally occurred during the negative phase of AO (-AO). Peaks in the frequency of extreme rainfall events occurred during different AO phases for the northern cluster, which implied that extreme rainfall in that cluster may be modulated by a greater variety of mechanisms than for the southern cluster. Extreme rainfall in the south (north) somewhat favored +MEI and -AO (-MEI and -AO) conditions, though the weak correlation suggested that other phenomena also considerably influence extreme rainfall. ARs, examined as a potential source of smaller-scale variability, occurred during 77.1% of southern cluster and 97.6% of northern cluster events. Preliminary work suggests that 74.5% of AR-related southern events and 46.3% of AR-related northern events had ARs originating south of 23.5˚ N (i.e., in the tropics). However, for both clusters, integrated vapor transport (IVT) showed no significant difference between ARs that caused extreme rainfall and those that did not. Future work will focus on further investigation of ARs, as well as examination of the role of the MJO and the use of empirical orthogonal functions to identify leading modes of variability. Case studies will also be examined to provide a more comprehensive analysis of extreme rainfall events.