Relationships between Pacific and Atlantic ocean sea surface temperatures and U.S. streamflow variability

Abstract

An evaluation of Pacific and Atlantic Ocean sea surface temperatures (SSTs) and continental U.S. streamflow was performed to identify coupled regions of SST and continental U.S. streamflow variability. Both SSTs and streamflow displayed temporal variability when applying the singular value decomposition (SVD) statistical method. Initially, an extended temporal evaluation was performed using the entire period of record (i.e., all years from 1951 to 2002). This was followed by an interdecadal‐temporal evaluation for the Pacific (Atlantic) Ocean based on the phase of the Pacific Decadal Oscillation (PDO) (Atlantic Multidecadal Oscillation (AMO)). Finally, an extended temporal evaluation was performed using detrended SST and streamflow data. A lead time approach was assessed in which the previous year's spring‐summer season Pacific Ocean (Atlantic Ocean) SSTs were evaluated with the current water year continental U.S. streamflow. During the cold phase of the PDO, Pacific Ocean SSTs influenced streamflow regions (southeast, northwest, southwest, and northeast United States) most often associated with El Niño–Southern Oscillation (ENSO), while during the warm phase of the PDO, Pacific Ocean SSTs influenced non‐ENSO streamflow regions (Upper Colorado River basin and middle Atlantic United States). ENSO and the PDO were identified by the Pacific Ocean SST SVD first temporal expansion series as climatic influences for the PDO cold phase, PDO warm phase, and the all years analysis. Additionally, the phase of the AMO resulted in continental U.S. streamflow variability when evaluating Atlantic Ocean SSTs. During the cold phase of the AMO, Atlantic Ocean SSTs influenced middle Atlantic and central U.S. streamflow, while during the warm phase of the AMO, Atlantic Ocean SSTs influenced upper Mississippi River basin, peninsular Florida, and northwest U.S. streamflow. The AMO signal was identified in the Atlantic Ocean SST SVD first temporal expansion series. Applying SVD, first temporal expansions series were developed for Pacific and Atlantic Ocean SSTs and continental U.S. streamflow. The first temporal expansion series of SSTs and streamflow were strongly correlated, which could result in improved streamflow predictability.

Spatial cross‐correlation patterns of European low, mean and high flows

“…Tootle and Piechota, 2006;Barlow et al, 2001;. In the context of this study, however, the pronounced similarity to the equivalent patterns found in precipitation and temperature is a sufficient proof that the space-time patterns of low-frequency runoff follow closely the atmospheric drivers.…”

“…Krakauer and Fung, 2008;Small et al, 2006;Tootle and Piechota, 2006] and large scale atmospheric circulation patterns [e.g. Barlow et al, 2001;Rajagopalan et al, 2000;Tootle and Piechota, 2006;Tootle et al, 2005].…”

“…Thus, the spatial scale of a continental river system may be predominantly determined and constrained by the drainage network and the temporal scale is likely to be primarily controlled by hydrodynamic routing. However, regions of simultaneous variation of seasonal river flow from large number of stations from all over the United States were not in correspondence with large-scale drainage basins [Lins, 1997] but rather related to features of large-scale atmospheric variability [Tootle and Piechota, 2006;Barlow et al, 2001]. Similarly, process controls of the seasonality of runoff (i.e.…”

“…Lins, 1997;Gudmundsson et al, 2011] that are related to large-scale atmospheric drivers [e.g. Barlow et al, 2001;Tootle and Piechota, 2006]. However, the variance of annual runoff varies largely among catchments [for a global analysis see McMahon et al, 2007] and is on average larger than the variance induced by annual precipitation and evapotranspiration alone [as demonstrated for US streamflow by .…”

Complex networks for streamflow dynamics

“…The data are average monthly values (not anomalies). During the past few decades, a large number of studies have investigated the above streamflow data set (or a part or variant of it) in many different contexts (e.g., Slack and Landwehr, 1992;Kahya and Dracup, 1993;Vogel and Sankarasubramanian, 2000;Sivakumar, 2003;Tootle and Piechota, 2006;Patil and Stieglitz, 2012;Sivakumar and Singh, 2012;Kiang et al, 2013). Some of these studies have explicitly addressed the connections of streamflow between the stations, including in the context of data correlations, catchment similarities, and other measures; see, Patil and Stieglitz (2012) and Kiang et al (2013) for some recent studies.…”