Climate Influences on Water Surface Levels of Devils Lake, North Dakota
Nancy Steinberger – Federal Emergency Management Agency
Ellie Baldwin – MetropolitanStateCollege of Denver
Jeffrey D. Niemann – ColoradoStateUniversity
Abstract
Since 1940, the surface level of DevilsLake in North Dakota has risen over 50 feet, inundating homes, businesses, and property. Over $450 million has been spent for flood mitigation, including constructing levees and raising roads (Vecchia, 2008). In order to improve projections of future lake level trends, a better understanding of the climate conditions associated with rising and falling lake levels is needed. The purpose of this project is to investigate the climate variables associated with long-term, annual, and seasonal variability of the surface levels of DevilsLake. The project consists of three main parts. First, a composite analysis was performed using the years and seasons with the largest increases and decreases in lake levels to identify anomalous atmospheric variables in the National Centers for Environmental Prediction (NCEP) reanalysis dataset. Second, a correlation analysis was performed between the monthly, seasonal, and water-year timeseries of lake levels and atmospheric variables from the NCEP reanalysis dataset over a mapped domain space. Finally, the influence of the Pacific Decadal Oscillation (PDO), Atlantic Multi-decadal Oscillation (AMO), and sunspots on the long-term behavior of the lake was investigated using correlation analysis. Based on the NCEP composite and correlation analyses, several atmospheric variables appear to be associated with rising and falling lake levels. One of the strongest responses was due to the jet-stream level wind. Based on Osborne (2000), southwesterly jet-stream level winds tend to be associated with increases in lake levels, whereas a reduction in the strength of the jet-stream winds, or a change in direction, results in decreased lake levels. The results of the composite analysis confirm that this is an importantinfluence on DevilsLake levels, with 9 out of 10 seasonal composites displaying the expected behavior in the lake level. The mapped correlations between lake levels and jet stream zonal and meridional winds also produce the expected result, but the correlations are not as strong. Correlations were also identified between the lake level monthly timeseries and the three climate timeseries investigated in this study: PDO, AMO, and sunspot number. Water-year and winter seasonal correlations were significant for AMO and sunspot number but not for PDO.