Precipitation in Aberdeen, SD: data analysis approach

The daily, monthly and annual sum of precipitation was analyzed for station Aberdeen-Airport, South Dakota (NOAA COOP #390020) with traditional normative statistical descriptions of precipitation variability and with methods of data analysis. The goal for research was formulated as finding a way to forecast extremes, such as the flooding of Aberdeen in May 2007 after a 30-hour, 8.42 inch rain event. Daily data were available, analyzed and cannot be considered as having any use for the forecast. The observations of 1423 monthly sums of precipitation have the following characteristics: mean = 1.802 in.; geometric mean = 0.856 in.; median = 1.26 in.; minimum = 0.0; maximum = 12.39 in. Simplified Fourier analysis was performed for monthly data, and 31 cyclic components were determined and used in a model to represent the data with periods from 4 to 379 months. The biggest amplitude from those components belongs to period of 12 months (1.396), and the next largest amplitude (0.261) had a period length of 6 months. The lowest amplitude (0.033) has a period of 75 months; the longest period 379 months has an amplitude of approximately twice that height (0.066). Fourteen periods have amplitudes higher that 0.1. The same analysis was performed on data obtained from the High Plains Regional Climate Center that was collected between January 1932 and April 2006. The model has mean equal 1.623 in. and the same two periods: 12 and 6 months, - with the biggest amplitudes: 1.315 and 0.258, and the model showed the increase in precipitation during May 2007 (3.494 in) that bigger that in April (2.965 in), June (3.115 in) 2007, and May (2.246 in) 2006. Mean annual precipitation for the time interval 1891- 2007 was 21.627 in. Simplified Fourier analysis presented a model of annual variability with eight cyclic components. The biggest amplitude (2.26) has the longest period of 46 years; the smallest (0.336) has a period of 31 years. Five components (46, 3, 10, 24, and 8 years) have amplitudes larger that 1.0 The same analysis for annual sum for time interval 1932-2006 presents mean = 19.514, and a model with five components of 3, 7, 14, 18, and 26 years with sequential altitudes 1.525, 1.451, 1.084, 1.066, and 1.451. The model shows that 2007 annual precipitation is moderately greater than the long-term mean value. During 118 hydrological years, the mean of annual sum for hydrologic year is 21.617 in., the minimum is equal 9.07 and maximum is equal 41.44. The highest mean of 3.704 in. for monthly sum during this time interval has June, the second of 2.948 in. has May; but May has the highest maximum equal 12.39 and June only has 10.91. The factor analysis on the matrix of monthly sum of precipitations {P118*12,13} shows that the annual sum mostly associated with May and August; the seasonal variability may be described by five factors (cumulative variance reflected by the model is equal 57.4%). The seasonal regime may be well represented by the factor scores; four factors charts (for factors 1, 3, 4, and 5) show different degrees of decrease in sum of precipitations for analyzed interval. In contrast, for factor 2 slightly increased. The assumption of possible forecast for precipitation and connected flooding was considered based on the obtained results. The highest daily sum of precipitation may be expected in month with biggest monthly sums in May and June. The time series of annual sum and monthly sums of precipitations have cycles with periods of 24, 31 and 46 years, and 286 and 379 months; traditional tools of characterization variability have to be applied to the longest as it is possible time series