The frequency and timing of frost events and the length of the growing season are critical limiting factors in many human and natural ecosystems. This study investigates the temporal and spatial variability of the date of last spring frost (LSF), the date of first fall frost (FFF), and the length of the frost-free season (FFS) in the Great Lakes region of the United States using two gridded reanalysis datasets for the period of 1980-2010. Across most of the Great Lakes region, there has been a negative trend for the LSF and a positive trend for the FFF, leading to a lengthening of FFS. The three variables vary spatially across the region and exhibit large interannual variability. Empirical orthogonal function (EOF) analyses indicate that the first mode for the three variables, which accounts for about 30% of the total variance, is in phase across most of the Great Lakes region and that it appears to be related to Pacific North American (PNA) pattern in the case of LSF and to Pacific Decadal Oscillation (PDO) in the case of FFF and FFS. The main reason for these connections is that the 200-hPa geopotential height anomaly over North America induced by a Rossby wave train influences the strength of the trough over North America, which in turn affects surface temperatures over the Great Lakes region. The second mode, explaining about 10% of the total variance, mainly shows the out-of-phase variability between the Great Lakes and the surrounding land and it appears to be related to the circumglobal teleconnection (CGT) in the case of LSF and again to PDO in the case of FFF and FFS.
Yu, Lejiang; Zhong, Shiyuan; Bian, Xindi; Heilman, Warren E.; Andresen, Jeffrey A. 2014. Temporal and spatial variability of frost-free seasons in the Great Lakes region of the United States. International Journal of Climatology. 34(13): 3499-3514.