Spatial and Temporal Variability of Springtime Freeze Events and their Impact on Tart Cherry Production in the Great Lakes Region

In contrast to the majority of food and fiber crops produced in the Great Lakes Region where the primary climatological limiting factor is the relative lack or abundance of water, the primary limiting factor for tart cherry crops is springtime freeze events following initial phenological development of the crop which damage buds and reduce yields. This study analyzed the spatial and temporal variability of springtime freeze events in Michigan, the impact of the freeze events on tart cherry yields, and the use of gridded climatic datasets in a tart cherry yield model. The spatial and temporal variability of radiation freeze events over complex topography in spring was analyzed using datasets for 1960-2015 for 11 stations across western Lower Michigan. The influence of the Great Lakes plays a major role in the spatial variability; there is greater damage at stations at increasing distances from the lakes. In general, radiation freeze events were found to be relatively more common and severe away from the Great Lakes and in lower elevation areas, which prevents economically viable fruit production. The temporal variability was examined by looking at long term data from two individual station sites weather data series, Maple City in northwest Michigan and Hart in west central Michigan. To illustrate the relationship between freezes and tart cherry yields, a model from Zavalloni et al. (2006) that simulates impacts of damage based on phenological stages was applied. For both the Maple City and Hart stations, the largest number of damaging freeze events over the past 50+ years was found to occur during the full bloom phenological stage. Over time, phenological development has tended to begin earlier in the growing season at Maple City. The date of the vegetative side green stage has become earlier by approximately 1 day per decade since 1960. The changes in dates of phenology over time are not symmetrical within the phenological stages, with relatively greater changes in the earlier stages. The effects of spatial aggregation of freeze impacts over an area were examined using 2 versions of the tart cherry yield model. The first version used temporal shifts in Maple City weather data for 1960-2015 to represent differences in phenology between Maple City and other locations. The second version of the yield model used weather observations from 29 automated weather station sites from Michigan's Enviro-weather Network and NWS COOP for 2005-2015. Both these versions of the model utilized kriging to spatially interpolate between stations, and both versions used cherry production data to spatially weight the model output. The use of gridded datasets was evaluated in the second version of the tart cherry yield model. A smoothed data field dataset and a reanalysis dataset were run through the tart cherry yield model and were spatially weighted. The use of gridded datasets were found to be less suitable for this application than the individual station site observations.