Apple Cultivar Frost Vulnerability During Apple Bloom

Welcome to the Western Agricultural Research Center Apple Frost Modeling dashboard created in collaboration with the Montana Climate Office. This resource provides information describing the “frost risk” to various apple cultivars during critical phenological phases across the state of Montana. This analysis has been conducted for a historical 30 (1991-2020) and 10 (2011-2020) year period. We have also conducted this analysis using future climate projections for the next 20 years (2021-2040). Additionally, this resource quantifies the reduction in frost risk associated with transitioning production from early blooming cultivars to mid- or late blooming varieties. Finally, we compute the “observed frost occurrence”, or the number of days per year that minimum daily temperatures drop below 25 ˚F between the relevant growing degree day range for early, mid and late blooming cultivars from 2011-2020. A more detailed description of the methods used to conduct this analysis, including the specific growing degree day ranges used for each cultivar, can be found within the “methods” tab below.


Frost Risk Maps

Risk Mitigation

Frost Occurrence (2011-2020)


Frost Risk Mapping:

“Frost” during bloom was defined as air temperature (for any duration of time) less than or equal to 25˚F when bloom models predict 50% or greater full bloom. Predicted apple bloom is based on single sine (ss) growing degree day (ssDD) accumulation following January 1 with a lower threshold of 42 ˚F and an upper threshold of 77.6 ˚F (more information in the single sine method can be found at Note, this method assumes that endodormancy chill requirements have been met by January 1. “Early, mid-, and late-blooming” apple categories are separated by geographical region; west and east of the continental divide. Early, mid-, and late-blooming ssDD categories are defined as: 333-513 ssDD (early - west), 277-421 ssDD (early - east), 390-593 ssDD (mid - west), 432-565 ssDD (mid - east), 441-689 ssDD (late - west), 470-607 (late - east).

To produce maps of “frost risk” (e.g. tab 1) we followed a stepwise process. First, we calculated the cumulative ssDD over each year and for each location in Montana using the single sine method. We conducted this analysis over various time periods; 1991 - 2020, 2011 - 2020 and 2021 - 2040. Note, datasets used for this analysis included gridMET (for historical time periods) and MACA for (future time periods). More information on these data can be found below. Next, we located all areas/days (at a 4km resolution) that fell within the ssDD range of interest for “early, mid-, and late-blooming” apple blooming categories for the western and eastern portions of Montana independently. This geographical difference was defined as west and east of the continental divide. For each of these locations/days we extracted the associated minimum temperature for that location/day. Using these data, we developed a pixel specific distribution of minimum temperatures that fall within the ssDD range of interest. Using this minimum temperature dataset, a pixel specific empirical cumulative distribution function (ECDF) was used to estimate the probability that minimum temperatures will be equal to or less than 25F during any day for the period of interest. In general, frost risk is much higher in the south central portions of Montana, when compared to western and north eastern Montana.

NOTE: All historical meteorological data used here is from the gridMET dataset. For more information on this dataset please visit: All future climate data was from the MACA dataset. For more information on this dataset please visit: Future data assumed an RCP8.5 scenario and was generated by the NOAA’s Geophysical Fluid Dynamics Laboratory earth system model GFDL-ESM2M;

Risk Mitigation:

To compute the reduction in frost risk associated with transitioning from earlier to later blooming cultivars, we simply calculated the difference in frost risk probabilities between two cultivars for a given time period (e.g. 1991-2020). For example, the “Mid instead of Early (1991-2020)” map in tab 2 shows the difference in frost risk if one were to choose a mid-blooming cultivar instead of an early-blooming cultivar. Mathematically this map represents P(mid-blooming frost) - P(early-blooming frost), were P(x) represents the empirical probability of observing x.

Observed number of frosts

To aid in the interpretation of the frost risk maps, we have also computed the number of frost events that have occurred during the ssDD ranges described above over a recent 10 year period (2011-2020). Maps in tab 3 “Frost Occurrence (2011-2020)” show days with frost per year for each location and cultivar.

Data Download


Produced by the Montana Climate Office

Supported by the Western Agricultural Research Center

Montana Forest & Conservation Experiment Station

University of Montana

32 Campus Drive

Missoula, MT 59812

Phone: (406) 243-6793