California Chilling - Materials & Methods

A full description of the methods is given in the following paper:
Luedeling E, Zhang M and Girvetz EH, 2009. Climatic changes lead to declining winter chill for fruit and nut trees in California during 1950-2099. PloS ONE 4 (7), e6166.
The publication can be accessed (for free) here:

Because the full details are given there, here is only a summary of the methods:

The analysis was based on both hourly and daily records of historic weather throughout California, as well as on a range of climate projections by statistically downscaled Global Climate Models. For each of 20 climate scenarios, a weather generator and observed correlations between hourly and daily temperatures were used to generate 100 years of hourly weather data. This is not a time series, but represents 100 replicates of a typical year for each scenario, with variation introduced by a random function of the weather generator. These records were then fed into a winter chill model (the 'Dynamic Model') to arrive at seasonal totals of Chill Portions, a unit used to quantify the amount of winter chill available per season. The output was then a population of 100 seasonal winter chill totals for each climate scenario.

Safe Winter Chill

It is not sufficient for a grower of tree crops to fulfill the chilling requirement of his/her trees, only occasionally, or in an average year. To ensure high productivity, requirements must be met every year, or at least in most years. In this study, the amount of chilling that is exceeded every year with a 90% probability was interpreted as Safe Winter Chill. That means, that if a grower grows a tree cultivar, whose chilling requirement equals the Safe Winter Chill value of his growing region, the tree's requirements will be met with 90% probability in a given climate scenario. Mathematically, Safe Winter Chill is the 10% quantile of the population of 100 seasonal chill totals. All maps available on this website show Safe Winter Chill.

Chilling Model

Many growers in California do not use the Dynamic Model for assessing winter chill. Instead, they use older models, such as Chilling Hours or the Utah Model. We regret that at present this may restrict the usefulness of much of the information presented here for many growers. The reason for this is simply that the older models, particularly the Chilling Hours model, have been shown to be inaccurate under warm winter conditions, such as those prevailing in California. In addition to that, our own work has shown that these models are very sensitive to increases in temperature. This means, that if the Chilling Hours model were used in the same way as the Dynamic Model here, predicted winter chill losses would be a lot higher than those shown on this website. Given an increasing amount of evidence that the Chilling Hours is not reliable in warm climates, and unsuitable for climate change analysis, we therefore refrained from putting projections with this model up on this site. The purpose of this website is to provide information on the quantitative extent of expected changes, and we did not want to publish information which we believe to be inaccurate exaggerations of expected changes. In the long term, we hope that the Dynamic Model, which appears to be the most useful model existing at present, will be adopted by more and more growers and extension services. In the short term, we can only apologize for this mismatch between methods used here and established grower practice. Because the Dynamic Model is still unfamiliar to many, to provide a frame of reference we have compiled a list of chilling requirements for many temperature tree crops from the literature. View table here.

<< Overview >> Results