About the Science
The technology and research behind Grape Climate.
Work in Progress
This website is in its early days, and I am continuously working to add more features and detailed content. My goal is to provide a transparent look into the AI development process, the scientific models behind the recommendations, and the data sources used. Please check back for updates as the site evolves.
For more details on how the model works, I gave an explanation on the Sustainable Winegrowing podcast by the Vineyard Team.
Listen to the PodcastScoring Methodology
Popularity Score
The Popularity Score is related to the amount of growing area that my AI model predicts a particular grape will have in a given region. In practical terms, it indicates whether the climate is expected to support extensive cultivation of that variety. The percentile view shows how this score compares to all other model responses. For example, a score in the 99th percentile means it's in the top 1% of all predictions. In an average wine region, one might expect about 13 grape varieties to score above the 99th percentile and 65 to score above the 95th.
Suitability Score
The Suitability Score is designed to help growers find "specialist" varieties that may be less common but are perfectly suited for a specific climate, which is also useful for future climate scenarios. A high score indicates that your region's climate is very similar to where that grape is currently most popular in our dataset. This scaling reduces the score for major international varieties and highlights lesser-known ones. For example, if a climate is very similar to where a niche variety is almost exclusively grown, the suitability score will be very high. The percentile view shows that a given climate is among the best for a grape, as defined by this relative popularity, not by an absolute, directly measurable suitability.
For more details on the scoring methodology and model development, please refer to my M.Sc. thesis.
Read the ThesisAbout the developer
Growing up in Germany’s Baden-Württemberg and Canada’s Okanagan Valley, my passion for wine and the environment was sparked by these beautiful regions. This led me to study Bioresource Engineering at McGill University, where I earned my B.Eng. and M.Sc. and have contributed to research for over five years.
My goal is to provide solutions to pressing problems in the wine industry that can be answered by data and I am continuously expanding my skills in AI and the wine industry to do this. If you have any questions or would want to get in touch for personalized services I am one button press away :)
- Joel Z. Harms