By Robert Crassweller
The reliability of the Cornell carbohydrate thinning model forecast is dependent on weather forecasts.'
As a result of funding from the Extension Committee of the State Horticultural Association of PA, all Pennsylvania growers have access to the Cornell MaluSim Carbohydrate model via the NEWA web site. You can choose the site closest to you, enter your green tip and full bloom dates, and run the model to get an estimate of the carbohydrate status of your trees. You also have access to all the other features of the NEWA site including insect and disease models. We have been using this system since 2013, with the addition of new or improved models in recent years as well as a wider geographical distribution of weather stations.
The Cornell carbohydrate thinning model is a tool that can help you decide when or what to spray based on weather conditions. We know that weather conditions can change based on location, and even though weather forecasting has greatly improved, there are still some inconsistencies that can occur. We have always stated that the carbohydrate model forecast is only as good as the weather forecast used. Fortunately, changes in the weather forecast have been included in the model prediction process.
In one of the recent runs of the model, I started in the morning and was then interrupted by the phone and other duties. I did not get back to continue running the additional sites until after lunch. When I went back to check the validity of the model outputs from what I had done several hours earlier I found that there were slight changes. The changes were not large but they did exist. I contacted Alan Lakso, and he in turn contacted Keith Eggleston who is the Regional Climatologist at the Northeast Regional Climate Center and operates the MaluSim model operations. Keith indicated that the model output can change during the day.
When the carbohydrate program receives a request for the model it pulls all the available data from the weather station you selected through the most recent hourly report to update the model. The rest of the model is filled in with the forecast data. Since the data are updated every hour, if you run the model 3 to 4 hours later, it will have 3 to 4 more hours of actual data that will replace the forecast data which was used earlier in the day, providing nearly real time data. The values will change as more observed data come into the database to replace less precise forecast data. Alan Lakso noted that, in reality, the change in data is only a gram or two of carbohydrate and usually does not affect the recommendation.
This variability, however, emphasizes the need to check the model if the weather deviates from what was predicted in the latest run. This is why we always show the current day’s prediction and carbohydrate balance as well as the next 3 days. Remember that the balance and the recommendation is a product of current weather plus predicted for the current date and the next three day predictions.