Around-the-clock and accurate weather data are at the core of DAS. But when they are missing, DAS cannot make model predictions. While insect and plant development models are less prone to disruption, disease models cannot run without the necessary weather input. This has to do with the differences in biology that determine the model input requirements.
In general, the speed of insect, pathogen, and plant development is governed by temperature. Each species has its own favored temperature range. But disease pathogens also need rain events and/or certain periods of wetness to infect the trees. And that is where accurately predicting infection events gets tricky without proper weather data.
DAS collects real-time weather data from local station networks to calculate current model conditions. That data then gets thoroughly checked for errors and missing values, which can happen when a sensor goes bad or the external weather service is down. This is a critical step, because invalid or missing data interrupt model calculations.
To minimize these interruptions, DAS substitutes invalid or missing values with data from other sources as much as possible. For example, daily maximum and minimum temperatures can be easily replaced with daily forecast values from weather forecasting services. That way, all models that use this type of data input, continue to provide predictions, namely the insect and plant development models.
DAS already uses these daily forecast temperature data to make predictions about pest, disorder, or plant development for up to 16 days into the future. Beyond that, up to 42 days ahead, DAS uses the local 10-year average of the temperature values for the particular dates of interest.
Let’s go back to disease models. What makes predicting infections so tricky? Pathogen development happens on a much faster scale than that of insects or plants. Therefore, models run several times a day using hourly temperature, precipitation, leaf wetness, relative humidity data from the local weather station networks.
If those records are missing, however, filling in hourly data from forecast weather services is not possible, because weather forecasts do not provide the same kind of data. For example, forecast weather services do not provide information on leaf wetness, and they can only give estimates for the probability and amount of precipitation. We have all experienced that the accuracy of such forecasts can vary.
However, for disease control strategies that focus on eradication sprays after an infection occurred, it is crucial to accurately predict the time when that infection period began. Certain fungicides only work within a narrow window of time after infection started. So, for the lack of reliable hourly forecast wetness data, DAS currently cannot predict disease infection events when local real-time data is unavailable.
We make every effort to keep all DAS models running with adequate and accurate weather data, because any model output is only as good as the model input.
Ute Chambers