How ducks and data science led to rain in Tennessee

Recently I was contacted by one of our dealers who had a grower that was questioning the accuracy of the rainfall data being reported by AgDNA. Our rainfall tracking service was telling him it had rained overnight but he was adamant there wasn’t a cloud in the sky.

This was very unusual given most growers are blown away by the accuracy of our rainfall reporting. In fact, some customers setup rain gauges on various fields to try and catch the system out — with no luck.

So we contacted our API weather partner WDT Inc and provided them with some background on what was being reported. Within the hour they had an answer. By replaying the historical data for that region and time period they were able to determine that a flock of ducks were taking off every morning at around 7 AM alongside the local radar station.

The take off pattern and the way the ducks dispersed in flight looked like a rainfall event as far as the rainfall modelling algorithms were concerned. However, once the data scientists had this new information they were able to fine-tune their algorithm filters to remove the anomaly for this region.

As far as our software was concerned, it had stopped raining!

Data Science and the Real World

With all new technologies there is a significant amount of real world practicality that needs to be accounted for. In the case of weather prediction and rainfall modelling, unforeseen events such as a flock of ducks taking off can skew the data. This is where data scientists are able to clean the data to continually improve accuracy.

This is one of the ongoing challenges for the agricultural sector as data science and predictive analytics make their way into the value chain of crop production. Whether it’s modelling the weather, predicting plant growth, or looking for yield enhancing insights throughout the season — technology must continue to evolve and account for unforeseen real-world scenarios.

This is what makes AgTech and the emerging next generation predictive analytic platforms so exciting for agricultural producers. As big-data rolls in, algorithms get smarter, the accuracy of outputs increase and the resulting insights become more valuable.

In the meantime it is up to the AgTech and data science community to ensure the feedback loop from customer to engineering is rapid and ongoing. This will ensure timely adjustments and fine tuning of the predictive models to ensure accuracy and maximise return on investment for crop producers.

If only ducks could make it rain!