icon_play

Wanted: A New Kind of Weather Service, by Machines and for Machines

Weather forecasting has come a long way in the last few decades. Satellite photos, radar, weather observation stations around the world, and new forecasting models have all contributed to an increase in forecasting accuracy.

But one thing hasn’t changed: weather forecasting remains a human-centric activity. How will tomorrow’s rain affect my morning commute? What are the chances my Caribbean vacation will be ruined by a hurricane? When should residents be evacuated from a rising river flood zone?

But weather can also have a big impact on business, particularly on critical infrastructure like trains, trucks, farm equipment, and other expensive industrial assets. Yet, there are only a few solutions that provide insights about weather’s impact on industrial equipment.

At Uptake, we began working on industrial-focused weather solutions during a customer product demonstration. The customer noticed that a locomotive was losing traction and immediately left our software to review third-party weather data and see if snow was affecting the train’s operation. In that moment, we committed to a project to integrate weather data and analysis into every one of our asset monitoring products.

Machine learning now enables extremely accurate, hyper-local forecasts to inform machine-specific recommendations based on those predictions. Our solution integrates weather and climate data from myriad global sources—lightning observation posts, remote sensors, satellite and radar, global climate archives, even the sensors on the industrial assets we monitor. We also incorporate a menu of public and private forecasts from the world’s leading sources and blend data through our proprietary algorithm. By running all this data through the Uptake weather engine, we can provide accurate, and more importantly, relevant and actionable insights for each customer sector.

What makes this sort of information truly valuable are the insights and recommendations data science can provide to help customers take the right actions at just the right time to reduce costs and optimize operations. For example, the expected lifetime of components like air filters or batteries can be dramatically affected by weather. Fleet service managers can better understand which alerts from the machine can be ignored and which repairs might become more urgent when weather data is included in condition monitoring.

In the future, weather analytics for various industries could make an even greater difference:

Energy: Suppose you have a wind farm with hundreds of turbines producing electricity. With an accurate forecast of wind conditions, you can plan maintenance for low-wind days, and you can maximize turbine operation on high-wind days. Having a pinpoint, near-real-time forecast in conjunction with real-time asset health predictions can also help you sell at a good price by accurately predicting how much electricity you can commit to delivering in the next-day or next-hour markets.

Agriculture: Climatic factors impact every operational decision made by a farmer, but today decisions are made through gut feelings. By incorporating weather analytics with soil moisture and nitrogen sensors, as well as irrigation controls, Uptake can automatically optimize irrigation schedules. This will not only reduce water and energy costs and minimize fertilizer run-off, but keep the crops at the ideal moisture band to significantly improve crop yields. Weather also affects agricultural machinery. For example, it makes a difference to harvesting equipment if sugar cane is harvested when it is light and dry or wet from rain, which makes it tough and causes greater wear to harvester blades. Understanding the chances of rain during harvest season not only helps farmers prevent equipment breakdowns, it can help equipment dealers better manage blade inventory and plan service staffing levels.

Transportation: Global shipping companies may have hundreds of boats crossing the Pacific on any given day. Insights about wind speed and direction, and the height and direction of swells can help shippers optimize routes to minimize fuel consumption, reduce route time, and avoid unnecessary risk.

The old model of human analysis and forecasting cannot possibly deliver relevant weather insights at the scale and speed needed to optimize the performance of thousands of assets across the globe.

With machine learning, we can allow machines, trucks, mining equipment, and even buildings to communicate directly with weather data—and provide insights and recommendations that people can act on to reduce downtime and increase performance.

Edwin Campos supports weather data science and strategy at Uptake.