How Better Weather Data Will Improve Safety and Energy Efficiency

August 14, 2019

Weather forecasts can make or break your day. If the analysis is correct, you wear the right clothing and all goes well. If it’s wrong, you can get soaked, you can overheat; the list goes on.

If you’re baffled by the lack of accuracy, it’s probably worth noting that the forecast itself isn’t that old. In fact, a weather predicting body wasn’t created until 1854 when the British government launched the Meteorological Department of the Board of Trade, which was focused on improving wind charts for ships.

Since then, weather technology has come a long way. We currently have an array of tools, including satellites that use laser pulses to measure wind velocity with extreme accuracy. The broad array of weather data we’re able to generate is crucial for research on weather prediction and climate trends.

What we use this data for goes well beyond figuring out whether we should bring an umbrella to work, and the number of applications is growing as the quality of data grows.

Using this weather data, researchers are working on some cutting-edge applications that could have a significant impact on public health, energy use and more:

Preventing heat stroke: Researchers in Japan are combining anatomical models, core temperature algorithms, thermoregulatory response models, solar radiation models and weather data to predict the heat stroke threshold with 90% accuracy. That could enable cities to take precise preventative action during heat waves and save lives.
Predicting power outages: A team from the University of Connecticut is building a new predictive model for power outages, incorporating wind, precipitation and soil data, as well as utility company data, land cover data and vegetation cover data. Their work has greatly reduced the margin of error of previous efforts. Advances like this could lead to a more resilient electric grid.
Accurately estimating precipitation: Doppler radar is a familiar part of our forecasting toolkit, but it has traditionally relied on human eyes to identify clouds that are likely to cause precipitation. Researchers in China are working to make it much more accurate by using deep learning instead, which could mean much more accurate forecasts.
Reducing energy use in data centers and industrial buildings: When multiple facilities perform the same task, weather data can be used to delegate to the most efficient one. For example, by using a weather-aware scheduling algorithm, researchers proactively delegated workloads to data centers that had the smallest temperature difference between indoors and out, which resulted in a 40% savings in cooling energy. A similar strategy is being applied to the scheduling of key activities in industrial buildings (machine work, loading, packing and office work) so that they require the least amount of HVAC energy.

The feasibility of this research is largely dependent on the sensors that provide this weather data. That will be an interesting technological frontier to watch, especially as we usher in the era of 5G and the sensors that come along with it.