Hyper-Local Weather Data Provides Accuracy to Central Hudson Gas & Electric

By Jim Foerster

Weather stations make guesswork a thing of the past. Utilities can effectively determine how much energy they need to generate and purchase for any given day thanks to the specific temperature data that is readily available. This leaves little room for error and saves the utility money.

Photo courtesy: Schneider Electric

Weather plays a key role in the amount of energy used in the U.S., and severe weather is often to blame for leaving neighborhoods in darkness. To tackle weather head on, utilities use weather forecasts to make real-time decisions that ultimately add or shed load on the grid.

However, inaccurate weather forecasts can dramatically affect the amount of energy put on the grid, and in turn put a utility at risk of outages. To combat the possibility of negative effects, local weather stations provide hyper-local, point-based forecasts that allow utilities to make informed, business critical decisions – all with one goal in mind: to decrease outages throughout the seasons.

Like many utilities, this was a primary goal for Central Hudson Gas & Electric (Central Hudson). Extending from the suburbs of metropolitan New York City north to the Capital District in Albany, Central Hudson delivers natural gas and electricity to approximately 300,000 electric customers and 78,000 natural gas customers. Given the large area and diverse topography of Central Hudson’s service territory, areas can experience varying conditions simultaneously.

Central Hudson needed a clear assessment of what weather would affect its entire service area, and when, to determine what outages were on the horizon.

Left in the Dark

With outage management in mind, Central Hudson set out to develop an outage prediction model. Using historical weather data and outage history reports, an outage prediction model generates an estimate of the impact an approaching weather system will have on a utility’s service area. Due to the varying topography of Central Hudson’s service area, an outage prediction model was an important tool to develop for accurate, effective and prompt outage management.

However, the utility had a problem. The most critical component of an outage prediction model is a strong database of highly accurate historical weather data. For Central Hudson, the best source for this was the National Weather Service (NWS). While NWS is a reliable and accurate weather forecaster, very few of its weather stations are located within Central Hudson’s service area.

Public meteorological services, such as NWS, are tasked with providing forecasts that are accurate at a very general level, commonly generating an hourly forecast for a broad geography from centrally located weather stations. And its geographically specific forecasts are often based on weather models, as opposed to the collection of hard data from local weather sensors.

While these forecasts provide the information needed for the vast majority of citizens and businesses, even slight differences in temperature, precipitation, and wind speed or direction can have significant impacts on weather dependent businesses.

For example, other utilities have found just one half a degree improvement in temperature accuracy can save them more than $500,000 a day through improved demand forecasts. A storm’s destructive force also can vary widely depending on altitude, topography, or even human development meaning damage in one area may be completely different in another area ten miles away.

The historical weather data Central Hudson needed was not available in much of its service territory, including some of the more topographically diverse locations – which is where the utility needed it most.

As a utility that spans from the Hudson River Valley up to the Catskill Mountains and back down through the Poughkeepsie area of Upstate New York, there are a multitude of factors that affect the weather in its region. Extremely different weather situations could be present within a general forecast area of 10 miles.

“In 2011 we experienced an ice storm that severely impacted customers in three towns in Northern Dutchess County for several days, but because of temperature differences, customers as few as five miles away experienced only rain and saw virtually no interruption of service,” said Tim Hayes, T&D Operations Services &

Emergency Response at Central Hudson.

Because of its diverse territory, if the NWS were to show Albany experiencing snow, icy rain could be affecting another area in Central Hudson’s service territory simultaneously and deliver a very different impact on the utility’s assets.

Temperature, precipitation – including the percent likelihood, type, and amount – wind speed and direction, sunlight exposure, and possibly the occurrence of hazardous weather conditions can vary across locations, even those in close proximity.

Central Hudson needed accurate, hyper-local weather data for each unique weather profile in its service area in order to develop an outage prediction model that would be accurate and effective.

Hyper-local forecasts, which are point-based, are for a specific latitude and longitude, and time.

These forecasts are made possible through local weather stations.

With accurate weather information, Central Hudson would be able to identify areas within its service territory where outages were more likely to occur, and target those areas with preemptive measures.


In order to obtain the localized weather data it needed, Central Hudson installed twenty-four weather stations across the Hudson River Valley to deliver highly accurate, location-specific weather data.

With access to hyper-local weather data, Central Hudson began to build a more reliable and effective historical weather database.

Precise weather conditions are now reported with real-time notifications, allowing Central Hudson to be better aware of weather influences on its service.

Having hyper-local forecasts also provides Central Hudson with the opportunity to review location-specific data when multiple outages occur.

Using this data, the utility can compare the previous day’s outages with the previous day’s weather for a specific area of its territory.

This helps Central Hudson better determine what type of weather its energy systems withstood, and in what areas. Moving forward, Central Hudson hopes to integrate this weather data into its outage history to develop an outage management forecasting system.

Sunny Skies Ahead

Now Central Hudson has the hyper-local weather data it needs to create an outage prediction model. Important weather conditions are observed with pinpoint accuracy rather than generalized data.

Looking to its future, Central Hudson sees combining historical outage and weather data to inform tree-trimming efforts, which is one of the utility’s biggest concerns. For example, if a certain area in its service territory is repeatedly hit by a severe storm in mid-July, the utility can make sure it trims trees in those neighborhoods, keeping power lines free of falling branches.

Central Hudson also could use the data to see which parts of its territory require more workers on call when a storm is coming, send earlier warnings out to customers about expected severe weather, or detect areas where underground wiring would be beneficial.

Moving forward, the variety or severity of wind and precipitation will be correlated with the previous number of assets affected in storms to develop a model that predicts the number of outages.

These predictions will change the way Central Hudson reacts to severe weather and will transform its outage management.

The Big Weather Picture

While outage prevention is a major benefit for Central Hudson, other utilities found hyper-local weather stations to have additional valuable uses, such as aiding in the reduction of energy consumption.

Energy is an integral part of our modern world, and will only continue to be as more infrastructure comes on the grid.

Accurate data from hyper-local forecasts can have a huge impact on building and facility energy usage.

In the U.S., residential and commercial buildings account for 39 percent of energy demand.

Up to one third of that usage is the result of heating and cooling. This number is highly dependent on weather – and could be lowered with hyper-local forecasts.

Weather stations make guesswork a thing of the past. Utilities can effectively determine how much energy they need to generate and purchase for any given day thanks to the specific temperature data that is readily available.

This leaves little room for error and saves the utility money, while providing customers with the best possible service.

Hyper-local weather stations have transformed Central Hudson from a utility that reacts to bad weather, into a utility that will be able to predict the impact a storm will have on its assets – before the storm even forms.

The utility’s ability to ensure power is flowing to its diverse territory has improved significantly. With localized weather data, Central Hudson no longer needs to fear its customers will be left in the dark.


Jim Foerster is director of Product Management, Weather, at Schneider Electric.

No posts to display