Why Weather Source?

Weather Source is Big Weather & Analytics

Weather Source, LLC was founded in 2004 to solve the challenge of providing industry leading, hyper-local past, present and forecast weather data that seamlessly integrates with business data and empowers our clients to optimize for weather and significantly improve profits.

Our science team is a unique fusion of crazy smart meteorologists, climate scientists, and data & analytics scientists who are passionate about delivering game-changing Big Weather & Analytics solutions via the Weather Source OnPoint Weather Platform.

How big is Big Weather?  OnPoint Weather supports nearly 2 million global points and provides a continuum of past, present and forecast weather on hourly and daily increments for each point back to January 2000. In total this equates to over 12 trillion accurate weather observations.

What Makes OnPoint Weather Special?

It all starts with understanding what OnPoint Weather is engineered to do, which is to provide businesses with hyper-local, accurate, and reliable weather information designed for business use.  Sure, there are plenty of sources of weather information, but only OnPoint Weather is engineered for Big Weather & Analytics.

Data Density, Proximity and Relevance

Back in 2004 Weather Source was supporting clients with weather station data that was cleaned to correct errors and fill gaps.  This proved to be useful for some clients provided their business interests were close enough to the weather station locations.  However, as the distance from the weather stations to the locations of interest increased, the utility of the clean weather data dropped due to the lack of correlation caused by the increase distance.  This challenge led to the development of OnPoint Weather, which due to its hyper-local data density ensured that every business location could be supported with relevant weather information. Comparatively there are less than 10,000 viable weather stations around the global, whereas OnPoint Weather support nearly 2 million points (see: Why Resolution Matters).

Data Quality and Accuracy

Every data source, whether it be financial data, demographics, or weather data is prone to errors, which can be extremely costly when these data are used for the purpose of improving business and profits. In the world of weather data, errors are a daily occurrence as data is generated and transmitted from many thousands of weather sensors that monitor weather conditions across the globe.  Left uncorrected, these errors would corrupt and degrade the utility of important weather information.  Leveraging the deep history Weather Source has with data cleaning, we employ data cleaning methods throughout all areas of the OnPoint Weather Production System.  Weather data from weather stations, satellites, weather analysis models, radar and other sources are challenged, flagged and corrected. The resulting OnPoint Weather product is clean, accurate, reliable and robust for business use.

Data Consistency

There are two key areas where consistency is important. First, when using past weather to build statistical relationships that will later be driven by forecast weather, it is critical that the past weather and forecast weather be statistically consistent.  This is assumed by most to be the norm, but unfortunately, this is typically not the case.  A typical situation is where one builds a regression model using station data, then drives the regression model using a forecast. However, it is very likely that the forecast is operating at a bias compared to the station data.  This is a common, yet typically unknown issue that is the result of weather stations and weather forecasts operating as independent systems.  The OnPoint System is built on a common framework where past weather and forecast weather are inherently statistically consistent.

The second area of consistency is related to the availability of weather parameters. An unfortunate truth is there is a significant amount of variation regarding what weather is observed at weather stations.  For example, in the US, it is common for only a subset of weather stations to record snowfall observations.  This can lead to issues where analytics developed for one weather type can’t be implement at all weather station sites since the required weather parameters aren’t available at all sites. This issue is even worse when one attempts to implement analytics to international locations.  OnPoint Weather solves this problem by supporting a broad and consistent set of weather parameters for each of the nearly 2 million supported points.


Climatology is essentially the statistics of weather over time.  Climatology (or ‘climo’ for short) can be useful in two ways.  The first allows one to compare weather to what should be ‘normal’ weather for a place and time. Weather parameters such as temperature, humidity, wind and precipitation can be differenced with climo to get a ‘departure from normal’.  Often it is the departure from normal that has the biggest influence on consumers.  For example, a September temperature of 50F may be 20 degrees below normal and inspire consumers to think ‘hunker down’, whereas 50F in February may be 20 degrees above normal and might inspire consumers to think ‘get outside’.  Both temperature are 50F degrees, but one is below normal and the other above, and this causes a different consumer response.

The second use of climatology is as a long-range forecast.  Most dependable forecasts can only provide a reliable forward view of several days. A few more advanced forecasts can look forward a few weeks with some skill.  Beyond that, longer range forecasts tend to be sketchy.  A helpful alternative is to use climatology to get a solid estimate of what to weather to expect for any location and any point in time.  Weather Source’s OnPoint Climatology supports nearly two million points globally with climatology information on hourly and daily increments.  The OnPoint Climatology provides information on weather parameter means and standard deviations. In addition, OnPoint Climatology provides valuable ‘frequency of occurrence’ information for parameters like precipitation and snowfall. The frequency of occurrence provides insight into how often certain precipitation amounts occur. For example, how often does snowfall in the range of 1.0 to 2.5 inches occur.

Comparative Matrix

Feature OnPoint® Weather Government Weather Forecast Companies
Quality and accuracy design designed for analytics

Weather Source has been the leader in clean weather data algorithms for over a decade.

OnPoint Weather YES Government Weather No, severely limited Forecast Companies Slightly better than government data, but not designed for this.
Past, present and forecast weather that is statistically consistent

The elephant in the room most don’t realize is an issue.

OnPoint Weather YES Government Weather No Forecast Companies No
Climatology to provide context of the difference from normal weather and as a long range forecast

Consumer behavior is often driven by departures from normal.

OnPoint Weather YES Government Weather No Forecast Companies No
Consistent weather parameter set across space & time

Allows you to use your analytics everywhere without exceptions.

OnPoint Weather YES Government Weather No Forecast Companies No
Broad set of parameters to support a wide range of business needs OnPoint Weather YES Government Weather No Forecast Companies Slightly better than government data, but not designed for business intelligence.
Fast, on-demand access to targeted weather information OnPoint Weather YES Government Weather No Forecast Companies Limited