NOAA revises heating degree day data back to Jan.

By Joe Silha

NEW YORK, May 14 (Reuters) - The National Oceanic and Atmospheric Administration said this week it had revised heating degree day calculations going back to January due to data problems that sometimes overstated total heating degree days (HDDs), particularly in some Midwestern states.

"This was a data problem. A computer analysis program was putting out corrupted temperature grids, which are the basis for degree day calculations for states and regions," a source at the Climate Prediction Center said, adding HDD data for specific U.S. cities was not affected.

He said the problem was first discovered in late April, when the biggest degree day discrepancies were reported, particularly in states like Ohio, Indiana and Illinois.

At that time, the Climate Prediction Center realized it was calculating more heating degree days than actually existed, which would lead energy industry computer models that analyze supply and demand to overestimate heating needs.

A heating degree day is the measure of how far the average outside temperature falls below 65 degrees Fahrenheit and is used to reflect demand for energy to heat homes and businesses.

"This was the data I plugged into my model to give me my EIA (Energy Information Administration inventory) projection. I thought my model was way off, but with revised heating degree day numbers, it looks a lot better," said Craig Solberg, senior weather analyst at Forecast Trading, a commodity trading advisory firm in New York.

While overstating HDDs should prompt computers to estimate that more natural gas was being burned each week for heating, a review of Reuters weekly poll data for the period, which surveys industry expectations for weekly inventory changes, showed no consistent pattern between the average estimate and the actual storage draw or build.

The Climate Prediction Center is part of NOAA's National Weather service and is known for its climate forecasts, but it also collects and produces daily, weekly and monthly data and maps on precipitation, temperatures and degree days useful to meteorologists as well as energy and commodity traders.