In this paper, we examine reliability of wind-speed forecasts made by The Weather Channel (TWC) and AccuWeather (AWX). Specifically, we analyze n-day-ahead forecasts, where n ranges from 1 to 9 for TWC and from 1 to 14 for AWX. Our dataset includes approximately 370 000 wind-speed forecasts, collected over a one-year period at 51 observation stations across the state of Texas. We segment our analysis by the four wind-power classes used in Texas. In addition to verifying the forecasts made by both providers in an absolute sense, we also make relative comparisons between the forecasting strategies of both providers.

We find that TWC forecasts, while not perfectly calibrated, demonstrate positive skill over the observation window. However, THC short-term forecasts exhibit some bias (they over-forecast wind speed), especially within the lower wind-power class regions.

AWX forecasts exhibit positive skill the first six days of the forecasting period, but are negative thereafter. By the end of the forecast period AMX's wind speed forecasts are independent of the observations.

The differences in skill observed between the two providers are driven by differences in their forecasting procedures. For example, AWX tends to concentrate its wind forecasts around 6-10 mph and 14-16 mph, with a noticeable gap in the likelihood of forecasting wind speeds around 12 mph. This behavior is difficult to explain and detracts from the quality of AWX wind-speed forecasts. TWC exhibits no such irregularities.

In sum, both providers provide skilled forecasts in the near term. Overall, however, TWC's forecasts are more skillful than those of AWX, especially for long-term forecasts.