Some Weather Models May Be Overstating Cooling Demand

When analyzing weather model guidance, it is always important to take into account what the biases of certain weather models may be. With this understanding, energy traders may be able to better process the raw output released from weather models. Each week, we release a Model Verification Update on Wednesdays that outlines what are the main biases of weather models are how that may impact current forecasts. 

This week, we saw that the CMC ensembles from 2 weeks ago performed exceptionally better than the GFS ensembles for the same time. This may be due to the fact that the GFS ensembles have more of a bias to warm lower levels of the atmosphere in the long-range as compared to the CMC ensembles, as can be seen below. 

Meanwhile, we are seeing that across the Northern Hemisphere operational weather model guidance also seems to have a slight warm bias. The operational GFS is too warm from the 500mb layer all the way down to the surface, and appears to have the warmest bias at the surface. The operational ECMWF has the warmest bias around the 500-700mb layer and is performing better with surface temperatures, not having much of a warm bias. Finally, the operational CMC across the Northern Hemisphere does not appear to have a bias below 500mb, unlike the other models.

These trends are consistent with a lot of what we are seeing in the long-range as well. If these trends are true, then we need to account for a warm bias this summer among a number of the major models, especially the American GFS model at the surface. The result of this would be forecasts that do not result in quite as much cooling demand as they would seem to. Of course, a lot of these trends are minimal (a degree or two here or there does not significantly diminish demand on the aggregate). However, we will need to continue to track this bias for our major weather models to be slightly too warm across the Northern Hemisphere, and especially in the Pacific North America (PNA) region. 

Natural gas prices have rallied significantly the past few weeks, and the most recent leg higher appears at least in part due to significantly warmer forecasts in the medium and long-term. We saw CPC guidance over the last two days trend significantly warmer in the 8-14 Day timeframe, as seen below. Their new update should be expected by around 3:30 PM.

While there is no doubt that over the past 48 hours weather models have trended warmer in this direction, this verification analysis indicates there is a chance that as these forecasts progress into the short-term, they may cool off somewhat to account for model biases. If in a week the heat that you see the weatherman talking about now turns out to not be quite as strong as expected, this model warm bias could be at play.