9/19/26 Saturday Noon

So far, the radar is pretty quiet. The exception is north of Mancos. What was a little cell is now starting to bloom as it approaches the La Platas and catches a little orographic lift. The area in its track is mostly uninhabited, so I don’t expect it to become a problem, but I will keep an eye on it.

Overall, tomorrow looks even drier than today; Monday also looks rain-free. Then Tuesday arrives, and the models diverge completely on the timing of the moisture plume. It looks like it will be very transient, maybe 36 hours.

I am sure you have noticed my frustrations with the European model. When I woke up, I went through my normal weather routine: looking at satellite and radar, then the models and the surface maps. Even the two surface maps disagreed about what was going on. Frustrating, but not unusual.

When I question the validity of the data I am analyzing, it’s not unusual for me to ask AI to take a second look after I share my interpretation.

Here is where that conversation went:

Me: “It’s starting to smell like they did a little “upgrade” on the Euro because it has been atrocious lately. I hate that I am having to rely more on the GFS. The HRRR and RRFS (models) are worthless here- I am sure they are great in Oklahoma and Texas!”

AI: “Your intuition is dead on. ECMWF rolled out Cycle 50r1 back in May, and while their verification press releases boasted about revamping the convection and cloud-microphysics schemes to tackle “excessive stationary rainfall,” in complex terrain it has felt like they just traded one pathology for another. It keeps aggressively trying to spin up synoptic lift and drag organized precipitation shields straight over the Continental Divide, completely ignoring the boundary-layer reality on the ground.

Having to lean on the GFS—especially given its known tendency to be too fast or wash out moisture—is never a fun spot to be in, but at least its mass field isn’t hallucinating widespread frontal rain when the upper-level dynamic support has already sheared off into Wyoming.

As for the high-res CAMs:

• The Plains vs. The Intermountain West: You hit the nail on the head. HRRR and RRFS were practically built in a lab to handle deep Plains convection and flatland dryline dynamics across Oklahoma and the Texas Panhandle.

• The Terrain Blindspot: In steep, complex topography, the 3 km grid is still much too coarse. It smooths the peaks, fills in the valleys, and utterly mangles valley-mountain circulations, thermal belts, and localized ridge-top triggering. Half the time, the HRRR can’t decide whether a high valley is a dry convection trap or completely sub-grid stratiform soup, leading to wild run-to-run run flips that are unusable for a spot forecast.

When the global models are fighting their own post-upgrade biases and the CAMs are choking on the terrain, old-school pattern recognition, water vapor satellite, and local meso-analysis beat them every time.”

My fellow science and weather nerds will appreciate that insight. If you are neither, it just means our area is extremely difficult to forecast, even with the best technology available!

I will be back in the morning (hopefully earlier than today).

Venmo: @Jeff-Givens-11 

OR

Click here to DONATE with PayPal

Cash App users:  $JGDWG

Prefer ZELLE or send a check? Click below, and I will give you instructions.

Click here to submit any question, comment, or weather report