Question of the week: Will we get recovery?

Last week, the overarching question plaguing the SFE daily was a fairly simple one: Will we get recovery? Due to low moisture across the Gulf of Mexico from a front that swept through during the weekend of May 7-8, instability was hard to come by last week. Since the necessary requirements for severe convection are lift, shear, and instability, this question is extremely relevant to our daily activities. On the other hand, it gave our participants the chance to deal with a forecast scenario that is fairly common, if unusual for May: how do we forecast when we have high shear, but relatively low instability? Last week, CAPE values were typically around 1500J/kg or less - sufficient for severe storms, but low enough that the questions of whether that CAPE would be realized and whether significant severe weather would occur.

Climatologically, May is favorable for severe weather across the Central Plains because of an overlap in CAPE values, which increase from winter to summer, and shear values, which decrease from winter to summer. Therefore, the spring maxima in severe weather is expected due to the potential for overlap between high values of these ingredients. The sounding climatology of Norman, Oklahoma for MLCAPE and 0-6km shear shows this nicely, with the upper quartile of MLCAPE values on 11 May at 00Z exceeding 807.13 J/kg and the upper quartile of 0-6km shear on 11 May at 00Z exceeding 44.44 kts:

However, instability requires both heat and moisture, and moisture was slow to recover after the previous weekend's cold front. Dewpoints at Dallas on Monday morning were 5-10 degrees below normal at 850 mb, precluding mixing down of moisture. Monday also held the issue of early morning clouds and precipitation, which would be common throughout the week. Overall, the models were quite keen on recovering the moisture faster than actuality. It felt like every day last week, we were saying "the deep moisture might start to come back tomorrow." While we anticipated that moisture in the Gulf would be a problem for the week of 9 May - 13 May, on Monday things were already recovering more slowly than we thought. Even when we got sufficient moisture to provide some instability Monday, our thinking was that the highest CAPE would be displaced from the strongest shear. As we saw, sufficient moisture existed on 9 May to produce significant tornadoes, 

Tuesday produced a similar problem, although a pocket of moisture was evident at 850 mb across Arkansas and Missouri. Moisture at 850 mb is something we look at daily in the SFE, to determine how deep any moisture we might be seeing at the surface is. However, the moisture we saw on Tuesday being advected northward from the Gulf of Mexico was flowing right into ongoing convection in southern Illinois and Indiana. This morning convection ended up overturning the atmosphere enough that on Tuesday, the severe weather occurred almost exclusively south of the area of morning convection. Also on Tuesday, we finally saw some high precipitable water values (~1.75") coming back into the Gulf. 

On Wednesday, the recovery we had to think of was across Oklahoma, in the wake of a cold pool created by storms in Texas. The outflow from these storms caused a drop in dewpoints across Oklahoma, as was evident from several mesonet sites:

Temperature (red) and dewpoint (green) for Norman, OK. Note the nearly twenty degree drop around midnight on the 11th!

Precipitable water values also dropped in the wake of this system - the high values of 1.5"-2" were located in far south Texas. Missouri was another forecast area of interest on Wednesday - while morning convection was ongoing, models consistently indicated that there would be recovery behind that system. However, this conceptual model ended up failing, and the ongoing convection intensified and produced the severe reports across Missouri and into Illinois on Wednesday. 

MPAS, a convection-allowing model that runs out 5 days, recovered instability quickly behind the ongoing convection Wednesday, much moreso than what actually occurred. MPAS was far from the only model guilty of a quick recovery on Wednesday - all CAMs in the experiment that extended beyond a 36 hour forecast showed some signs of recovery for Wednesday until the runs initialized on the actual day of the event.

We began to see some larger-scale recovery on Thursday, but overall our shear was not very strong. Ongoing convection was once more a factor in our area of interest, which was over Texas. This day was the least active of the week, and we were aware that the outflow from the morning storms would likely prevent any further storms. Despite the beginning of the recovered dewpoints, the morning convection did not significantly intensify, and only a few hail reports occurred in our area of interest. 

Luckily, we were able to look at a 5-day convection allowing ensemble (MPAS), and find some sort of moisture coming into play by the following Monday (the day of this writing). Whether that moisture and subsequent forecast scenario played out or not is the subject of tomorrow's blog. 

Friday was an interesting case, in that a narrow tongue of instability was available along the front. This can be seen in the RUC reanalysis of surface-based CAPE:

While models captured this area of CAPE, they suggested that recovery would occur in Missouri behind the ongoing morning convection. If this sounds familiar, it's because we also contemplated recovery behind morning convection in Missouri on Wednesday! As was the case on Wednesday, the severe weather didn't precisely adhere to our mental models. On Friday, there was no recovery over Missouri, evident from the lack of CAPE there in the previous image. Instead, the axis of greatest severe weather extended from south-central Kansas to the Texas Panhandle. Friday was a day when the CAMs had trouble determining how far the moisture would return up the front - they suggested much more recovery than what occurred. We got burned by this in our forecasts, ending up with high probabilities and an area of significant severe weather in Missouri that did not verify. 

Hail (top) and total severe (bottom) experimental forecasts (left) and practically perfect forecasts (right) for Friday, 13 May 2016

So, what can we learn from all of this? First, that forecasting moisture return can be difficult for both humans and models. Particularly when we're dealing with unseasonably low moisture, determining where the moisture would be sufficient for storms in an area with enough shear for severe storms can be quite difficult. Second, that ongoing convection is a forecast consideration that can severely affect the CAMs. This week had issues with CAMs recovering moisture too quickly, and often this was in the wake of morning convection. Finally, I think this week points to the importance of incorporating observations with the model guidance. While we considered the lack of available moisture each day in our forecasts, perhaps we bought into the CAMs and their fast recovery too much this week. As of Monday, 16 May, the moisture is returning to much of the Gulf, and we're on to new challenges!