One exciting thing about EUREC4A is that it puts us in a position to link to really exciting developments in modeling the atmosphere with Global Storm Resolving Models, i.e., models with grid spacings on scales of less than 5 km and on which (in the atmosphere) non-hydrostatic motions become important.
Recently (tonight actually) we have been putting the final touches on a draft manuscript describing the first ever intercomparison of such models. The models are fantastic because: (i) they look cool; (ii) they use physics, rather than statistical hypotheses (parameterization) to manage the convective energy transport in the vertical; (iii) they link much more naturally to the observations (and impacts); and (iv) out of the box (without much tuning) they give a pretty good approximation to the atmosphere, especially when it comes to precipitation.
Two things relevant about this effort for today’s post. The first is the role of shallow precipitation. Attached is a zonal average of thirty days of simulated precipitation (starting after a ten day spinup, where all models started from the same initial data and use the same SST). What you see is that there are two places of discrepancy, on the poleward side of the winter (SH) storm tracks, and just south of the Equator. It looks like these models have a double ITCZ problem. Or maybe the observations are missing precipitation from shallow convective systems. The consistency of the models, in their global numbers, and in their features make the hypothesis that it is the data that is off competitive with the idea that it is the models that are off.
The second plot shows an example of strong precipitation event from a very shallow cloud that passed over the site earlier today (okay, yesterday, when it was still Day 7). The maximum k-band reflectivity was 32 dBZ, which is actually not that large, but the cloud had its top at only 2200 m! So far this season we haven’t seen very much deeper clouds, but by the time they reach 4 km it is easy for an isolated clouds to have rain rates over 100 mm/h.
I said two things are relevant about today’s post. The other thing I wanted to point out is how all the simulations look quite realistic, but quantiatively they can differ substantially in their representation of clouds, with net absorbed shortwave radiation across the models straddling the observations by ±15 W/nm^2. I am convinced that this new type of simulation (eg., the same video I gave in Day 2’s post: https://youtu.be/ji4nno-fsvw) will have a huge effect on climate science, and is the future of climate modelling; but I am also equally convinced that it introduces a new quality of problems we need to begin working on… especially if we want to couple to the ocean and predict the fate of scum; but not tonight, now its bedtime.
PS if Pier Siebesma reads this far maybe he can post a link to the simulations they have been doing, as I saw an awesome video he presented in a big university lecture (where he was beautifully garmented), or agree to tell us more about them in a future post.