Turbulent Kinetic Energy (TKE) evolution and spin-up time

[Arty]

Hello,

I read on the forum about spin-up time that WRF is quite rapid to stabilize : for mesoscale, 12h is commonly advised. My case runs nested domains (7km and 2.333 km) at time-step for d01 at 40 seconds (on climate modeling time-scale).

I've been advised to look at TKE in the first (few) month(s) of my simulation to determine ideal spin-up time. I wonder however how to proceed using U10 and V10 (and other Us and Vs at lower pressure levels for example).
My research weren't successful on that matter. I looked at Skamarock et al., 2004 (mentionned by Kelly somewhere on the forum) to plot log-log TKE vs. wavelength but it seems to focus on free troposphere and above ; hence I'm not quite sure that I can use it for surface winds.

I've been told to compute TKE as (u'² + v'²), with u' = u - u_mean(time) over a given period (respectively for v') ; then summing/avering TKE over all domain. I'm also told to expect increasing TKE over time at the beginning of the run until it stabilizes, showing that model's spinned-up : but that is not what I observe with the quick script I wrote to do so.

Is there anyone here who'd be nice enough to help and provide some hints or methodological references ? Do I miss something ?

Thanks for your help

 

[Ming Chen]

Hi Arty,
I am not quite sure what your research is focused on. When we apply WRF for weather simulation, we usually recommend a spin-up period of 9-12 hour.

Here is a paper that explores optimal spin-up time. Please take a looka nd hope it is helpful for you.

Developing spin-up time framework for WRF extreme precipitation simulations

Despite the wide application of the Weather Research and Forecasting (WRF) model in extreme precipitation simulations, there is a lack of consensus an…

www.sciencedirect.com

 

[Arty]

Hi Arty,
I am not quite sure what your research is focused on. When we apply WRF for weather simulation, we usually recommend a spin-up period of 9-12 hour.

Here is a paper that explores optimal spin-up time. Please take a look and hope it is helpful for you.

Developing spin-up time framework for WRF extreme precipitation simulations

Despite the wide application of the Weather Research and Forecasting (WRF) model in extreme precipitation simulations, there is a lack of consensus an…

www.sciencedirect.com

Hi Ms. Chen,

My bad, I should have described a little bit more what my research is focused on. I'm working on dynamical downscaling of central South Pacific island of Tahiti within 2 nested domains forced by previous WRF downscaling, itself based on CMIP5 MMM data ; resolutions are set up as follows (my domains) :

Original GCM data = 105 km
d00 = 21 km (wrfout* I'm using to force my own domains below) (6-hourly outputs) [Dutheil, 2019]
d01 = 7 km (1-hourly output, 40 seconds time-step)
d02 = 2.333 km (1-hourly output, 13+1/3 seconds time-step)

I'm running climate experiments over 25 years (1991-2016). I wonder about spin-up as I read different approaches from different sources : 12 hours, 1 month, 1 year, 2 years... It is not clear how the kind of experiment affect the (required) spin-up time. Tahiti is a really small island and I'm not quite sure how the surface (eg water content) will affect this spin-up time.

I'll have a look at the paper you provided though ; in the meantime if you have any other insight based on the aforementioned details, I'd be be glad to ear you out.