Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Upper boundary condition: "Constant pressure"? What happens at the model top?


New member

I am new to WRF and since I will have to do LES in the future, first of all I try to understand how the model works. My latest simulation was a 48-hours idealized LES with almost default settings (only surface layer physics of Jimenez et al., 2012), but with an initial zonal wind of u = 10 m/s on every level added in the input sounding file. I also turned off the TKE heat flux (set to zero), because I wanted to see if that causes the wind speed to slow down to zero over time due to friction at the lowest levels (which it does not).

Although I read the most recent documentation of the WRF model V4, it is not enough for me to answer the question what exactly drives the model. Why does the wind speed keep relatively stable around 10 m/s at the model top (here 2.000m) and at relatively decent speed in middle levels (see the time average of the vertical cross section of the u-component in figure 1)? Which forcing is at work? I understand that the sounding values are just initial input values and not constant input.

My professor believes that there must be some kind of pressure gradient at the model top, but in the documentation, they state that there is „constant pressure level at top boundary along a material surface“. Could this statement be misleading? If it was constant, the pressure gradient at the highest model level would have to be zero which would slow down the wind to zero over time due to friction, at least from my understanding.
But I see that the horizontal wind components (spatially averaged) are oscillating around a mean value (see attached figure 2) and that the pressure perturbation term at the highest grid cell is changing over time, even though the change is very small. I calculated the pressure gradient terms (dp/dx, dp/dy) from the pressure perturbation array, these values are also very small (on the order of 10^{-4} hPa/km for dp/dy, 10^{-5} for dp/dx) on the highest grid cell, but not zero or some sort of quasi-zero (10^{-15} or so).
I understand that this small gradient should be enough to keep the wind speed around its initial level of 10 m/s at the top boundary grid cell and thus serves as some sort of a model "driver". Is this analysis correct and does the model check somewhere in the background that the originally set wind speeds (u = 10m/s, v = 0 m/s) are values that need to be maintained at the model top and then force the pressure perturbation term to take "suitable" values? If the analysis is wrong, what keeps the wind speed at the almost constant level, or what exactly happens at the model top?

Best regards,


  • Figure 1.png
    Figure 1.png
    17.1 KB · Views: 1
  • Figure 2.png
    Figure 2.png
    53.4 KB · Views: 1
Dear Ming Chen,

thanks for your reply. Please find attached my namelist.input. It should be the default setting except some changes in the time_control and no TKE heat flux.



  • namelist.input
    5 KB · Views: 3