Cases in London:

Default subroutines outputs high roughness length (z0) for grids with low frontal area index.

• Issue 1: $\hat{\psi}_m$ is set to zero at the highest level (observation) and that has greater impact in the calculation of $\hat{\psi}_m$ for low rise neighbourhoods.

  • Proposed change: Update the code to include a blending height ($Z_w$), which represents the height up to which the roughness sublayer (RSL) exerts influence -> $\hat{\psi}_m(Z_r) = 0$. Eq. 21 from Grimmond (1999, JAM)
    $Z_w = z_H + 1.5 (D_x - L_x)$

    To calculate the average interelement spacing in the along wind direction ($D_x$) and the mean breadth of the roughness elements ($L_x$) for a uniform grid with squared buildings, you will need the following information: surface area of the grid ($SurfaceArea$), the number of building ($N_buildings$) and the plan area index $\lambda_P$, where

    $Dx = \sqrt{SurfaceArea / N_{buildings}}$
    $Lx = \sqrt{D_x^2 * PAI}$

• Issue 2: In the default code $\beta = u^* / U_H$ is parametrised using $\beta_N = 0.4$

  • Proposed change (Initial): Update $\beta_N$ according to plan area index ($\lambda_P$) - results at the top right figure.

• Proposed change (Current): Include parameterisation of $\beta_N$ according to frontal area index ($\lambda_F$) and $H' = H_{std} / H_{avg}$, where $H_{std}$ and $H_{avg}$ are standard deviation and average buildings height. $\beta_N$ is parameterised using UrbanTALES dataset for realistic neighbourhood cases - result below.