Evaluating the impacts of cumulus, land surface and ocean surface schemes on summertime rainfall simulations over East-to-southeast Asia and the western north Pacific by RegCM4

Abstract

This study evaluates the sensitivity of summertime rainfall simulations over East-to-southeast Asia and the western north Pacific in the regional climate model version 4 (RegCM4) to cumulus (including Grell with Arakawa–Schubert type closure, Grell with Fritsch–Chappell type closure, and Emanuel), land surface (Biosphere–atmosphere transfer scheme or BATS, and the community land model or CLM) and ocean surface (referred to as Zeng1, Zeng2 and BATS1e in the model) schemes by running the model with different combinations of these parameterization packages. For each of these experiments, ensemble integration of the model was carried out in the extended boreal summer of May–October from 1998 to 2007. The simulated spatial distribution, intensity and inter-annual variation of the precipitation, latent heat flux, position of the subtropical high and tropical cyclone genesis patterns from these numerical experiments were analyzed. Examinations show that the combination of Emanuel, CLM and Zeng2 (E-C-Z2) yields the best overall results, consistent with the fact that physical mechanisms considered in E-C-Z2 tend to be more comprehensive in comparison with the others. Additionally, the rainfall quantity is found very sensitive to sea surface roughness length, and the reduction of the roughness length constant (from 2 × 10⁻⁴ to 5 × 10⁻⁵ m) in our modified BATS1e mitigates the drastic overestimation of latent heat flux and rainfall, and is therefore preferable to the default value for simulations in the western north Pacific region in RegCM4.

Appendix: Ocean surface flux schemes in RegCM4

BATS1e scheme uses constant ocean surface roughness length. Roughness length for wind (z_0m), temperature (z_0h) and humidity (z_0q) are all set to be the constant value of 2 × 10⁻⁴ m. Then neutral exchange coefficient for momentum C _MN is calculated through

$$C_{\text{MN}} = \left( {\frac{k}{{\log (z/z_{{0{\text{m}}}} )}}} \right)^{2}$$

(1)

In this equation, k is von Kármán constant and equals 0.4, z is the height of lowest model level. Exchange coefficient for momentum (C _M) depends on the bulk Richardson number Ri _B

$$C_{\text{M}} = C_{\text{MN}} (1 + 24.5\sqrt { - C_{\text{MN}} Ri_{\text{B}} } ),\;{\text{if}}\;Ri_{\text{B}} < 0$$

(2)

$$C_{\text{M}} = \frac{{C_{\text{MN}} }}{{1 + 11.5Ri_{\text{B}} }},\;{\text{if}}\;Ri_{\text{B}} > 0$$

(3)

Exchange coefficients for heat (C _H) and moisture (C _E) are assumed to be the same as C _M. Then momentum, sensible heat and latent heat fluxes (τ, H and LE) are given by

$$\tau = \rho C_{\text{M}} u^{2}$$

(4)

$$H = - \rho c_{p} C_{\text{H}} u\varDelta \theta$$

(5)

$$LE = - L_{e} \rho C_{\text{E}} u\varDelta q$$

(6)

Here, u is the wind speed at lowest model level, Δθ(Δq) is the difference of potential temperature (specific humidity) between the lowest model level and the Earth’s surface, ρ is air density, c _p is specific heat of air, and L _e is latent heat of vaporization.

In Zeng1 and Zeng2 schemes, z_0m, z_0h and z_0q are dependent on friction velocity u _*. The only difference between these two schemes is the calculation of z_0m:

$$z_{{ 0 {\text{m}}}} = 0.0065\frac{{u_{*}^{2} }}{g}_{ } \left( {\text{for Zeng1}} \right)$$

(7)

$$z_{{ 0 {\text{m}}}} = 0.013\frac{{u_{*}^{2} }}{g} + 0.11\frac{\nu }{{u_{*} }}_{ } \left( {\text{for Zeng2}} \right)$$

(8)

In both schemes, z_0h and z_0q are calculated as

$$z_{{ 0 {\text{h}}}} = z_{{ 0 {\text{q}}}} = \frac{{z_{{ 0 {\text{m}}}} }}{{\exp (2.67(\frac{{u_{*} z_{{ 0 {\text{m}}}} }}{\nu })^{0.25} - 2.57)}}$$

(9)

Here ν is the kinematic viscosity of air. The stability parameter ζ is related to Ri _B as:

$$\zeta = Ri_{\text{B}} \frac{{\log (z/z_{{0{\text{m}}}} )}}{{1 - 5Ri_{\text{B}} }},{\text{ if}}\;Ri_{\text{B}} > 0$$

(10)

$$\zeta = Ri_{\text{B}} \log (z/z_{{ 0 {\text{m}}}} ),{\text{ if}}\;Ri_{\text{B}} < 0$$

(11)

Finally, τ, H and LE are given by:

$$\tau = \rho \frac{{k^{2} }}{{[\ln (z/z_{{0{\text{m}}}} ) - \psi_{\text{M}} (\zeta ) + \psi_{\text{M}} (\zeta z_{{0{\text{m}}}} /z)]^{2} }}u_{{}}^{2}$$

(12)

$$H = - \rho c_{\text{p}} \frac{{k^{2} }}{{[\ln (z/z_{{0{\text{m}}}} ) - \psi_{\text{M}} (\zeta ) + \psi_{\text{M}} (\zeta z_{{0{\text{m}}}} /z)][\ln (z/z_{{0{\text{h}}}} ) - \psi_{\text{H}} (\zeta ) + \psi_{\text{H}} (\zeta z_{{0{\text{h}}}} /z)]}}u\varDelta \theta$$

(13)

$$LE = - \rho L_{e} \frac{{k^{2} }}{{[\ln (z/z_{{0{\text{m}}}} ) - \psi_{\text{M}} (\zeta ) + \psi_{\text{M}} (\zeta z_{{0{\text{m}}}} /z)][\ln (z/z_{{0{\text{q}}}} ) - \psi_{\text{H}} (\zeta ) + \psi_{\text{H}} (\zeta z_{{0{\text{q}}}} /z)]}}u\varDelta q$$

(14)

Here $\psi_{\text{M}}$ and $\psi_{\text{H}}$ are atmospheric stability functions proposed by Dyer (1974):

$$\psi_{\text{M}} (\zeta ) = \ln [(\frac{1 + x}{2})^{2} (\frac{{1 + x^{2} }}{2})] - 2\arctan (x) + \frac{\pi }{2}$$

(15)

$$\psi_{\text{H}} (\zeta ) = 2\ln (\frac{1 + y}{2})$$

(16)

with

$$x = (1 - 16\zeta )^{1/4}$$

(17)

$$y = (1 - 16\zeta )^{1/2}$$

(18)