Impact of horizontal diffusion, radiation and cloudiness parameterization schemes on fog forecasting in valleys

Abstract

Fog and low stratus forecasting experiments have been carried out with the numerical weather prediction model ALADIN on a case of long lasting fog. The model has been used with different radiation, cloud diagnosing and horizontal diffusion schemes, different representation of orography, increased vertical resolution and with or without prognostic condensates and turbulent kinetic energy (TKE). Some of the numerical set-ups are able to reproduce the fog (low stratus) field as seen in the satellite images as well as the measured 2m temperature and relative humidity diurnal cycles. The results show that cloud diagnosing schemes and overlap assumptions play a more important role than a more sophisticated radiation scheme, or introduction of prognostic cloud water, ice, rain, snow or TKE. More realistic orography representation and a more physical horizontal diffusion scheme significantly improve the modelled low stratus and 2m temperature in the areas with variable orography.

Appendix

The cloud diagnosing scheme that was operational in 2004 diagnoses cloudiness using the formula

$$ N_{\text{S}} = \alpha \sqrt {{\frac{{q_{\text{cs}} }}{{q_{\text{sat}} }}}} $$

(1)

where α = 0.7, q _sat is saturated specific humidity and q _cs is condensed stratiform specific humidity computed as

$$ {q_{\text{cs}} = q_{{{\text{c}}\max }} }\left( {1 - {\text{e}}^{{ - \beta {\frac{{q - r_{\text{c}} q_{\text{sat}} }}{{q_{{{\text{c}}\max }} }}}}} } \right) $$

(2)

if and 0 otherwise.In this formula, q _c max is the maximum sustainable water content, β = 0.4, q is specific humidity and r _c is the critical mesh averaged humidity producing a cloud.

When the model uses prognostic cloud water and ice, liquid water and ice specific contents are added to the water vapour specific humidity, so q+q _l+q _i is used instead of q in the equation for q _cs. The alternative equation used here is

$$ {q_{cs} = q_{c\max } }\left( {1 - e^{{ - \beta {\frac{{q + q_{l} + q_{i} - r_{c} q_{sat} }}{{q_{c\max } }}}}} } \right), $$

(3)

and it accounts for the presence of the prognostic condensate species.