Abstract:
Fog, especially extremely dense fog, has adverse effects on transportation. This paper utilizes the fog droplet spectrum data measured using the FM-100 fog drop spectrometer at the Shouxian National Climate Observatory in January 2019 and the contemporary conventional meteorological observation data to investigate the microphysical characteristics of fog with different intensities. Based on the analysis of the relationships between the visibility (
V) and liquid water content (
L) and the number concentration (
N) of fog droplets and RH (relative humidity), various visibility parameterization schemes were established. The results showed that: (1) With the increase of fog intensity, the water content in fog increased significantly, with average values of 0.003 g m
−3, 0.01 g m
−3, and 0.09 g m
−3 during the periods of fog, dense fog, and extremely dense fog, respectively. When the
L was greater than 0.02 g m
−3, the proportion of extremely dense fog reached 95%. (2) The
N and droplet size increased with the increase in fog intensity. From fog to dense fog, the
N increased significantly by 67%, while from dense fog to extremely dense fog, the droplet size increased significantly, and the average diameter (
D) and effective radius (
Re) increased by 62% and 135%, respectively. When the
Re was greater than 4.7 μm, the proportion of extremely dense fog reached 95%. (3) All spectra distributions of the droplet number concentration for fog, dense fog, and extremely dense fog exhibited a bimodal structure with major peaks close to the end of the small particles. The spectrum type of the extremely dense fog was a Deirmendjian distribution, while that of the dense fog and fog was a Junge distribution. The fog water mass concentration spectrum was characterized by multipeaks for extremely dense fog with maximum peak appearing at 21.5 μm, while the dense fog and fog exhibited a bimodal distribution and single peak type, respectively, with maximum peak at 5 μm. (4) Both
L and
N were inversely correlated with visibility, with
L showing the highest correlation coefficient with visibility. Four kinds of visibility parameterization schemes were established using the full sample and segmented method. Test results indicated that the visibility subsection fitting scheme based on
L was the best.