We first examine the result of experiment TH10, which represents the simplified synoptic inversion of our studying upslope fog event. The results demonstrate the moistening processes in the valley after the initiation of valley winds and the vertical turbulences are explicitly simulated. Under the topographic uplifting effect, the convection in the TaiwanVVM model develops. The convection could only develop up to 3000 m because of the limitation by the capping inversion, result in the accumulation of liquid cloud water around 2000 to 3000 m (Fig. 8). The convections moisten the upper part of the valley boundary layer, which is the crucial process of the upslope fog development previous studies did not identify. Fig. 9 shows the evolution of moisture on the lower and upper levels of Xitou valley. The resulting moistening of the valley by both the valley winds and moist convections could be found. The valley winds converge while flowing into the V-shaped valley. The supply of water vapor by the valley winds convergence keeps moistening the bottom of the valley. The water vapor mixing ratio in the valley at 1300 m (black box in Fig.9) increase from 8.12 g kg-1 (10 am) to 8.75 g kg
-1 (3 pm). Although the valley winds induced by the heating difference on topography only
exist on the bottom level, the water vapor mixing ratio of the upper level in the valley still
increases gradually (from 6.05 g kg-1 at 10 am to 7.07 g kg-1 at 3 pm). It indicates that the moistening of the upper valley level should come from lower level moisture by vertical turbulence mixing and convections. By consistent valley winds and convections development, the moisture transport upward into the upper level of the boundary layer in the valley. The increased condensation of liquid water leads to the thicker low-level clouds and hence the lowering of the cloud base in the afternoon. Eventually, the cloud base touches the ground and the valley fog forms (Fig. 10).
In the experiment TH10 (CTRL), we simulate the orographic effects on low-level moisture convergence and consequently cloud base lowering and the fog formation. The initiation of valley winds in the morning and the cloud base lowering before the fog formation in the afternoon are both consistent with the observations of our studying fog event. Actually, in this experiment, we found that the observed valley winds are only part of the local circulation in the Xitou valley, which also includes the moist convections triggered by the uplifting effect of topography. By prescribing an initial profile with strong inversion as 50 K km-1, these convections could not penetrate the inversion and make the
moisture confined in the valley to promote the fog formation. Fig.11 demonstrates the schematic of the three-dimensional flow structure of local circulation in the Xitou valley.
The upslope valley winds converge in the valley, twisted up, and initiate convections. The
17
capping inversion confines the convections, and airflows become horizontal after convections developed to the level of inversion. The upslope valley winds observed on the valley surface only reveals part of the whole local circulation. The three-dimensional structure of local circulation over complex topography could be better identified by the high-resolution cloud-resolving simulation.
Since the inversion capping limit convections development and trap the moisture in the valley, it could be a crucial factor to control the duration of the fog. The simulations with varied inversion strength show that the experiment with weaker inversion strength results in a shorter fog duration. For example, the fog is barely formed on the valley surface, and the fog is lasting only 0.67 hours in the experiment which inversion strength is set as weak as 15 K km-1 (TH03 in Fig. 12). The comparison of the liquid water path evolution of experiments (Fig. 13) suggests that liquid water and moisture could not be held in the valley in experiment TH03. The incapability of trapping moisture in the valley in weak inversion simulation could also be found on the water vapor mixing ratio distribution. Fig.
14 shows the comparison of wind fields and water vapor on the valley bottom level (1300
m) and upper level (2000 m) at 3 pm. The moisture and circulation characteristics of experiment TH08 are similar to the results in experiment TH10. The valley winds and
more humid environment are at the bottom of the valley while the moistening of the upper level could also be found in both experiments.
On the other hand, the results of the weakest inversion strength experiment (TH03) show that the moisture in the upper level of the valley is drier than the other experiments.
At 3 pm, the water vapor mixing ratio at 2000 m (red box in Fig. 14) of TH08 and TH10 experiments are comparable 7.00 and 7.07 g kg-1 respectively, while it is only 6.33 g kg
-1 in TH03 experiment. It is worth noting that the strength of the capping inversion can
influence the moisture within the boundary layer. By examining the conditions of the boundary layer in the valley (Fig. 15), we found that with stronger capping inversion (more than 25K km-1 respectively), both the liquid water and water vapor are kept under the inversion layer. In experiment TH03, the prescribed inversion is too weak so that it could be overcome and eliminated by convections. Once the inversion no longer exists, the convections develop to a higher level and transport moisture upward to the free atmosphere. Without the inversion to confine moisture in the valley, the fog is barely formed, and the duration is shorter in the experiment TH03. The sensitivity experiments
suggest that the temperature inversion strength in the Xitou valley is the crucial factor in the duration of the fog.
19