High Resolution WRF Simulation of Water Vapor Transport to the Upper Troposphere/Lower Stratosphere during Hurricane Ingrid (2013)

Atmospheric water vapor is a potent greenhouse gas, and its variations in the upper troposphere and lower stratosphere (UTLS) have important climate impacts. However, the water vapor budget of tropical cyclones (TCs) and their impact on the UTLS remain understudied. This paper describes high-resolution simulations of TC Ingrid (2013) using the Weather Research and Forecasting (WRF) model to calculate its water vapor budget. WRF was used in a nested configuration, with the finest grid spacing being 1.333 km. Our WRF simulations of Ingrid were verified using in situ aircraft data (DC-8 and ER-2) when both aircraft coordinated their flights through Ingrid. The verifications revealed a good simulation in terms of water vapor mixing ratio, TC structure, and environmental wind shear. Since our simulation agreed well with observations and our understanding of TCs, we used the WRF model output as though it were observed data. Calculations of water vapor flux (t h^-1) and water vapor flux density (t h^-1 km^-2) were made for Ingrid as a whole. Maxima of upward vapor flux and flux density were located at altitudes of 1.2 km, while maxima of downward vapor flux and flux density occurred at 1 and 1.2 km, respectively. Maximum net vapor flux and flux density were found near 1.7 km. The research then focused on the 100 hPa and 875 hPa levels. The 875 hPa level is above the top of PBL near the level of maximum upward net flux, and 100 hPa is slightly above the tropopause. Results showed positive (upward) net water vapor flux at both altitudes and at all times during the period. Thus, even a weak Category 1 TC like Ingrid can transport a large amount of water vapor through the tropopause. Calculations also were made for the various regions of Ingrid (eye, eyewall, and outer rainband). Areas of ascent and descent were simulated within Ingrid's eye at 875 hPa, leading to a small net positive (upward) water vapor flux and flux density. Strong simulated vertical motions at 875 hPa outside the eye were associated with simulated deep convection within both the eyewall and outer rainbands. However, since the area of the outer rainbands (96.2% of the total TC) was much larger than the eyewall (3.3% of the total TC), the rainbands contributed more to the water vapor flux. At 100 hPa, the largest downward net flux density was in the eye region, even though it covered only a tiny area (0.5%) of the total TC. Gravity waves at 100 hPa were observed in the banded nature of vertical motion at this altitude. We compared Ingrid's simulated altitudes and magnitudes of hydration and dehydration with the September 2013 monthly average. Ingrid and its environment were moister than the monthly average at all altitudes except the lower stratosphere. We also calculated differences in water vapor mixing ratio from the beginning to end of the simulation. Results in the UTLS showed greatest dehydration at ~ 15.5 km, widespread hydration at ~ 18.5 km, and a reversal back to dehydration at ~ 20.8 km. To further investigate these results we explored 1) deep convection that overshoots into the stratosphere and 2) storm tops whose relative humidity with respect to ice (RHi) exceeds 100%. Overshooting convection was observed in satellite imagery of Ingrid, and the simulation revealed numerous overshooting cells with tops as high as 2.3 km above the tropopause. The areal coverage of the simulated overshooting tops was much less than that of the total area of Ingrid, varying from 0.002% to 0.128%. Simulated upward water vapor flux density by overshooting tops at 100 hPa was as much as 21.5 times greater than that of the TC as a whole. Thus, very small areas of overshooting convection have a major impact on vertical transport of water vapor to the UTLS. The intense convection provides a direct transport mechanism of water vapor into the stratosphere. Thus, the frequency and penetrating altitude of overshooting convection is crucial to the resulting moistening and drying of the tropical tropopause layer (TTL). Relative humidity with respect to ice (RHi) is also a factor influencing hydration and dehydration in the atmosphere. Deep convection can hydrate the lower stratosphere by ice sublimation, or it can dehydrate the region by ice sedimentation. The effectiveness of ice sedimentation and the initial environmental TTL RHi determines whether deep convection hydrates or dehydrates the lower stratosphere. Our simulated average maximum RHi values showed supersaturation with respect to RHi between ~ 5 km to 17.25 km, and subsaturation from ~ 17.25 km to 21 km. Supersaturation (subsaturation) with respect to ice from 5 km to 17.25 km (~ 17.25 km to 21 km) would therefore be thought to result in dehydration (hydration). However, the vertical profile of Ingrid revealed dehydration from ~ 14.5-17.5 km, hydration from ~ 17.5-20 km, and dehydration from ~ 20-21 km. Thus, RHi alone cannot completely explain these layers of hydration and dehydration in our simulation. The results suggest that a variety of factors impact the altitude and magnitude of hydration and dehydration in the UTLS. We hypothesize that our simulated dehydration from ~ 14.5-17.5 km is likely due to ice sedimentation, gravity waves, and cirrus clouds; hydration within ~ 17.5-20 km is due to downward transport of water vapor due to methane oxidation, ice sublimation by deep convection, and “cirrus lofting”; and the dehydration from ~ 20-21 km is likely due to gravity waves and ice sedimentation.