Statistics of Water Vapor Variability in the Tropics from Airborne Lidar

• Main Author: Sadykov, Aidar
• Format: Thesis
• Language: unknown
• Published: 2019
• Subjects: Lidar; narval
• Online Access: https://elib.dlr.de/131209/

The tropics are a forge of Earth’s climate, since they store a considerable part of terrestrial water vapor, which is a key greenhouse effect contributor. Water vapor is extremely variable in space and time due to inhomogeneous distribution of humidity sources and sinks intertwined with complex atmospheric dynamics. This presents challenges to water vapor parametrization in numerical climate and weather models given the ongoing climate change, effectively increasing the uncertainty of predictions. In order to bridge the gap and elucidate water vapor variability, we analyze WALES (a differential absorption lidar) and dropsonde data on relative humidity and water vapor mass mixing ratio from NARVAL, an airborne campaign near the Lesser Antilles in December 2013 and August 2016. Firstly, we identify four physically different atmospheric condition classes of regions in lidar curtains: winter undisturbed, summer undisturbed, summer Saharan dust and summer tropical storm; and inspect them with the help of three statistical methods: statistical moments, probability density functions and Fourier analysis. As the result, characteristic behavior of water vapor in the planetary boundary layer, the cloud layer and the free atmosphere is described in terms of mean, standard deviation and skewness profiles of relative humidity; probability density functions of water vapor mass mixing ratio across ranging horizontal slices of the cloud layer; and Fourier power spectra of mean class relative humidity profiles. Vertical lidar profiles feature high variability in the cloud layer at altitudes between 500 m and 2500 m (above mean sea level), where humid clouds are interleaved with drier gaps between them. The free atmosphere between 2500 m and 9000 m typically contains only minor humidity oscillations around a very dry central value due to intermittently occurring wetness bubbles. Storms, however, are capable of bringing wet air parcels above the capping inversion and disturbing the entire altitude range of the free atmosphere. Above 9000 m relative humidity variance increases again, being fueled by the Hadley circulation. Differences between winter and summer, disturbed and undisturbed regions manifest only in the free atmosphere and the cloud layer, while the planetary boundary layer is stable and does not exhibit noticeable dependence on the season or the disturbance status. The free atmosphere is wetter and more variable in summer, because of the closer proximity to the Intertropical Convergence Zone. Additionally, the lower cloud layer boundary moves from around the 445 m − 565 m altitude level in summer to 616 m in winter. Lastly, a comparison between the lidar measurements and the output of the Icosahedral Nonhydrostatic (ICON) model for numerical weather prediction is conducted. There is generally a good agreement between lidar and model mean values of water vapor mass mixing ratio. Nevertheless, natural water vapor variability is underrepresented in the ICON output, which is reflected by standard deviation of mass mixing ratio being 1.2 − 3.8 times lower than in lidar curtains, which indicates issues with the model algorithm. Moreover, probability density functions of water vapor mass mixing ratio across ranging 100 − m − thick horizontal slices of the cloud layer exhibit bimodality in 87.5%−100.0% of considered lidar curtains, while the corresponding simulated ICON curves are predominantly unimodal (12.5% bimodality rate). The model has difficulties representing cloud layer heterogeneity in the tropics, being unable to resolve both peaks of mass mixing ratio in a horizontal slice. Instead, it returns only a single, merged maximum lying between two respective lidar modes. This reveals a crucial physical process in the cloud layer might be not implemented or reasonably parametrized in the weather model. A follow-up study could make an attempt to isolate a reason for the described cloud layer patterns in lidar and ICON data; and evaluate their agreement comprehensively under various atmospheric conditions.