Search for: All records

In this paper, we consider a coupled chemotaxis-fluid system that models self-organized collective behavior of oxytactic bacteria in a sessile drop. This model describes the biological chemotaxis phenomenon in the fluid environment and couples a convective chemotaxis system for the oxygen-consuming and oxytactic bacteria with the incompressible Navier–Stokes equations subject to a gravitational force, which is proportional to the relative surplus of the cell density compared to the water density. We develop a new positivity preserving and high-resolution method for the studied chemotaxis-fluid system. Our method is based on the diffuse-domain approach, which we use to derive a new chemotaxis-fluid diffuse-domain (cf-DD) model for simulating bioconvection in complex geometries. The drop domain is imbedded into a larger rectangular domain, and the original boundary is replaced by a diffuse interface with finite thickness. The original chemotaxis-fluid system is reformulated on the larger domain with additional source terms that approximate the boundary conditions on the physical interface. We show that the cf-DD model converges to the chemotaxis-fluid model asymptotically as the width of the diffuse interface shrinks to zero. We numerically solve the resulting cf-DD system by a second-order hybrid finite-volume finite-difference method and demonstrate the performance of the proposed approach on a number of numerical experiments that showcase several interesting chemotactic phenomena in sessile drops of different shapes, where the bacterial patterns depend on the droplet geometries. 

Domain adaptation techniques using deep neural networks have been mainly used to solve the distribution shift problem in homogeneous domains where data usually share similar feature spaces and have the same dimensionalities. Nevertheless, real world applications often deal with heterogeneous domains that come from completely different feature spaces with different dimensionalities. In our remote sensing application, two remote sensing datasets collected by an active sensor and a passive one are heterogeneous. In particular, CALIOP actively measures each atmospheric column. In this study, 25 measured variables/features that are sensitive to cloud phase are used and they are fully labeled. VIIRS is an imaging radiometer, which collects radiometric measurements of the surface and atmosphere in the visible and infrared bands. Recent studies have shown that passive sensors may have difficulties in prediction cloud/aerosol types in complicated atmospheres (e.g., overlapping cloud and aerosol layers, cloud over snow/ice surface, etc.). To overcome the challenge of the cloud property retrieval in passive sensor, we develop a novel VAE based approach to learn domain invariant representation that capture the spatial pattern from multiple satellite remote sensing data (VDAM), to build a domain invariant cloud property retrieval method to accurately classify different cloud types (labels) in the passive sensing dataset. We further exploit the weight based alignment method on the label space to learn a powerful domain adaptation technique that is pertinent to the remote sensing application. Experiments demonstrate our method outperforms other state-of-the-art machine learning methods and achieves higher accuracy in cloud property retrieval in the passive satellite dataset. 

Atmospheric gravity waves are produced when gravity attempts to restore disturbances through stable layers in the atmosphere. They have a visible effect on many atmospheric phenomena such as global circulation and air turbulence. Despite their importance, however, little research has been conducted on how to detect gravity waves using machine learning algorithms. We faced two major challenges in our research: our raw data had a lot of noise and the labeled dataset was extremely small. In this study, we explored various methods of preprocessing and transfer learning in order to address those challenges. We pre-trained an autoencoder on unlabeled data before training it to classify labeled data. We also created a custom CNN by combining certain pre-trained layers from the InceptionV3 Model trained on ImageNet with custom layers and a custom learning rate scheduler. Experiments show that our best model outperformed the best performing baseline model by 6.36% in terms of test accuracy. 

Abstract. In this study, we developed a novel algorithm based on the collocatedModerate Resolution Imaging Spectroradiometer (MODIS) thermal infrared (TIR)observations and dust vertical profiles from the Cloud–Aerosol Lidar withOrthogonal Polarization (CALIOP) to simultaneously retrieve dust aerosoloptical depth at 10 µm (DAOD10 µm) and the coarse-mode dusteffective diameter (Deff) over global oceans. The accuracy of theDeff retrieval is assessed by comparing the dust lognormal volumeparticle size distribution (PSD) corresponding to retrieved Deff withthe in situ-measured dust PSDs from the AERosol Properties – Dust(AER-D), Saharan Mineral Dust Experiment (SAMUM-2), and Saharan Aerosol Long-Range Transport and Aerosol–Cloud-InteractionExperiment (SALTRACE) fieldcampaigns through case studies. The new DAOD10 µm retrievals wereevaluated first through comparisons with the collocated DAOD10.6 µmretrieved from the combined Imaging Infrared Radiometer (IIR) and CALIOPobservations from our previous study (Zheng et al., 2022). The pixel-to-pixelcomparison of the two DAOD retrievals indicates a good agreement(R∼0.7) and a significant reduction in (∼50 %) retrieval uncertainties largely thanks to the better constraint ondust size. In a climatological comparison, the seasonal and regional(2∘×5∘) mean DAOD10 µm retrievals basedon our combined MODIS and CALIOP method are in good agreement with the twoindependent Infrared Atmospheric Sounding Interferometer (IASI) productsover three dust transport regions (i.e., North Atlantic (NA; R=0.9),Indian Ocean (IO; R=0.8) and North Pacific (NP; R=0.7)). Using the new retrievals from 2013 to 2017, we performed a climatologicalanalysis of coarse-mode dust Deff over global oceans. We found thatdust Deff over IO and NP is up to 20 % smaller than that over NA.Over NA in summer, we found a ∼50 % reduction in the numberof retrievals with Deff>5 µm from 15 to35∘ W and a stable trend of Deff average at 4.4 µm from35∘ W throughout the Caribbean Sea (90∘ W). Over NP inspring, only ∼5 % of retrieved pixels with Deff>5 µm are found from 150 to 180∘ E, whilethe mean Deff remains stable at 4.0 µm throughout eastern NP. To the best of our knowledge, this study is the first to retrieve both DAOD andcoarse-mode dust particle size over global oceans for multiple years. Thisretrieval dataset provides insightful information for evaluating dustlongwave radiative effects and coarse-mode dust particle size in models.