AI for cloud classification
Course Overview
Embark on a captivating journey into the world of cloud systems and how exciting present-day artificial intelligence developments can help us pursue data-driven approaches for process understanding in nature. This course unravels the intricacies of cloud structure, distribution, and their vital role in the climate system. Overcome challenges specific to AI and earth science datasets, exploring satellite measurements and their post-processing.Dive into the neural architecture, optimization based on physical properties, and the generalization capacity of trained networks. Explore the significance of trade wind cumulus cloud systems, organizational variability, and dimensionality reduction. Evaluate the agreement between human and machine-identified cloud organizations, transitions, and atmospheric intrusions.Understand the importance of cloud observation and the transformative potential of machine learning in interpreting satellite measurements. This course is a must for those passionate about cloud dynamics and their impact, providing insights for weather prediction and climate models.
Details
Lecturer
Overview
Embark on a captivating journey into the world of cloud systems and how exciting present-day artificial intelligence developments can help us pursue data-driven approaches for process understanding in nature. This course unravels the intricacies of cloud structure, distribution, and their vital role in the climate system. Overcome challenges specific to AI and earth science datasets, exploring satellite measurements and their post-processing.Dive into the neural architecture, optimization based on physical properties, and the generalization capacity of trained networks. Explore the significance of trade wind cumulus cloud systems, organizational variability, and dimensionality reduction. Evaluate the agreement between human and machine-identified cloud organizations, transitions, and atmospheric intrusions.Understand the importance of cloud observation and the transformative potential of machine learning in interpreting satellite measurements. This course is a must for those passionate about cloud dynamics and their impact, providing insights for weather prediction and climate models.
Importance of cloud systems for Climate change and Renweable energy
In this video, we will talk about cloud systems, their structure and distributions, and how exciting present-day artificial intelligence developments can help us pursue data-driven approaches for process understanding in nature.
Past state-of-the-art research studies and possible limitations
In this video we will talk about • our approach on how do we plan to overcome the limitations of the previous studies • further challenges that are specific to data sets in case of AI and earth science community. • satellite measurement and its post-processing used for our study.
Deep learning Architecture and evolution of understanding
In this video we will carry forward our last discussion and will move on to describe the neural architecture. • different components of the neural architecture and why they are used. • how the network evolves and improves its learning through multiple iterations.
Integrating physical knowledge to find optimzed classes
In this video we will • try to find out the optimized classes based on their physical properties. • visualize in the representation space, the distribution of the optimized classes.• study the physical properties of the identified classes using auxiliary satellite products.
Gaining inference of the identified classes by the neural network
In this video we will see • how to gain inference about the identified cloud regimes and how distinct they are as far as other physical fluxes are concerned • What are the cloud types and are they physically meaningful?
Generalization: How does the network perform on unseen data
In this video we will talk about generalization capacity of the trained neural network
Application for solar enery using transfer learning from the pre-trained neural network
In this video we will explore some details about the feature space of the trained neural network.
Significance of trade wind cumulus clouds & their organizational variability over tropics
In this video we will understand the significance of trade wind cumulus cloud systems with regards to climate change and then discuss briefly about their organizational variability over tropics.
Visual features of cloud organizations and exploiting them to capture the diversity
In this video we will dive a little bit into visual features of cloud organizations and further how can we use them to train a neural a network that can learn to capture the full diversity of cloud systems in tropics.
Finding optimal boundary conditions in the continous space
Here in this video you will see how does the dimensionally reduced version of the high dimensional space looks like. Also after that you will learn how to put optimal boundary conditions in that continuous setup.
Extension of agreement between humans and machines for the identified cloud organizations
In this video we will discuss the extent of agreement between human identified cloud organizations and machines classes.
Capturing transition in cloud organziations
In this video we will talk about catching transistion of one cloud system to other.
Dry - wet atmospheric intrusions
In this video a brief introduction is given about dry-wet atmospheric intrusions over Barbados region in North Atlantic trades.
Importance of Satellite Cloud Remote Sensing
Clouds as fascinating and important features in the climate system and how representing clouds in weather prediction and climate models is a major challenge
Satellite Cloud Remote Sensing Introduction
In this video, we will dive into how machine learning can help interpreting satellite measurements of clouds.
How can machine learning help cloud research?
Importance of cloud observation, challenges of satellite remote sensing of atmospheric features and possibilities of ML for remote sensing of clouds and atmospheric features
