Effective Tutorial Notebooks for Cloud Computing Workflows using Remote Sensing Datasets

Abstract

Remote sensing research relies on high volumes of openly accessible data products from satellites spanning different time periods, data formats, and sizes. Traditionally, resources for data access have been limited to individual satellites based on funding mandates for specific missions, siloing tutorials and increasing complexity and cost for users to synthesize datasets to produce novel results. In this notebook, we demonstrate challenges and lessons learned in creating a tutorial notebook for onboarding users combining multiple datasets in a cloud computing environment. This notebook was presented to XX users onboarding to the CryoCloud JupyterHub, with overwhelmingly positive feedback. The notebook guides users to stream ICESat-2 altimetry data, SWOT altimetry data, Dynamic Atmosphere Corrections, background elevation and ice velocity, run a tide model, and plot a comparison of the ICESat-2 and SWOT elevations over a portion of an ice shelf. Through creating this tutorial, we learned three key ideas—speed, flexibility, and interest—that shape an effective data tutorial notebook. Effective tutorials are quick; no cell takes more than a minute to run to help users understand if the workflow is correct without investing significant time. This is often accomplished by using small test datasets but should also have tips for scaling up workflows to larger datasets. Effective tutorials are flexible, showing a full range of options so users can select which meets their needs; they are complete, so users can use the tutorial as a reference. Finally, effective tutorials are interesting, solving a real problem that engages the user. By striving to produce tutorials aligned with these key ideas, we believe they can be an effective tool for reducing the barrier to entry in remote sensing research, leading to more users and interesting science. This approach can be generalized to other fields with large open-access datasets.