Introduction

Overview

This tutorial contains jupyter notebooks demonstrating various steps of a typical scientific workflow including accessing, processing and visualizing remote sensing data. The structure is as follows:

1) Data access

• Access ITS_LIVE data stored as Zarr data cubes in an AWS S3 bucket.

2) Reading and working with a larger-than-memory dataset

• Illustrate different strategies for manipulating and organizing a large dataset using Xarray, Zarr, and Dask.

3) Working with raster and vector data

• Parse geographic metadata with cf_xarray.

• Handle projections and coordinate reference system information with GeoPandas, Rioxarray and PyProj.

• Spatial subset of vector data with GeoPandas.

• Spatial subset of raster data using vector data with Rioxarray.

4) Initial inspection and analysis of velocity data for a single glacier

• Handle projections and coordinate reference system information with GeoPandas, Rioxarray and PyProj.

• Visualize raster and vector with background maps data using Xarray, GeoPandas, and Contextily.

• Calculate and examine data coverage along a given dimension using Xarray label-based indexing and selection.

• Use available metadata to interpret and organize dataset,

  • Use xr.DataTree or groupby() to separate dataset using metadata,

• Use Xarray and scipy.stats to calculate and visualize summary statistics along a given dimension.

• Perform dimensional computations, reductions and visualizations using Xarray resample(), groupby() and FacetGrid.

5) Exploratory analysis and visualization of multiple glaciers

• Combine raster and vector data into a multi-dimensional vector data cube using Xvec.

• Read and write vector data cubes to disk using Xvec methods that rely on cf_xarray to encode and decode metadata.

• Interactive visualization of vector data cube using Xvec and GeoPandas.

• Use Xarray plotting tools to visualize data from a vector data cube.

Relevant Concepts

This tutorial will spend a lot of time discussing the following concepts, if they’re unfamiliar to you, we recommend first heading to Relevant Concepts.

1. Larger than memory data

2. Dask

3. Chunking

Learning goals

This tutorial demonstrates how to use xarray for scientific investigation of remote sensing data. By the end of this tutorial, you should be able to:

• Read Zarr data from AWS S3 cloud object stores,

• Understand different approaches for working with larger-than-memory data,

• Use GeoPandas to manipulate and visualize vector data,

• Clip raster data by vector data objects using GeoPandas and RioXarray,

• Parse geographic metadata with cf_xarray,

• Build and use vector data cubes with Xvec,

• Write multi-dimensional data to disk as Zarr data cubes and understand how different metadata storage can affect these operations.

Software

For instructions on setting up a computing environment needed for this tutorial, see Software.

Data

For more background on the data used in this tutorial, head to Tutorial Data.

To get started with this tutorial, make sure you’ve followed the instructions on the Software page for downloading the necessary material and setting up a virtual environment, then head to the first notebook.