/glacierhack_2018

DEM differencing and time series analysis

Primary LanguageJupyter Notebook

Glacierhack 2018 (Project Title)

DEM differencing and time series analysis

Collaborators

  • Elad Dente - Hydro-Geomorphology
  • Håvard Holm - Applied Mathematics
  • Daniel Howard - Applied Mathematics
  • Michelle Hu - Hydrology - Snow Water Runoff
  • Lynn Kaack - Applied Mathematics
  • Joachim Meyer - Computer Science and Software Development
  • Wei Wei - Geophysics - Ice Sheets and Ocean Interactions

Team Lead

  • Shashank Bhushan - Glaciology and Geospatial Image Analysis

Data Science Lead

  • Friedrich Knuth - Data Science Methods and Geospatial Image Analysis

The problem

  • Can we quantify inter-annual changes in digital elevation models that represent glacial mass balance?
  • Can we improve upon time series analysis methods cappturing changes in digital elevation models (DEMs)?
  • What can we learn from image analysis and statistical methods (machine learning), applied to this 4 dimensional array?
  • How do our solutions perform at scale? Can we leverage the xarray stack and processing power of a Pangeo? Pangeo is a kubernetes powered jupyterhub configuration that enables distributed data processing and analysis through dask and xarray.

Relevance (So What? - Application Example)

  • Predict the fate of glaciers and impact for water resource management.
  • Explore if methods developed for this dataset can be applied to other glacier systems, such as glaciers that experience periodic surges.
  • Learn new data science methods.

Dataset

Khumbu Time Series

Specific Questions

  • Can we quantify inter-annual changes in digital elevation models that represent glacial mass balance?
  • Is the trendfitting of glacial mass change robust to systems that experience high variability? (What kind of math makes sense?) -- math folks
  • Can we improve upon time series analysis methods cappturing changes in digital elevation models (DEMs)?
  • What can we learn from image analysis and statistical methods (machine learning), applied to this 4 dimensional array?
  • How do our solutions perform at scale? Can we leverage the xarray stack and processing power of a Pangeo? Pangeo is a kubernetes powered jupyterhub configuration that enables distributed data processing and analysis through dask and xarray.
  • How much water is being released / added to the system?
  • Can we calculate velocities from changes in elevation and create a velocity map/vector field for the glacier?

Proposed methods/tools

Tools/Libraries

Relevant python libraries

Products

  • Notebook for visualization
  • Interactive 2D/3D widgets

Goals

  • Explore stacking raster DEMs and conversion to nD xarray object (basic elevation time series manipulations)
  • Explore stacking raster DEMs and conversion to Dask array (perform distributed computing)
  • Explore bridge between a dataset (Python) and Google Earth Engine (Javascript)

Background reading

See the Wiki