ICESAT-2HackWeek / xtrak

Repo for the cross-track project group, for fine-scale spatiotemporal interpolation between ICESat-2 tracks

xtrak

Fine-scale spatiotemporal interpolation between ICESat-2 tracks

Collaborators

Project Lead: Tim Bartholomaus, University of Idaho

Data Science Lead: Fernando Paolo, NASA Jet Propulsion Laboratory

Lavanya Ashokkumar, University of Arizona

Taryn Black, University of Washington

Allison Chartrand, Ohio State University

Jukes Liu, University of Maine

Evan Carnahan, University of Texas

The Problem

How well can we interpolate between ICESat-2 tracks?

ICESat-2 products will eventually include quarterly (91-day) gridded land ice elevations. We want to be able to also estimate elevations effectively continuously through time, so that we can get rough elevation profiles on shorter time scales. We will interpolate elevations between neighboring tracks to create gridded elevation data from which we can extract our elevation profiles. We do not expect the shape of an elevation profile to change significantly between the collection of neighboring tracks, so interpolating between them should be safe.

Pie-in-the-sky goal: essentially continuously updating elevation profiles, such that we can enter a location and date and get an estimated elevation profile.

Application Example

Interpolate ICESat-2 tracks to build elevation profiles of Zachariae Isstrom, an outlet glacier in northeastern Greenland.

Most tracks over Greenland outlet glaciers are across-flow rather than along-flow, so it is hard to get a single temporally-consistent elevation profile of a glacier. By interpolating between neighboring tracks, we can build an along-flow elevation profiles of Zachariae Isstrom roughly every two weeks.

Sample Data

• Operation IceBridge, ATM profiles over Zachariae Isstrom, 2014 & 2018. ~1.3 MB as CSV, from NSIDC.
• ICESat-2 tracks over Zachariae Isstrom. ~1.8 GB as HDF5, from NSIDC.

Specific Questions

Can we reasonably estimate elevation z(x,y,t) for any arbitrary space (x,y) and time (t)?

Existing Methods

Airborne and satellite altimetry (e.g. Operation IceBridge ATM)

Proposed Methods/Tools

• pandas, geopandas
• linear interpolation with scipy.griddata

Workflow

Get data

script	input data	output data	notes
data_raw/downloadData.sh	NSIDC	data_raw/ATL06evan	over ZI, all dates
manual download	NSIDC	data_raw/ATM_20**
manual download	NSIDC	data_raw/ATL06lavanya	over ZI, December only

Subset and extract variables of interest

script	input data	output data	notes
IS2ZachExtract.ipynb	data_raw/ATL06evan	data_prod/ICESat2ZachData.csv
ATMprocessing.ipynb	data-raw/ATM_20**	data_prod/ATMprof_20**_slopes.csv

Error checks, smoothing, and filtering

script	input data	output data	notes
CheckForBadISatTransects.ipynb	data_prod/IceSat2ZachData.csv	data_prod/ZachISatData_wSmooth.csv
SmoothAtm.ipynb	data_prod/ATMprof_20**_slopes.csv	data_prod/ATMprof_20**\slopes_wSmooth.csv

Identify crossover points

script	input data	output data	notes
Intersections.ipynb	data_prod/ZachISatData_wSmooth.csv, ATMprof_20**_slopes_wSmooth.csv	data_prod/InterX_ATM20**_AllSmooth.csv

Get residuals

script	input data	output data	notes
calculate_residuals.ipynb	data_prod/InterX_ATM2014_AllSmooth.csv	data_prod/residuals.csv

Interpolation

script	input data	output data	notes
fit_residuals.ipynb	data_prod/residuals.csv	plots