Reproducibly estimating and evaluating supraglacial lake depth with Landsat 8 and other multispectral sensors

摘要

Lakes which form on the surface of ice sheets (supraglacial lakes) play an important role in the ice sheet hydrological system, serving as temporary reservoirs for meltwater that can lead to ice fracture and associated ice flow speedups or ice shelf disintegration. Satellite imagery can be used to monitor the extent and volume of supraglacial lakes. In this study, we summarize and expand on methods for doing this that have been optimized for the multispectral imager aboard Landsat 8. This report fully documents the data and software (i.e., MATLAB and GDAL scripts) used in these analyses, linking citeable and freely accessible data (in ACADIS and Figshare) to published figures via shared (GitHub), described (OntoSoft), and citeable (Zenodo) code as described in a full provenance diagram. In addition, new water supraglacial volumes for the Sermeq Kujalleq (Jakobshavn) region for 2015 are presented and compared to results from 2014. Lakes in 2015 largely behaved similarly to 2014, with maxima and minima occurring in recurrent areas; disparate behavior is explained both by aliasing and interannual temperature variability. The challenges of achieving full openness, documentation, and reproducibility are discussed, as well as examples where higher standards of accessibility and provenance would have facilitated the original research. Still, the interdependence of documentation can make accessibility challenging to achieve. However, if best practice is integrated into research processes and workflow, as demonstrated here, it will facilitate more complete and more effective reproducibility and reuse of geoscience methods and results.