search

r-barnes / faster-unmixer

A faster implementation of the sediment inverse/unmixing scheme proposed in Lipp et al (2021).
5 stars 1 forks source link

Solve for the amount of material contributed from each area in addition to composition #15

Closed AlexLipp closed 1 year ago

AlexLipp commented 1 year ago

Currently we assume each sub-basin contributes an amount of material proportional to its area, implicitly assuming constant erosion rates/run-off. However, if multiple tracers are used simultaneously, the actual amount of material being contributed from each catchment could be solved for as a free variable.

This would be a relatively major new feature as it would require solving for multiple elements simultaneously (this could be rescued from previous versions) and we would need to think carefully about how it is regularised. Potentially very impactful however as it would allow long-term erosion rates and run-off to simply be estimated from geochemical observations downstream.

AlexLipp commented 1 year ago

As this is exploratory and probably beyond scope of current target I'm closing this as an issue, but will explore in a fork in future.