GEOHYD Lunch Seminar: Ensemble-based retrospective analysis of the seasonal snowpack

Welcome to our GEOHYD Lunch Seminar Friday 27th of September @ 12:15 in Aud 1, The Geology building. The seminar is helt by Kristoffer Aalstad, Doctoral Research Fellow, Dept. of Geosciences.

Seminar by Kristoffer Aalstad, Doctoral Research Fellow, Dept. of Geosciences.

"Ensemble-based retrospective analysis of the seasonal snowpack"

Abstract: 

The seasonal snowpack, with its high albedo, low thermal conductivity and large water storing capacity, is a key component of the terrestrial energy, water, and carbon balance. At the same time, it is widely acknowledged that accurate estimation of the distribution of snow water equivalent (SWE) remains an outstanding challenge, especially in complex terrain. This is unfortunate because, as a modulator of the terrestrial balances, SWE is an essential climate variable.

Since the dawn of the satellite era, the depletion of snow-cover retrieved from optical satellite sensors has been used to reconstruct SWE in a deterministic manner. The basic idea is to use the remotely sensed date of disappearance of the snowpack to perform a backwards calculation to reconstruct SWE using modeled snowmelt rates. Despite some success, this approach is limited given that it completely ignores uncertainties in the forcing, model, and retrievals. Recently, ensemble-based snow retrospective analysis (reanalysis) that accounts for these uncertainties has emerged as a robust alternative to traditional reconstruction. In this approach, the assimilation of the remotely sensed depletion of fractional snow-covered area (fSCA) is used to constrain an ensemble of modeled realizations of the seasonal snowpack. Coincidentally with this emergence, a new generation of optical satellite sensors, such as Sentinel-2 MSI, have been launched into orbit to complement the already existing satellite climate data record.

In this work, multiple satellite sensors are leveraged in the pursuit of ways to further improve the ensemble-based reanalysis approach to SWE reconstruction. A modular reanalysis framework is presented with four components: a topographic downscaling routine, a simple snow model, fSCA retrieval algorithms, and ensemble-based data assimilation schemes. This framework can in principle be applied anywhere on Earth. This work focuses on three disparate study sites: the Brøgger peninsula in high-Arctic Svalbard archipelago, the Mammoth Lakes Basin in the Californian Sierra Nevada, and the Swiss Alps. Spectral unmixing is shown to provide the means of retrieving unbiased hill-slope scale fSCA from optical satellite sensors. It is also shown that an iterative ensemble smoother algorithm can outperform the data assimilation schemes that have previously been proposed for snow reanalysis. Through the experimental framework, a taste is provided of what is possible when combining robust ensemble-based data assimilation with emerging remotely-sensed data streams in the reanalysis exercise.​

About the seminar:

This seminar is offered by the Section for Geography and Hydrology at the Department of Geosciences, University of Oslo. The seminar is announced as a lunch seminar so bring your lunch if you want to. 

Everyone is welcome, and especially students. 

The lunch seminar team​​​.

Published Sep. 25, 2019 12:37 PM - Last modified Sep. 25, 2019 12:37 PM