Can we use semi-qualitative observations to forecast the Arctic sea ice?

Dr Abhishek S. Shah.Dr Abhishek S. Shah.The answer was given during Dr. Abishek Shah´s the doctoral dissertation on 16. September at the Mathematical Institute, University of Bergen. His thesis dealt with the methods of data assimilation that blend observations and a numerical model into a forecasting system as used routinely by for example weather forecasts. The data assimilation method is introduced in the modelling system to improve the accuracy of the forecast by optimizing the forecast to better match observed parameters.

In the Arctic Ocean and sea ice forecasting system, established and developed at the Nansen Center during the last two decades, the preferred data assimilation method is the Ensemble Kalman Filter (EnKF). This coupled ice-ocean forecasting system (TOPAZ) is in operational use by Met-Norway providing weekly forecast
s for the entire Arctic Ocean during the next ten days, as a part of the European Copernicus Marine Forecasting and Monitoring Services.

The Arctic Ocean, in contrast to its natural beauty, is a remote and hostile environment, which is more efficiently monitored by remote sensing satellites than by field expeditions. One of the most critical questions in Arctic climate is: how thick is the sea ice? Some satellites can measure the ice thickness but are technically limited to a range of values. The sea ice thickness values falling “out-of-range” are so far ignored by data assimilation methods like the EnKF, even though they do provide some information in the shape of inequalities: knowing that the value is above the detection limit is more useful than having no observation at all. Such out-of-range inequality is called “semi-qualitative” observation.

Dr Shah introduces in his thesis a new data assimilation method called the Ensemble Kalman Filter – Semi-Qualitative (EnKF-SQ) that is able to assimilate all values both within and outside of the observed range. The first part of the thesis concentrates on the formulation and the mathematical properties of the EnKF-SQ, followed by a real-scale test using the same coupled ice-ocean model of the Arctic used for operational predictions. The tests in the doctoral studies of Dr Shah show that the EnKF-SQ is able to take benefit from the semi-qualitative observations and adds value to the satellite data to be assimilated into the forecasting system.

Dr. Abishek Shah has conducted his PhD. studies working under Nordic Center of Excellence EmblA: Ensemble-based data assimilation for environmental monitoring and prediction led by the Nansen Center. His supervisors have been Dr. Laurent Bertino and Mohamad El Gharamti.

From left: Alf Hartvig Øien (UoBergen), Laurent Bertino (NERSC),  Dr. Abhishek Shah, Tijana Janjic-Pfander (Hans Ertel Center for Weather Research,  German Weather Service), Erik Hanson (UoBergen) and François Massonnet (UCLouvain).From left: Alf Hartvig Øien (UoBergen), Laurent Bertino (NERSC), Dr. Abhishek Shah, Tijana Janjic-Pfander (Hans Ertel Center for Weather Research, German Weather Service), Erik Hanson (UoBergen) and François Massonnet (UCLouvain).

 

Citation: Shah, Abishek (2019): Stochastic data assimilation of observations with a detection limit. Doctoral thesis University of Bergen, Department of Mathematics. 09/2019. 97 pp. ISBN 9788230856741.