Model Uncertainty

There is growing consensus that land surface models need to be confronted with a wide range of data to constrain uncertainty in parameters, initialize surface states, and address model structural uncertainty. Model uncertainty can be reduced by sampled isolation, parameter constraints, and ensemble approaches, and arises from internal variability, data errors, and the parameterization of sub-grid scale processes. One example of a tool for addressing uncertainty in model structures and parameterizations is data assimilation. The potential of improving model uncertainty is untapped and possible effects on future models are unknown. Here we feature a series of talks given at the 2021 virtual workshop on ‘Tackling Technical Challenges in Land Data Assimilation’ the talks feature a range of DA methods used within the community spanning techniques used by numerical weather prediction and land surface modeling. 

Introduction and Orientation to the Webinar Series

Speakers: Andy Fox (Joint Center for Satellite Data Assimilation), Hannah Liddy (AIMES), Jana Kolassa (NASA GSFC), Natasha MacBean (Indiana University), Tristan Quaife (Univesity of Reading)

Date aired: 14-16 June 2021

Technical challenges of coupled land-atmosphere data assimilation for operational NWP and reanalyses

Speaker: Patricia De Rosnay (ECMWF)

Date aired: 14-16 June 2021

About: Patricia De Rosnay gives a presentation called: Technical challenges of coupled land-atmosphere data assimilation for operational Numerical Weather Prediction and reanalyses.

Topic: Applicability of data assimilation approaches across different land modeling groups

Monitoring land surface variables with LDAS-Monde
Speaker: Bertrand Bonan (CNRM)

Date aired: 14-16 June 2021

About: Bertran Bonan from CNRM gives a presentation on Monitoring land surface variables with LDAS-Monde: focus on assimilation approaches and applications to kilometric-scale spatial resolutions.

Topic: Applicability of data assimilation approaches across different land modeling groups

Experiences on terrestrial model parameter optimisation based from CCDAS using multiple observations

Speaker: Marko Scholze (Lund University)

Date aired: 14-16 June 2021

About: Marko Scholze from Lund University gives a presentation on: Experiences on terrestrial model parameter optimisation based from the Carbon Cycle Data Assimilation System using multiple observations.

Topic: Applicability of data assimilation approaches across different land modeling groups

Land hydrology data assimilation – Are we on the right track?

Speaker: Sujay Kumar (NASA GSFC)

Date aired: 14-16 June 2021

About: Sujay Kumar from NASA GSFC gives a presentation: Land hydrology data assimilation – Are we on the right track?

Topic: Applicability of data assimilation approaches across different land modeling groups

The added value of brightness temperature assimilation for global soil moisture estimation

Speaker: Jianzhi Dong (MIT)

Date aired: 14-16 June 2021

About: Jianzhi Dong from MIT gives a presentation called: The added value of brightness temperature assimilation for global soil moisture estimation.

Topic: Emerging techniques in land data assimilation

Hybrid data assimilation methods for land surface modelling

Speaker: Ewan Pinnington (University of Reading)

Date aired: 14-16 June 2021

About: Ewan Pinnington from the University of Reading gives a presentation on Hybrid data assimilation methods for land surface modelling.

Topic: Emerging techniques in land data assimilation

Gaussian process emulators for efficient Bayesian calibration of process-based models

Speaker: Istem Fer (Finnish Meteorological Institute)

Date aired: 14-16 June 2021

About: Istem Fer from the Finnish Meteorological Institute presented on Gaussian process emulators for efficient Bayesian calibration of process-based models.

Topic: Emerging techniques in land data assimilation

Estimating the full observation error covariance matrix

Speaker: Joanne Waller (UK Met Office)

Date aired: 14-16 June 2021

About: Joanne Waller from the UK Met Office gives her presentation on: Estimating the full observation error covariance matrix.

Topic: Emerging techniques in land data assimilation

Enhanced streamflow forecasting using ensemble data assimilation

Speaker: Moha El Gharamti (NCAR/UCAR)

Date aired: 14-16 June 2021

About: Moha El Gharamti from NCAR/UCAR presents on Enhanced streamflow forecasting using ensemble data assimilation.

Topic: Emerging techniques in land data assimilation

Using the temporal dynamics of surface soil moisture to deal with biases when calibrating LSMs

Speaker: Nina Raoult (LSCE)

Date aired: 14-16 June 2021

About: Nina Raoult from LSCE gives a presentation on using the temporal dynamics of surface soil moisture to deal with biases when calibrating land surface models.

Topic: Challenges in dealing with observations

Using the ITRDB records as century-long benchmarks for land-surface models

Speaker: Jina Jeong (Vrije Universiteit Amsterdam)

Date aired: 14-16 June 2021

About: Jina Jeong from Vrije Universiteit Amsterdam gave a presentation on: Using the International Tree-Ring Data Bank (ITRDB) records as century-long benchmarks for land-surface models.

Topic: Challenges in dealing with observations

Want to learn more?

If you are interested in model uncertainty, take a look at our webpage for the Land Data Assimilation Working Group. You’ll find information on our webinar series as well as recent developments in the field.