Hybrid Levenberg–Marquardt and weak-constraint ensemble Kalman smoother method

Mandel, Jan; Bergou, El Houcine; Gürol, Selime; Gratton, Serge; Kasanický, Ivan

doi:10.5194/npg-23-59-2016

Cited by 20 publications

(9 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This restricts the class of problems for which it is unbiased but makes it more tractable (Oliver, 2017). Similar methods were proposed and studied by Bardsley et al (2014), Liu et al (2017), and Morzfeld et al (2018).…”

Section: Rmlmentioning

confidence: 99%

Revising the stochastic iterative ensemble smoother

Raanes

Stordal

Evensen

2019

Nonlin. Processes Geophys.

View full text Add to dashboard Cite

Abstract. Ensemble randomized maximum likelihood (EnRML) is an iterative (stochastic) ensemble smoother, used for large and nonlinear inverse problems, such as history matching and data assimilation. Its current formulation is overly complicated and has issues with computational costs, noise, and covariance localization, even causing some practitioners to omit crucial prior information. This paper resolves these difficulties and streamlines the algorithm without changing its output. These simplifications are achieved through the careful treatment of the linearizations and subspaces. For example, it is shown (a) how ensemble linearizations relate to average sensitivity and (b) that the ensemble does not lose rank during updates. The paper also draws significantly on the theory of the (deterministic) iterative ensemble Kalman smoother (IEnKS). Comparative benchmarks are obtained with the Lorenz 96 model with these two smoothers and the ensemble smoother using multiple data assimilation (ES-MDA).

show abstract

Section: Rmlmentioning

confidence: 99%

Revising the stochastic iterative ensemble smoother

Raanes

Stordal

Evensen

2019

Nonlin. Processes Geophys.

View full text Add to dashboard Cite

show abstract

“…Gratton et al (2013) discuss application of GN in a trust region framework, which has the limitation that a portion of the computationally expensive outer loop increments will be rejected. Some authors have successfully applied the Levenberg-Marquardt algorithm in EnKF (e.g., Chen and Oliver, 2013;Mandel et al, 2016) by adding a regularization term to the cost function. That method requires one to perform the inner loop approximation of [H δv ] −1 multiple times, once for each value of a scalar regularization parameter.…”

Section: Nonlinear Optimizationmentioning

confidence: 99%

Four-dimensional variational inversion of black carbon emissions during ARCTAS-CARB with WRFDA-Chem

Guerrette

Henze

2017

Atmos. Chem. Phys.

View full text Add to dashboard Cite

Abstract. Biomass burning emissions of atmospheric aerosols, including black carbon, are growing due to increased global drought, and comprise a large source of uncertainty in regional climate and air quality studies. We develop and apply new incremental four-dimensional variational (4D-Var) capabilities in WRFDA-Chem to find optimal spatially and temporally distributed biomass burning (BB) and anthropogenic black carbon (BC) aerosol emissions. The constraints are provided by aircraft BC concentrations from the Arctic Research of the Composition of the Troposphere from Aircraft and Satellites in collaboration with the California Air Resources Board (ARCTAS-CARB) field campaign and surface BC concentrations from the Interagency Monitoring of PROtected Visual Environment (IMPROVE) network on 22, 23, and 24 June 2008. We consider three BB inventories, including Fire INventory from NCAR (FINN) v1.0 and v1.5 and Quick Fire Emissions Database (QFED) v2.4r8. On 22 June, aircraft observations are able to reduce the spread between a customized QFED inventory and FINNv1.0 from a factor of 3.5 (×3.5) to only ×2.1. On 23 and 24 June, the spread is reduced from ×3.4 to ×1.4. The posterior corrections to emissions are heterogeneous in time and space, and exhibit similar spatial patterns of sign for both inventories. The posterior diurnal BB patterns indicate that multiple daily emission peaks might be warranted in specific regions of California. The US EPA's 2005 National Emissions Inventory (NEI05) is used as the anthropogenic prior. On 23 and 24 June, the coastal California posterior is reduced by ×2, where highway sources dominate, while inland sources are increased near Barstow by ×5. Relative BB emission variances are reduced from the prior by up to 35 % in grid cells close to aircraft flight paths and by up to 60 % for fires near surface measurements. Anthropogenic variance reduction is as high as 40 % and is similarly limited to sources close to observations. We find that the 22 June aircraft observations are able to constrain approximately 14 degrees of freedom of signal (DOF), while surface and aircraft observations together on 23/24 June constrain 23 DOF. Improving hourly-to daily-scale concentration predictions of BC and other aerosols during BB events will require more comprehensive and/or targeted measurements and a more complete accounting of sources of error besides the emissions.

show abstract

Section: Nonlinear Optimizationmentioning

confidence: 99%

“…Open biomass burning (BB), which includes natural wildfires, deforestation, and agricultural waste and prescribed burning, accounts for 40 % of total global BC emissions, while anthropogenic energy related sources (e.g., on-and off-road diesel and gasoline engines, industrial coal, residential cooking and heating) make up the remaining 60 % (Bond et al, 2013). Future climate conditions that increase drought and fire prevalence (e.g., Spracklen et al, 2009) and increasingly regulated anthropogenic sources might lead to a reversal of these ratios in California (Mao et al, 2011) and globally (Jolly et al, 2015). In California, BB events have been shown to increase surface PM 2.5 concentrations by a factor of 3 to 5 (×3 to ×5), compared to non-fire periods (Wu et al, 2006).…”

Section: Introductionmentioning

confidence: 99%

Four dimensional variational inversion of black carbon emissions during ARACTAS-CARB with WRFDA-Chem

Guerrette¹,

Henze²

2016

Preprint

View full text Add to dashboard Cite

<p><strong>Abstract.</strong> Biomass burning emissions of atmospheric aerosols, including black carbon, are growing due to increased global drought, and comprise a large source of uncertainty in regional climate and air quality studies. We develop and apply new incremental 4D-Var capabilities in WRFDA-Chem to find optimal spatially and temporally distributed biomass burning (BB) and anthropogenic black carbon (BC) aerosol emissions. The constraints are provided by aircraft BC concentrations from the Arctic Research of the Composition of the Troposphere from Aircraft and Satellites in collaboration with the California Air Resources Board (ARCTAS-CARB) field campaign and surface BC concentrations from the Interagency Monitoring of PROtected Visual Environment (IMPROVE) network on 22, 23, and 24 June, 2008. We consider multiple BB inventories, including Fire INventory from NCAR (FINN) v1.0 and v1.5 and Quick Fire Emissions Database (QFED) v2.4r8. On 22 June, aircraft observations are able to reduce the spread between QFED&#8201;&#215;&#8201;&#8531; and FINNv1.0 from &#215;3.5 to &#215;2.1. On 23 and 24 June, the spread is reduced from &#215;3.4 to &#215;1.4. The posterior corrections to emissions are heterogenous in time and space, and exhibit similar spatial patterns of sign for both inventories. The posterior diurnal BB patterns indicate that multiple daily emission peaks might be warranted in specific regions of California. The U.S. EPA's 2005 National Emissions Inventory (NEI05) is used as the anthropogenic prior. On 23 and 24 June, the coastal California posterior is scaled by &#215;&#8201;&#189;, where highway sources dominate, while inland sources are increased near Barstow by &#215;5. Relative BB emission variances are reduced from the prior by up to 35&#8201;% in grid cells close to aircraft flight paths and up to 60&#8201;% for fires near surface measurements. Anthropogenic variance reduction is as high as 40&#8201;% and is similarly limited to sources close to observations. We find that the 22 June aircraft observations are able to constrain approximately 14 degrees of freedom of signal (DOF), while surface and aircraft observations together on 23/24 June constrain 23 DOF. Improving hourly to daily scale concentration predictions of BC and other aerosols during BB events will require more comprehensive and/or targeted measurements and a more complete accounting of sources of error besides the emissions.</p>

show abstract

Hybrid Levenberg–Marquardt and weak-constraint ensemble Kalman smoother method

Cited by 20 publications

References 45 publications

Revising the stochastic iterative ensemble smoother

Revising the stochastic iterative ensemble smoother

Four-dimensional variational inversion of black carbon emissions during ARCTAS-CARB with WRFDA-Chem

Four dimensional variational inversion of black carbon emissions during ARACTAS-CARB with WRFDA-Chem

Contact Info

Product

Resources

About