STUDY ON SPATIOTEMPORAL CHARACTERISTIC OF OBSERVATION IMPACT IN PROXY DATA ASSIMILATION

Abstract Two types of approaches are commonly used for estimating the impact of arbitrary subsets of observations on short-range forecast error. The first was developed for variational data assimilation systems and requires the adjoint of the forecast model. Comparable approaches were developed for use with the ensemble Kalman filter and rely on ensembles of forecasts. In this study, a new approach for computing observation impact is proposed for ensemble–variational data assimilation (EnVar). Like standard adjoint approaches, the adjoint of the data assimilation procedure is implemented through the iterative minimization of a modified cost function. However, like ensemble approaches, the adjoint of the forecast step is obtained by using an ensemble of forecasts. Numerical experiments were performed to compare the new approach with the standard adjoint approach in the context of operational deterministic NWP. Generally similar results are obtained with both approaches, especially when the new approach uses covariance localization that is horizontally advected between analysis and forecast times. However, large differences in estimated impacts are obtained for some surface observations. Vertical propagation of the observation impact is noticeably restricted with the new approach because of vertical covariance localization. The new approach is used to evaluate changes in observation impact as a result of the use of interchannel observation error correlations for radiance observations. The estimated observation impact in similarly configured global and regional prediction systems is also compared. Overall, the new approach should provide useful estimates of observation impact for data assimilation systems based on EnVar when an adjoint model is not available.

Download Full-text

A reduced-order Kalman smoother for (paleo-)ocean state estimation: assessment and application to the LGM

10.5194/gmd-2019-32 ◽

2019 ◽

Author(s):

Charlotte Breitkreuz ◽

André Paul ◽

Stefan Mulitza ◽

Javier García-Pintado ◽

Michael Schulz

Keyword(s):

Data Assimilation ◽

State Estimation ◽

Adjoint Method ◽

Computing Time ◽

Oxygen Isotopic Composition ◽

Proxy Data ◽

Overturning Circulation ◽

Control Variables ◽

State Reduction ◽

Ocean State Estimation

Abstract. Combining ocean models and proxy data via data assimilation is a powerful means to obtain more reliable estimates of past ocean states, but studies using data assimilation for paleo-ocean state estimation are rare. A few studies used the adjoint method, also called 4D-Var, to estimate the state of the ocean during the Last Glacial Maximum (LGM). The adjoint method, however, requires the adjoint of the model code, which is not easily obtained for most models. The method is computationally very demanding and does not readily provide uncertainty estimates. Here, we present a new and computationally very efficient technique to obtain ocean state estimates. We applied a state reduction approach in conjunction with a finite difference sensitivity-iterative Kalman smoother (FDS-IKS) to estimate spatially varying atmospheric forcing fields and to obtain an equilibrium model simulation in consistency with proxy data. We tested the method in synthetic pseudo-proxy data experiments. The method is capable of very efficiently estimating 16 control variables and reconstructing a target ocean circulation from sea surface temperature (SST) and oxygen isotopic composition of seawater data at LGM coverage. The method is advantageous over the adjoint method regarding that it is very easy to implement, it requires substantially less computing time and provides an uncertainty estimate of the estimated control variables. The computing time, however, depends linearly on the size of the control space limiting the number of control variables that can be estimated. We used the method to investigate the constraint of data outside of the Atlantic Ocean on the Atlantic overturning circulation. Our results indicate that while data from the Pacific or Indian Ocean aid in correctly estimating the Atlantic overturning circulation, they are not as crucial as the Atlantic data. We additionally applied the method to estimate the LGM ocean state constrained by a global SST reconstruction and data on the oxygen isotopic composition of calcite from fossil benthic and planktic foraminifera. The LGM estimate shows a large improvement compared to our first guess, but model-data misfits remain after the optimization due to model errors that cannot be corrected by the control variables. The estimate shows a shallower North Atlantic Deep Water and a weaker Atlantic overturning circulation compared to today in consistency with previous studies. The combination of the FDS-IKS and the state reduction approach is a step forward in making ocean state estimation and data assimilation applicable for complex and computationally expensive models and to models where the adjoint is not available.

Download Full-text

Some Experimental Lessons on Digital Filtering in the ALADIN-France 3DVAR Based on Near-Ground Examination

Monthly Weather Review ◽

10.1175/2010mwr3388.1 ◽

2011 ◽

Vol 139 (3) ◽

pp. 774-785 ◽

Cited By ~ 4

Author(s):

Claude Fischer ◽

Ludovic Auger

Keyword(s):

Data Assimilation ◽

Digital Filter ◽

Three Dimensional ◽

Weather Forecast ◽

Forecast Model ◽

Horizontal Wind ◽

Screen Level ◽

Incremental Formulation ◽

Weather Forecast Model ◽

Observation Impact

Abstract This paper deals with the characteristics and effects of digital filter initialization, as implemented in the operational three-dimensional variational data assimilation (3DVAR) system of the Aire Limitée Adaptation Dynamique Développement International (ALADIN)-France regional weather forecast model. First, a series of findings on the properties of the initialization of the model are discussed. Examples of initial spinup linked with inertia–gravity wave occurrence are shown, and the major sources for their generation are listed. These experimental results are compared with past and present experiences concerning the use and need for digital filter initialization. Furthermore, the impacts of switching to an incremental formulation of the filter in data assimilation mode are demonstrated. Second, the effects of the filter formulation on the results of an observation impact study are illustrated. The latter consists of implementing screen-level, 10-m horizontal wind information into the ALADIN 3DVAR analysis. There can, indeed, be some delicate interference between observation impact evaluation and the effects of filtering, at least on short-term forecasts. The paper is concluded with some general considerations on the experimental evaluation of spinup and the link between the assimilation system design and model state filtering.

Download Full-text

Comparing Two Approaches for Assessing Observation Impact

Monthly Weather Review ◽

10.1175/mwr-d-12-00100.1 ◽

2013 ◽

Vol 141 (5) ◽

pp. 1484-1505 ◽

Cited By ~ 17

Author(s):

Ricardo Todling

Keyword(s):

Data Assimilation ◽

State Space ◽

Adjoint Operator ◽

Additional Advantage ◽

Earth Observing System ◽

Natural Choice ◽

Observation Impact ◽

Observation Space ◽

The Impact ◽

Space Approach

Abstract Langland and Baker introduced an approach to assess the impact of observations on the forecasts. In that approach, a state-space aspect of the forecast is defined and a procedure is derived ultimately relating changes in the aspect with changes in the observing system. Some features of the state-space approach are to be noted: the typical choice of forecast aspect is rather subjective and leads to incomplete assessment of the observing system, it requires availability of a verification state that is in practice correlated with the forecast, and it involves the adjoint operator of the entire data assimilation system and is thus constrained by the validity of this operator. This article revisits the topic of observation impacts from the perspective of estimation theory. An observation-space metric is used to allow inferring observation impact on the forecasts without the limitations just mentioned. Using differences of observation-minus-forecast residuals obtained from consecutive forecasts leads to the following advantages: (i) it suggests a rather natural choice of forecast aspect that directly links to the data assimilation procedure, (ii) it avoids introducing undesirable correlations in the forecast aspect since verification is done against the observations, and (iii) it does not involve linearization and use of adjoints. The observation-space approach has the additional advantage of being nearly cost free and very simple to implement. In its simplest form it reduces to evaluating the statistics of observation-minus-background and observation-minus-analysis residuals with traditional methods. Illustrations comparing the approaches are given using the NASA Goddard Earth Observing System.

Download Full-text

Forecast Sensitivity Observation Impact in the 4DVAR and Hybrid-4DVAR Data Assimilation Systems

Journal of Atmospheric and Oceanic Technology ◽

10.1175/jtech-d-18-0240.1 ◽

2019 ◽

Vol 36 (8) ◽

pp. 1563-1575 ◽

Cited By ~ 2

Author(s):

Sung-Min Kim ◽

Hyun Mee Kim

Keyword(s):

Data Assimilation ◽

East Asia ◽

Forecast Error ◽

Integration Time ◽

Ensemble Forecasts ◽

Background Error ◽

Error Covariance ◽

Atmospheric Motion ◽

Observation Impact ◽

Atmospheric Motion Vectors

AbstractIn this study, the observation impacts on 24-h forecast error reduction (FER), based on the adjoint method in the four-dimensional variational (4DVAR) data assimilation (DA) and hybrid-4DVAR DA systems coupled with the Unified Model, were evaluated from 0000 UTC 5 August to 1800 UTC 26 August 2014. The nonlinear FER in hybrid-4DVAR was 12.2% greater than that in 4DVAR due to the use of flow-dependent background error covariance (BEC), which was a weighted combination of the static BEC and the ensemble BEC based on ensemble forecasts. In hybrid-4DVAR, the observation impacts (i.e., the approximated nonlinear FER) for most observation types increase compared to those in 4DVAR. The increased observation impact from using hybrid-4DVAR instead of 4DVAR changes depending on the analysis time and regions. To calculate the ensemble BEC in hybrid-4DVAR, analyses at 0600 and 1800 UTC (0000 and 1200 UTC) used 3-h (9-h) ensemble forecasts. Greater observation impact was obtained when 3-h ensemble forecasts were used for the ensemble BEC at 0600 and 1800 UTC, than with 9-h ensemble forecasts at 0000 and 1200 UTC. Different from other observations, the atmospheric motion vectors (AMVs) deduced from geostationary satellite are more frequently observed in the same area. When the ensemble forecasts with longer integration times were used for the ensemble BEC in hybrid-4DVAR, the observation impact of the AMVs decreased the most in East Asia. This implies that the observation impact of AMVs in East Asia shows the highest sensitivity to the integration time of the ensemble members used for deducing the flow-dependent BEC in hybrid-4DVAR.

Download Full-text

Multi-timescale data assimilation for atmosphere–ocean state estimates

Climate of the Past ◽

10.5194/cp-12-1375-2016 ◽

2016 ◽

Vol 12 (6) ◽

pp. 1375-1388 ◽

Cited By ~ 11

Author(s):

Nathan Steiger ◽

Gregory Hakim

Keyword(s):

High Resolution ◽

Data Assimilation ◽

Spectral Characteristics ◽

Low Frequency ◽

Meridional Overturning Circulation ◽

Climate System ◽

Time Averaging ◽

Proxy Data ◽

Multiple Timescales ◽

Frequency Components

Abstract. Paleoclimate proxy data span seasonal to millennial timescales, and Earth's climate system has both high- and low-frequency components. Yet it is currently unclear how best to incorporate multiple timescales of proxy data into a single reconstruction framework and to also capture both high- and low-frequency components of reconstructed variables. Here we present a data assimilation approach that can explicitly incorporate proxy data at arbitrary timescales. The principal advantage of using such an approach is that it allows much more proxy data to inform a climate reconstruction, though there can be additional benefits. Through a series of offline data-assimilation-based pseudoproxy experiments, we find that atmosphere–ocean states are most skillfully reconstructed by incorporating proxies across multiple timescales compared to using proxies at short (annual) or long (∼ decadal) timescales alone. Additionally, reconstructions that incorporate long-timescale pseudoproxies improve the low-frequency components of the reconstructions relative to using only high-resolution pseudoproxies. We argue that this is because time averaging high-resolution observations improves their covariance relationship with the slowly varying components of the coupled-climate system, which the data assimilation algorithm can exploit. These results are consistent across the climate models considered, despite the model variables having very different spectral characteristics. Our results also suggest that it may be possible to reconstruct features of the oceanic meridional overturning circulation based on atmospheric surface temperature proxies, though here we find such reconstructions lack spectral power over a broad range of frequencies.

Download Full-text