Operational Forecasting System Research Articles

Automated model optimisation using the Cylc workflow engine (Cyclops v1.0)

Abstract. Most geophysical models include many parameters that are not fully determined by theory, and can be “tuned” to improve the model's agreement with available data. We might attempt to automate this tuning process in an objective way by employing an optimisation algorithm to find the set of parameters that minimises a cost function derived from comparing model outputs with measurements. A number of algorithms are available for solving optimisation problems, in various programming languages, but interfacing such software to a complex geophysical model simulation presents certain challenges. To tackle this problem, we have developed an optimisation suite (“Cyclops”) based on the Cylc workflow engine that implements a wide selection of optimisation algorithms from the NLopt Python toolbox (Johnson, 2014). The Cyclops optimisation suite can be used to calibrate any modelling system that has itself been implemented as a (separate) Cylc model suite, provided it includes computation and output of the desired scalar cost function. A growing number of institutions are using Cylc to orchestrate complex distributed suites of interdependent cycling tasks within their operational forecast systems, and in such cases application of the optimisation suite is particularly straightforward. As a test case, we applied the Cyclops to calibrate a global implementation of the WAVEWATCH III (v4.18) third-generation spectral wave model, forced by ERA-Interim input fields. This was calibrated over a 1-year period (1997), before applying the calibrated model to a full (1979–2016) wave hindcast. The chosen error metric was the spatial average of the root mean square error of hindcast significant wave height compared with collocated altimeter records. We describe the results of a calibration in which up to 19 parameters were optimised.

On operational flood forecasting system involving 1D/2D coupled hydraulic model and data assimilation

In the context of hydrodynamic modeling, the use of 2D models is adapted in areas where the flow is not mono-dimensional (confluence zones, flood plains). Nonetheless the lack of field data and the computational cost constraints limit the extensive use of 2D models for operational flood forecasting. Multi-dimensional coupling offers a solution with 1D models where the flow is mono-dimensional and with local 2D models where needed. This solution allows for the representation of complex processes in 2D models, while the simulated hydraulic state is significantly better than that of the full 1D model. In this study, coupling is implemented between three 1D sub-models and a local 2D model for a confluence on the Adour river (France). A Schwarz algorithm is implemented to guarantee the continuity of the variables at the 1D/2D interfaces while in situ observations are assimilated in the 1D sub-models to improve results and forecasts in operational mode as carried out by the French flood forecasting services. An implementation of the coupling and data assimilation (DA) solution with domain decomposition and task/data parallelism is proposed so that it is compatible with operational constraints. The coupling with the 2D model improves the simulated hydraulic state compared to a global 1D model, and DA improves results in 1D and 2D areas.

Validation of Mountain Precipitation Forecasts from the Convection-Permitting NCAR Ensemble and Operational Forecast Systems over the Western United States

Abstract Convection-permitting ensembles can capture the large spatial variability and quantify the inherent uncertainty of precipitation in areas of complex terrain; however, such systems remain largely untested over the western United States. In this study, we assess the capabilities of deterministic and probabilistic cool-season quantitative precipitation forecasts (QPFs) produced by the 10-member, convection-permitting (3-km horizontal grid spacing) NCAR Ensemble using observations collected by SNOTEL stations at mountain locations across the western United States and precipitation analyses from PRISM. We also examine the performance of operational forecast systems run by NCEP including the High Resolution Rapid Refresh (HRRR) model, the NAM forecast system with a 3-km continental United States (CONUS) nest, GFS, and the Short-Range Ensemble Forecast system (SREF). Overall, we find that higher-resolution models, such as the HRRR, NAM-3km CONUS nest, and an individual member of the NCAR Ensemble, are more deterministically skillful than coarser models, especially over the narrow interior ranges of the western United States, likely because they better resolve topography and thus better simulate orographic precipitation. The 10-member NCAR Ensemble is also more probabilistically skillful than 13-member subensembles composed of each SREF dynamical core, but less probabilistically skillful than the full 26-member SREF, as a result of insufficient spread. These results should help guide future short-range model development and inform forecasters about the capabilities and limitations of several widely used deterministic and probabilistic modeling systems over the western United States.

Validation of a wave forecast system for Galway Bay

ABSTRACTAn operational wave forecast system for the area of Galway Bay is presented. Three numerical models are implemented: Weather Research and Forecasting (WRF), Wavewatch III (WWIII) and Simulating Waves Nearshore (SWAN). GFS (Global Forecast System) winds, from NCEP (National Centres for Environmental Prediction), are used to force the WWIII model, while WRF winds are used to force the SWAN model. Bathymetries for WWIII and SWAN were obtained from NOAA’s GEODAS (Geophysical Data System) data base and from GEBCO (General Bathymetric Chart of the Oceans), hosted by BODC. The wave generation model, WWIII, is used to simulate waves for the whole North Atlantic, whereas the SWAN model runs for the Irish west coast, with a nested grid for Galway Bay. Forecast products of 7 days are produced daily, with a warm up of 4 days. The validations of the forecast system were made for both wind and wave parameters with data from buoys located in the Irish coast and inside Galway Bay. Comparisons in the time domain as well as a statistical analysis were performed. Overall, the wave forecast system gave good results closely reproducing the real data, with a low error.

Subseasonal Forecasting with an Icosahedral, Vertically Quasi-Lagrangian Coupled Model. Part I: Model Overview and Evaluation of Systematic Errors

Abstract The atmospheric hydrostatic Flow-Following Icosahedral Model (FIM), developed for medium-range weather prediction, provides a unique three-dimensional grid structure—a quasi-uniform icosahedral horizontal grid and an adaptive quasi-Lagrangian vertical coordinate. To extend the FIM framework to subseasonal time scales, an icosahedral-grid rendition of the Hybrid Coordinate Ocean Model (iHYCOM) was developed and coupled to FIM. By sharing a common horizontal mesh, air–sea fluxes between the two models are conserved locally and globally. Both models use similar adaptive hybrid vertical coordinates. Another unique aspect of the coupled model (referred to as FIM–iHYCOM) is the use of the Grell–Freitas scale-aware convective scheme in the atmosphere. A multiyear retrospective study is necessary to demonstrate the potential usefulness and allow for immediate bias correction of a subseasonal prediction model. In these two articles, results are shown based on a 16-yr period of hindcasts from FIM–iHYCOM, which has been providing real-time forecasts out to a lead time of 4 weeks for NOAA’s Subseasonal Experiment (SubX) starting July 2017. Part I provides an overview of FIM–iHYCOM and compares its systematic errors at subseasonal time scales to those of NOAA’s operational Climate Forecast System version 2 (CFSv2). Part II uses bias-corrected hindcasts to assess both deterministic and probabilistic subseasonal skill of FIM–iHYCOM. FIM–iHYCOM has smaller biases than CFSv2 for some fields (including precipitation) and comparable biases for other fields (including sea surface temperature). FIM–iHYCOM also has less drift in bias between weeks 1 and 4 than CFSv2. The unique grid structure and physics suite of FIM–iHYCOM is expected to add diversity to multimodel ensemble forecasts at subseasonal time scales in SubX.

Subseasonal Forecasting with an Icosahedral, Vertically Quasi-Lagrangian Coupled Model. Part II: Probabilistic and Deterministic Forecast Skill

Abstract Subseasonal forecast skill of the global hydrostatic atmospheric Flow-Following Icosahedral Model (FIM) coupled to an icosahedral-grid version of the Hybrid Coordinate Ocean Model (iHYCOM) is evaluated through 32-day predictions initialized weekly using a four-member time-lagged ensemble over the 16-yr period 1999–2014. Systematic biases in forecasts by the coupled system, referred to as FIM–iHYCOM, are described in a companion paper (Part I). This present study (Part II) assesses probabilistic and deterministic model skill for predictions of surface temperature, precipitation, and 500-hPa geopotential height in different seasons at different lead times ranging from 1 to 4 weeks. The coupled model appears to have reasonable agreement with reanalysis in terms of simulated weekly variability in sea surface temperatures, except in extratropical regions because the ocean model cannot explicitly resolve eddies there. This study also describes the ability of the model to simulate midlatitude tropospheric blocking frequency, Madden–Julian oscillation patterns, and sudden stratospheric warming events—all of which have been shown to be relevant on subseasonal time scales. The metrics used here indicate that the subseasonal forecast skill of the model is comparable to that of several operational models, including the National Oceanic and Atmospheric Administration’s (NOAA’s) operational Climate Forecast System version 2 and the European Centre for Medium-Range Weather Forecasts model. Therefore, FIM–iHYCOM—as a participant in NOAA’s Subseasonal Experiment—is expected to add value to multimodel ensemble forecasts produced through this effort.

A Case Study for the Application of an Operational Two-Dimensional Real-Time Flooding Forecasting System and Smart Water Level Gauges on Roads in Tainan City, Taiwan

An operational two-dimensional real-time flood forecasting system has been developed to prevent urban inundation in Taiwan, and it uses the Delft-FEWS (Flood Early Warning System) platform to integrate the SOBEK models to forecast flooding. A new generation of smart water level gauges, integrating power charging, recording, transmission, and sensing into one, can monitor the inundation on roads in real-time. This study took Annan District in Tainan City, Taiwan, as the study object and presented the application of the flood forecasting system and smart water level gauges in the urban environment. The collected data from Storm 0611 and Typhoon Megi in 2016 were used to assess the accuracy of the flood forecasting system. The analysis of three water level stations in the drainage showed that the one-dimensional simulation results were fairly accurate. According to the 18 observed and simulated inundation depths on the roads, the true positive, false positive, false negative, and true negative values were 8, 0, 2, and 8, respectively. The accuracy, sensitivity, and precision of the flood forecasting system were 0.889, 0.800, and 1.000, respectively, indicating that the two-dimensional simulation results were highly accurate.

Influence of MeteOcean processes on MSYM sea level predictions in the Malacca Straits

The South China Sea region, and particularly the Malacca and Singapore Straits, are known by the complex tidal dynamics, which is influenced by the tidal propagation from Pacific and Indian Oceans. In spite of the dynamic complexity, the region is very relevant economically, especially concerning the growing oil drilling activities. To give support to accidental oil spill prevention and response, an operational oil spill forecast system was developed for the Strait of Malacca. The hydrodynamic system validation revealed good results, in general. However, besides all the modeling efforts, some discrepancies between observed and predicted sea levels were identified, mainly during neap tide and for specific tide-gauges. Therefore, the main aim of this study consists in researching the origin of these discrepancies by comparing predictions with available data and exploring their relation with the MeteOcean processes of the region. Initially sea level data for eight tide gauges was explored to get a general overview of the local tidal dynamics, and then the model performance for astronomic tide was assessed. Analysis of meteorological tides was also performed for three tide-gauges located in the Singapore and Malacca Strait, which are under the influence of the northeast or southwest monsoons. The results show that the differences between the observed and predicted sea levels in Singapore Strait are usually due to discrepancies in the meteorological tide induced by the surface wind stress acting over the Taiwan-Singapore axis, while in the Malacca Strait are mainly related with model limitations in reproducing the astronomical tide.

Impact of Coupling with an Ice–Ocean Model on Global Medium-Range NWP Forecast Skill

The importance of coupling between the atmosphere and the ocean for forecasting on time scales of hours to weeks has been demonstrated for a range of physical processes. Here, the authors evaluate the impact of an interactive air–sea coupling between an operational global deterministic medium-range weather forecasting system and an ice–ocean forecasting system. This system was developed in the context of an experimental forecasting system that is now running operationally at the Canadian Centre for Meteorological and Environmental Prediction. The authors show that the most significant impact is found to be associated with a decreased cyclone intensification, with a reduction in the tropical cyclone false alarm ratio. This results in a 15% decrease in standard deviation errors in geopotential height fields for 120-h forecasts in areas of active cyclone development, with commensurate benefits for wind, temperature, and humidity fields. Whereas impacts on surface fields are found locally in the vicinity of cyclone activity, large-scale improvements in the mid-to-upper troposphere are found with positive global implications for forecast skill. Moreover, coupling is found to produce fairly constant reductions in standard deviation error growth for forecast days 1–7 of about 5% over the northern extratropics in July and August and 15% over the tropics in January and February. To the authors’ knowledge, this is the first time a statistically significant positive impact of coupling has been shown in an operational global medium-range deterministic numerical weather prediction framework.

Setting up a water quality ensemble forecast for coastal ecosystems: a case study of the southern North Sea

Abstract Prediction systems, such as the coastal ecosystem models, often incorporate complex non-linear ecological processes. There is an increasing interest in the use of probabilistic forecasts instead of deterministic forecasts in cases where the inherent uncertainties in the prediction system are important. The primary goal of this study is to set up an operational ensemble forecasting system for the prediction of the Chlorophyll-a concentration in coastal waters, using the Generic Ecological Model. The input ensemble is generated from perturbed model process parameters and external forcings through Latin Hypercube Sampling with Dependence. The forecast performance of the ensemble prediction is assessed using several forecast verification metrics that can describe the forecast accuracy, reliability and discrimination. The verification is performed against in-situ measurements and remote sensing data. The ensemble forecast moderately outperforms the deterministic prediction at the coastal in-situ measurement stations. The proposed ensemble forecasting system is therefore a promising tool to provide enhanced water quality prediction for coastal ecosystems which, with further inclusion of other uncertainty sources, could be used for operational forecasting.

Improving the Computational Performance of an Operational Two-Dimensional Real-Time Flooding Forecasting System by Active-Cell and Multi-Grid Methods in Taichung City, Taiwan

An operational two-dimensional real-time flood forecasting system has been developed in Taiwan to prevent urban inundation. This system takes an hour to come up with forecasts for the next three hours, and the resolution of the forecasts is 40 × 40 m. This study used a large urban area of 126 km2 in downtown Taichung City for the case study and adopted the active-cell and multi-grid methods to meet the target by computing from data of a 12-h rainfall within one hour at 20 m × 20 m spatial resolution to provide faster forecasting and more hours for flood preparation. With the active-cell method, the Central Processing Unit (CPU) time was reduced by 65.04% from 659 m 29 s to 230 m 33 s under the 200-year return period storm. Further, with multi-grid methods, the CUP time was reduced by 73.98% from 230 m 33 s to 60 m 0 s. In general, the computing time of this model has been reduced 11-fold. The error validation coefficients of inundation areas were between 89.39~97.45% with an average error of depth between 1.06~3.22 cm.

Predicting Heat Stress to Inform Reef Management: NOAA Coral Reef Watch's 4-Month Coral Bleaching Outlook

The U.S. National Oceanic and Atmospheric Administration’s (NOAA) Coral Reef Watch (CRW) operates a global Four-Month Coral Bleaching Outlook system for shallow-water coral reefs in collaboration with NOAA’s National Centers for Environmental Prediction (NCEP). The Outlooks are generated by applying the algorithm used in CRW’s operational satellite coral bleaching heat stress monitoring, with slight modifications, to the sea surface temperature (SST) predictions from NCEP’s operational Climate Forecast System Version 2 (CFSv2). Once a week, the probability of heat stress capable of causing mass coral bleaching is predicted for four months in advance. Each day, CFSv2 generates an ensemble of 16 forecasts, with nine runs out to 45 days, three runs out to three months, and four runs out to nine months. This results in 28-112 ensemble members produced each week. A composite for each predicted week is derived from daily predictions within each ensemble member. The probability of each of four heat stress ranges (Watch and higher, Warning and higher, Alert Level 1 and higher, and Alert Level 2) is determined from all the available ensemble members for the week to form the weekly probabilistic Outlook. The probabilistic Four-Month Outlook is the highest weekly probability predicted among all the weekly Outlooks during a four-month period for each of the stress ranges. An initial qualitative skill analysis of the Outlooks for 2011-2015, compared with CRW’s satellite-based coral bleaching heat stress products, indicated the Outlook has performed well with high hit rates and low miss rates for most coral reef areas. Regions identified with high false alarm rates will guide future improvements. This Outlook system, as the first and only freely available global coral bleaching prediction system, has been providing critical early warning to marine resource managers, scientists, and decision makers around the world to guide management, protection, and monitoring of coral reefs since 2012. This has been especially valuable during the third global coral bleaching event that started in mid-2014 and extended into mid-2017. The Outlook system is an integrated component of CRW’s global decision support system for coral bleaching. Recent management actions taken in light of this system are discussed.

The long-term prediction of the oil-contaminated water from the Sanchi collision in the East China Sea

The condensate and bunker oil leaked from the Sanchi collision would cause a persistent impact on marine ecosystems in the surrounding areas. The long-term prediction for the distribution of the oil-polluted water and the information for the most affected regions would provide valuable information for the oceanic environment protection and pollution assessment. Based on the operational forecast system developed by the First Institute of Oceanography, State Oceanic Administration, we precisely predicted the drifting path of the oil tanker Sanchi after its collision. Trajectories of virtual oil particles show that the oil leaked from the Sanchi after it sank is mainly transported to the northeastern part of the sink location, and quickly goes to the open ocean along with the Kuroshio. Risk probability analysis based on the outcomes from the operational forecast system for years 2009 to 2017 shows that the most affected area is at the northeast of the sink location.

Forecasting West African Heat Waves at Subseasonal and Seasonal Time Scales

Abstract Early indication of an increased risk of extremely warm conditions could help alleviate some of the consequences of severe heat waves on human health. This study focuses on boreal spring heat wave events over West Africa and the Sahel and examines the long-range predictability and forecast quality of these events with two coupled forecasting systems designed at Météo-France, both based on the CNRM-CM coupled global climate model: the operational seasonal forecasting System 5 and the experimental contribution to the World Weather Research Programme/World Climate Research Programme (WWRP/WCRP) subseasonal-to-seasonal (S2S) project. Evaluation is based on past reforecasts spanning 22 years, from 1993 to 2014, compared to reference data from reanalyses. On the seasonal time scale, skill in reproducing interannual anomalies of heat wave duration is limited at a gridpoint level but is significant for regional averages. Subseasonal predictability of daily humidity-corrected apparent temperature drops sharply beyond the deterministic range. In addition to reforecast skill measures, the analysis of real-time forecasts for 2016, both in terms of anomalies with respect to the reforecast climatology and using a weather-type approach, provides additional insight on the systems’ performance in giving relevant information on the possible occurrence of such events.

A Prototype Regional GSI-based EnKF-Variational Hybrid Data Assimilation System for the Rapid Refresh Forecasting System: Dual-Resolution Implementation and Testing Results

A dual-resolution (DR) version of a regional ensemble Kalman filter (EnKF)-3D ensemble variational (3DEnVar) coupled hybrid data assimilation system is implemented as a prototype for the operational Rapid Refresh forecasting system. The DR 3DEnVar system combines a high-resolution (HR) deterministic background forecast with lower-resolution (LR) EnKF ensemble perturbations used for flow-dependent background error covariance to produce a HR analysis. The computational cost is substantially reduced by running the ensemble forecasts and EnKF analyses at LR. The DR 3DEnVar system is tested with 3-h cycles over a 9-day period using a 40/∼13-km grid spacing combination. The HR forecasts from the DR hybrid analyses are compared with forecasts launched from HR Gridpoint Statistical Interpolation (GSI) 3D variational (3DVar) analyses, and single LR hybrid analyses interpolated to the HR grid. With the DR 3DEnVar system, a 90% weight for the ensemble covariance yields the lowest forecast errors and the DR hybrid system clearly outperforms the HR GSI 3DVar. Humidity and wind forecasts are also better than those launched from interpolated LR hybrid analyses, but the temperature forecasts are slightly worse. The humidity forecasts are improved most. For precipitation forecasts, the DR 3DEnVar always outperforms HR GSI 3DVar. It also outperforms the LR 3DEnVar, except for the initial forecast period and lower thresholds.

First Successful Hindcasts of the 2016 Disruption of the Stratospheric Quasi‐biennial Oscillation

AbstractIn early 2016 the quasi‐biennial oscillation in tropical stratospheric winds was disrupted by an anomalous easterly jet centered at ~40 hPa, a development that was completely missed by all operational extended range weather forecast systems. This event and its predictability are investigated through 40 day ensemble hindcasts using a global model notable for its sophisticated representation of the upper atmosphere. Integrations starting at different times throughout January 2016—just before and during the initial development of the easterly jet—were performed. All integrations simulated the unusual developments in the stratospheric mean wind, despite considerable differences in other aspects of the flow evolution among the ensemble members, notably in the evolution of the winter polar vortex and the day‐to‐day variations in extratropical Rossby waves. Key to prediction of this event is simulating the slowly evolving mean winds in the winter subtropics that provide a waveguide for Rossby waves propagating from the winter hemisphere.

Assimilation of Feng-Yun-3B satellite microwave humidity sounder data over land

The ECMWF has been assimilating Feng-Yun-3B (FY-3B) satellite microwave humidity sounder (MWHS) data over ocean in an operational forecasting system since 24 September 2014. It is more difficult, however, to assimilate microwave observations over land and sea ice than over the open ocean due to higher uncertainties in land surface temperature, surface emissivity and less effective cloud screening. We compare approaches in which the emissivity is retrieved dynamically from MWHS channel 1 [150 GHz (vertical polarization)] with the use of an evolving emissivity atlas from 89 GHz observations from the MWHS onboard NOAA and EUMETSAT satellites. The assimilation of the additional data over land improves the fit of short-range forecasts to other observations, notably ATMS (Advanced Technology Microwave Sounder) humidity channels, and the forecast impacts are mainly neutral to slightly positive over the first five days. The forecast impacts are better in boreal summer and the Southern Hemisphere. These results suggest that the techniques tested allow for effective assimilation of MWHS/FY-3B data over land.

Dominant Modes of Subseasonal Variability of East Asian Summertime Surface Air Temperature and Their Predictions

Subseasonal variability of surface air temperature (SAT) over East Asia is analyzed using the NCEP–NCAR reanalysis of 34 Northern Hemisphere extended summers. An empirical orthogonal function (EOF) analysis is performed with pentad SAT data to identify the leading modes of subseasonal SAT variability. The first (EOF1) and second (EOF2) modes, which together account for about 35% of the total variance, correspond to a monopole structure of SAT anomaly in the whole East Asian region and a dipole structure with opposite signs of variability over the north and south East Asian continent, respectively. Lead–lag regressions are calculated in order to analyze how the large-scale atmospheric circulation evolves in association with the development of the leading SAT modes. An eastward propagation of the Rossby wave from the midlatitude Atlantic Ocean is observed about three pentads before EOF1. EOF2 is influenced by both the tropical Madden–Julian oscillation (MJO) and a midlatitude wave train. These results indicate that there is potential for prediction of the dominant SAT modes on the subseasonal time scale. The subseasonal prediction of the two dominant modes is further evaluated in the operational monthly forecasting system of Environment and Climate Change Canada (ECCC). The model shows a better forecast skill than the persistence forecast. The strength of the subseasonal signal in initial conditions impacts the forecast skill. The forecasts starting with strong EOF in the initial condition are more skillful than those initialized with weak EOF. The findings in the study contribute to improving the understanding of the subseasonal variability and SAT subseasonal forecasting in East Asia.

ENSO and Sea Surface Temperature Anomalies in Association with Canadian Wheat Yield Variability

ABSTRACTInterannual variations of spring wheat yields in Canadian agricultural regions are analyzed, together with the associated sea surface temperature (SST) anomalies in the northern hemisphere tropics and extratropics, from 1961 to 2015. The cubic trend is calculated and used to represent the trend related to advances in agricultural technology over this time period. The correlations between Canadian wheat yields at regional scales and the tropical El Niño–Southern Oscillation (ENSO) variability are not robust at any stage of the evolution of ENSO. Based on the power spectrum and cross-spectrum analysis, the most prominent yield variance is found in the Canadian Prairies, with a significant power peak of 4.5 years but does not co-vary significantly with interannual ENSO variability. ENSO weakly affects temperature and precipitation anomalies in the Canadian Prairie Region in summer—two important agroclimatic conditions for crop growth—and hence insignificantly impacts wheat yields. This indicates that there would be little benefit to including tropical ENSO indices in the operational wheat yield forecasting system. For Canadian wheat yield forecasting, attention should be paid to the preceding winter and spring SST anomalies in the northern extratropics. The SST anomalies associated with yields in the Canadian Prairie region and Central Region are generally stronger than those associated with yields in the Canadian Pacific Coast Region and eastern Maritime Region. In association with the Prairie Region and Central Region yields, SST shows pronounced anomalies in the mid-high latitudes of the North Pacific from winter to summer. The non-linearity of the SST anomalies associated with the Canadian yields is also clearly evident. Stronger (weaker) SST anomalies in the extratropical North Pacific correspond to low wheat yields in the Prairie (Central) Region, while weaker (stronger) SST anomalies correspond to high yields in the Prairie (Central) Region.

Optimal and scalable methods to approximate the solutions of large‐scale Bayesian problems: theory and application to atmospheric inversion and data assimilation

This article provides a detailed theoretical analysis of methods to approximate the solutions of high‐dimensional (>106) linear Bayesian problems. An optimal low‐rank projection that maximizes the information content of the Bayesian inversion is proposed and efficiently constructed using a scalable randomized SVD algorithm. Optimality results for the associated posterior error covariance matrix and posterior mean approximations obtained in previous studies are revisited and tested in a numerical experiment consisting of a large‐scale atmospheric tracer transport source‐inversion problem. This method proves to be a robust and efficient approach to dimension reduction, as well as a natural framework to analyze the information content of the inversion. Possible extensions of this approach to the nonlinear framework in the context of operational numerical weather forecast data assimilation systems based on the incremental 4D‐Var technique are also discussed, and a detailed implementation of a new Randomized Incremental Optimal Technique (RIOT) for 4D‐Var algorithms leveraging our theoretical results is proposed.