Workshop on Future US Earth System Reanalysis

The UK Met Office Hadley Centre produces various marine datasets (https://www.metoffice.gov.uk/hadobs/). Here we provide an overview of the HadIOD dataset.

HadIOD (Hadley Centre Integrated Ocean Database) is a database of global historical in situ ocean temperature and salinity observations from 1850-present. It brings together observations made by surface-only observing platforms (like ships and buoys) and observations from sub-surface ocean profiling platforms (like profiling floats and bathythermographs) and supplements these with metadata including quality flags, bias corrections and estimates of measurement uncertainty.

A target use for HadIOD is ocean and coupled reanalyses where climate-quality observations from the sub-surface ocean and near the ocean-atmosphere interface are needed for assimilation. It is important that ocean data users correct for known biases (which introduce artificial variability) and explore their application’s sensitivity to observational uncertainties. By combining this information in a single dataset, we also allow users to more easily utilize current understanding of observation error in their applications.

HadIOD brings together marine observations and metadata from various sources and thus relies on the work of many other people. It is intended to complement existing data collections.

For HadIOD data and a comprehensive user guide see: https://www.metoffice.gov.uk/hadobs/hadiod/. We welcome feedback from potential users.

You will need this passcode for the poster session: XiV8RZb5Re7

Nick Rayner

Met Office Hadley Centre

Please watch this 5 minute video: https://www.youtube.com/watch?v=OPpSHFvbkfE

The Met Office Hadley Centre Sea-surface temperature and sea ice data set, HadISST, version 2.4 provides a 20+ member ensemble of globally complete SST and sea ice concentrations for the period 1850 to present. It improves upon earlier versions of the data set used in 20CRv3, ERA-20C, CERA-20C and ERA5 reanalyses.

The data set is based on state-of-the-art satellite SSTs from the ESA SST CCI with 172 years of in situ data from ICOADSr3. Input data are bias-adjusted using buoys as a reference.

Interpolation is in two steps: a large-scale, low-resolution reconstruction using the full uncertainty model; and a mid-scale reconstruction using a non-stationary local covariance at 1° resolution.

Samples are drawn at each stage to generate a 20-member ensemble of gridded datasets. Each ensemble member is consistent with available observations, is homogeneous and has realistic variability from 1870. Reconstructed variability in the Pacific prior to 1870 is lower due to a lack of observations. The grids can be downscaled to 0.25° daily resolution.

By construction, the data set is structurally distinct from comparable datasets - like NOAA’s ERSST and JMA’s COBE-SST-2 - and would thus provide a complementary dataset for assessing overall uncertainty.

You will need this passcode for the poster session: XiV8RZb5Re7

Nick Rayner

Met Office Hadley Centre

NOAA/PSL and CIRES/CU has created a set of Ocean Web-based Reanalysis Tools (WRIT). The tools are designed to allow users to easily examine and compare ocean reanalyses and observational data. Data can be examined and analyzed prior to potential downloading. Tools include a map interface which allows users to visualize the data via maps of means, anomalies, and climatologies of variables an vertical cross-sections of variables at depth. Users can average together multiple years/seasons and also create difference maps, and cross sections. There is a time-series tool to extract/plot time-series from the gridded fields and read ocean/atmospheric index time-series including for example Nino3.4, PDO, or station based tidal data. Users can compare time-series, compute statistics, and plot wavelets or autocorrelations. New features include looking at runs, sorting output, and examining extremes. There is also a correlation page which corre lates climate index time-series with ocean variables producing maps and vertical cross-sections. Users can upload their own time-series on the correlation and time-series pages for analysis and plotting. Finally there is a profile page which plots vertical profiles, transects, and time by depth plots of ocean analyses. Ocean variables include temperature, salinity, and u/v currents at depth. Single level variables include SST, heat flux, sea surface height, wind fluxes, bottom temperature, and ice. Reanalyses currently included are the NOAA/GODAS, ECWMF ORAS5, ECMWF ORA20C, NASA ECCO V4, GLORYS, SODA 3.4.2 and SODA 3.7.2, BRAN2020. Observational datasets include EN4 temperature/salinity observation and the NOAA OISST, ERSSTv5 and HadISST SST datasets and GPCP Precipitation. The tools are designed to be easy to use and provide output in multiple formats. Documentation provides links to the original datasets. The webpages are available at https://psl.noaa.gov/data/writ/.

Cathy Smith

NOAA/PSL

Dramatic increase in atmospheric CO2 is the primary driver of climate change. The growth of atmospheric CO2 is driven by anthropogenic emissions, but partially compensated for by sinks of carbon into the land biosphere and oceans. The drivers of CO2 growth is a complex function of anthropogenic, terrestrial, and oceanic processes. Patterns of climate variability directly affect the airborne fraction through spatially-complex processes such as fires, gross primary productivity, and respiration, while atmosphere-ocean CO2 exchanges are modulated across entire ocean basins. Changes in climate frequency and intensity alters net carbon exchange, leading to carbon-climate feedbacks. The co-location of anthropogenic and natural fluxes complicates the attribution of CO2 changes to mitigation strategies, as proposed by the Paris Climate Accord.

The NASA Carbon Monitoring System Flux (CMS-Flux) project advances our ability to attribute the CO2 growth rate to constituent fluxes through a consortium of assimilation systems that ingest observations spanning atmospheric, terrestrial, oceanic, and anthropogenic carbon. This consortium includes GEOS-Chem, CARDAMOM, ECCO-Darwin, and MOMO-Chem. We survey the capability and scientific results of CMS-Flux over the last decade to quantify climate variability on carbon exchange, their impacts on carbon-climate feedbacks, and implications for future Earth System Science reanalysis and carbon mitigation strategies.

Kevin Bowman <kevin.bowman@jpl.nasa.gov>

Jet Propulsion Laboratory

Decadal and longer trends have been suspect in atmospheric reanalyses because of the lack of consistency between the different reanalyses. The new Conventional Observation Reanalysis(CORe) shows a good agreement with JRA-55 and ERA-5 trends for large regional averages in the satellite period (1979-2020). These three reanalyses use three different forecast models, two different data assimilation methods (Ensemble Kalman Filter, 4-D Var), and two different sets of observations (in-situ and atmospheric vectors, all observations that their systems can ingest). That these diverse data assimilation systems are producing similar trends, suggests that we are beginning to resolve the trends. In the pre-satelite period, CORe, JRA-55 and ERA-5 (BE, preliminary) show more differences.

Wesley Ebisuzaki

NOAA/NWS/NCEP/CPC

Atmospheric reanalyses can use external boundary conditions such as sea-surface temperature, sea ice and snow. The Conventional Observation Reanalysis (CORe) used the USAF snow depth analyses. The CORe project addressed the lack of snow analyses prior to 1979 by using the model's snow forecast. However, the snow forecast had to be adjusted otherwise the snow cover would be 3% too large (based on 2014 results). This would have led to a spurious cooling in the pre-1979 analyses. A simple correction scheme was developed to reduce much of this cooling.

Another problem was traced to the use of the observed snow depth. If the model was too warm, the snow would melt and increase the soil moisture. The next day, the model would be forced with the observed snow analyses, the snow would melt and again increase the snow moisture. The net result would be that scattered points would have enormously high soil moisture. This is a problem caused by a satellite-based analysis detecting snow cover from higher elevations than the model's surface elevation. This would always be a problem when the models do not resolve the mountains. This is a problem that needs to be addressed.

Wesley Ebisuzaki

NOAA/NWS/NCEP/CPC

Physically-consistent model-data syntheses provide insight into workings of dynamic systems difficult to attain otherwise. In particular, processes estimated along with the states elucidate mechanisms governing the observed systems. Here, we illustrate such products and tools of the "Estimating the Circulation and Climate of the Ocean" consortium (ECCO; https://ecco-group.org). Property budgets help ascertain controlling processes and adjoint models, in addition to their utility in estimation, quantify causal relationships. The budgets call for complete term-by-term estimation of the fluxes involved and adjoint modeling requires ready evaluation of pertinent model sensitivities. We demonstrate the utility of these products and tools in identifying mechanisms responsible for changes in the Arctic's Beaufort Sea. During the last three decades, freshwater content of the Beaufort Sea has risen 40%, rivaling that of the Great Salinity Anomaly of the 1970s. The analysis reveals winds altering both ocean circulation and sea-ice melt, resulting in the changes observed. Whereas direct wind-driven kinematic anomalies decay over weeks, the sea-ice-melt-driven diabatic change will persist for years to come. ECCO invites application of its estimation system and welcomes collaboration in advancing physically-consistent climate reanalysis.

Ichiro Fukumori <fukumori@jpl.nasa.gov>

Jet Propulsion Laboratory, California Institute of Technology

Historical reanalyses have become a widely used resource for analyzing weather and climate processes and their changes over time. In this poster I explore how further historical observations could improve historical reanalyses. Using version 3 of the "Twentieth Century Reanalysis" (20CRv3), I estimate the benefit of additional observations using an off-line assimilation approach with de-biased observations. For the case of the year 1807, which was an extreme heat year in Europe, assimilating additional pressure data improves the skill for pressure (though not for temperature), while assimilating temperature data improves the skill for temperature (but not for pressure). Assimilating both leads to substantially increased skill in both variables in a leave-one-out approach. Using the example of El Nino-associated drought in South Africa in 1877/78, I demonstrate the value of assimilating further ship-based pressure observations. Finally, the weather in Europe in 1926/27 is used to analyze the benefit of assimilating upper air and total column ozone observations. Both lead to improvements at the tropopause and in the middle troposphere, but not in the lower troposphere, where 20CRv3 is already nearly perfect. Finally, I show that a backward extension to 1781 seems possible, but further data rescue efforts are necessary.

Stefan Brönnimann

University of Bern

Feedbacks between convection and its thermodynamic environment play a crucial role in determining the distribution, evolution, and organization of tropical convection. This study examines thermodynamic-convection coupling in observations and reanalyses, and attempts to establish process level benchmarks needed to guide model development. Thermodynamic profiles obtained from the NOAA Integrated Global Radiosonde Archive, COSMIC-1 GPS radio occultations, and several reanalyses are examined alongside precipitation estimates from the Tropical Rainfall Measuring Mission. Cyclical increases and decreases in a bulk measure of lower tropospheric convective instability are shown to be coupled to the cyclical amplification and decay of convection. In situ and satellite observations differ systematically from reanalyses in their depictions of lower tropospheric temperature and moisture variations throughout these convective cycles. When using reanalysis thermodynamic fields, these systematic differences cause variations in lower free tropospheric saturation deficit to appear less influential in determining the strength of convection than is suggested by observations. Disagreements amongst reanalyses, as well as between reanalyses and observations, pose significant challenges to process level assessments of thermodynamic-convection coupling, which are needed to improve model representation of tropical convective variability.

Brandon Wolding <brandon.wolding@noaa.gov>

CIRES / NOAA

The Northeast U.S. Shelf (NES) Large Marine Ecosystem, extending from the Gulf of Maine to Cape Hatteras, is a dynamic region supporting some of the most commercially valuable fisheries in the world. The region is experiencing a dramatic change in response to increased fishing pressure and climate change, the combined effects of which create a significant challenge for fisheries stock assessment in this region. This study aims to provide a systematic assessment of several intermediate-to-high spatial resolution global ocean reanalysis products (CFSRv1v2, ECCO V5, ORAS5, SODA3.12.2, Bran2020, Glorys12v1, HYCOM3.0, and HYCOM3.1) against available in-situ and satellite observations. In-situ observations include tide gauges and temperature and salinity from various sources including shipboard hydrographic data, buoys and moorings on the NES. Our comparisons focus on the annual and interannual variability of ocean circulation, temperature and salinity, and the warming trends for the NES. Overall, the global ocean reanalyses products exhibit limited skill in the coastal environment. Higher-resolution reanalyses perform better than the coarser-resolution products. Of the higher resolution products, GLORYS has the best comparisons in key variables such as SSH, bottom, and water column temperature in the continental shelf.

Alma Carolina Castillo Trujillo

Woods Hole Oceanographic Institution

Despite decades of research, fundamental questions about the convectively coupled Kelvin waves (CCKWs) remain not fully answered, including the destabilization mechanisms and the mean state modulation. To deepen our understanding and to test simple models for CCKWs, we examine CCKW precipitation, vertical structure, and energetics in four modern reanalyses (RAs): ERA5, NASA MERRA-2, NCEP CFSR, and JRA-55.

The CCKW precipitation signal strength in the wavenumber-frequency domain and the geographical distribution of CCKW precipitation are reasonably represented in all RAs, although they commonly underestimate the amplitude of CCKW precipitation. Despite considerable inter-RA differences in the vertical structure of temperature and diabatic heating anomalies, the eddy available potential energy (EAPE) generation within the CCKWs is found to be associated with the second baroclinic mode whereas the first baroclinic mode damps CCKW EAPE in three out of four RAs. Geographically, the positive CCKW EAPE generation within the second baroclinic mode occurs in areas with high mean state sea surface temperature. Our results are supportive of the simple models for CCKWs in which CCKWs are destabilized within the second baroclinic mode component. Our results suggest that improving representation of stratiform processes is crucial for a better representation of CCKWs in RAs.

Mu-Ting Chien

University of Washington

Environment and Climate Change Canada is advancing with its Global and Regional Reforecasting and Surface and Precipitation Reanalysis System. In versions 2.0 and 2.1 of the reanalysis, the Global Deterministic Reforecasting System (GDRS) based on GEM Atmospheric model at 39 km resolution is initiated by ERA-Interim and its outputs are dynamically downscaled to 10 km resolution by Regional Deterministic Reforecasting System (RDRS). RDRS is coupled with Canadian Land Data Assimilation System (CaLDAS) and Canadian Precipitation Analysis (CaPA) to produce the surface and precipitation analysis. The version 2.0 is covering 2000-2018 period while version 2.1 is for the period between 1980 and 2018; the only difference in model setup between the 2 versions is bug correction related to the maximum snow density. The version 3.0 of the reanalysis system, presently under development, will replace ERA-Interim with ERA 5 and upgrade to the most recent version of the GEM model (GEM 5). The preliminary evaluation of switching from ERA-I to ERA5 will be presented as well as improvements related to GEM 5. The advantages and challenges related to supplementary hourly and 24-hourly precipitation databases not assimilated by operational analysis, especially Integrated Surface Database (ISD), used for 1980-2000 period, will also be discussed and illustrated.

Milena Dimitrijevic

ECCC

To improve the understanding of emission variability and the processes controlling the atmospheric composition, the Tropospheric Chemistry Reanalysis version 2 (TCR-2) was conducted for the period 2005-2020 at 1.1_ horizontal resolution based on an assimilation of multi-constituent observations of ozone, NO2, CO, HNO3, and SO2 from multiple satellite sensors. Surface emissions and the chemical concentrations of various species are simultaneously optimized using an EnKF data assimilation, which was efficient for the correction of the entire tropospheric profile of various species and its year-to-year variations. The evaluation results demonstrate the capability of the chemical reanalysis to improve understanding of the processes controlling variations in atmospheric composition, including long-term changes in near-surface air quality and emissions. The estimated emissions can be employed for the elucidation of detailed distributions of the anthropogenic and biomass burning emissions of co-emitted species in all major regions. Chemical reanalysis can play a crucial role in assessing the changes and efficacy of short-lived climate pollutants (SLCP), that are an increasingly important component of greenhouse gas budgets, and evaluating climate model simulations. Improving the observational constraints in tropospheric chemistry reanalysis, including the continued development of satellite observing systems, will provide important implications for future Earth System Science reanalysis.

Kazuyuki Miyazaki

NASA JPL

In collaboration with the Global Modeling and Assimilation Office at NASA and State University of New York at Albany, NOAA is developing capability to assimilate observations to create the first ever aerosol reanalysis at this institution. The observations include Aerosol Optical Depth (AOD) derived from Moderate Resolution Imaging Spectroradiometer (MODIS) and AErosol RObotic NETwork (AERONET) direct sun measurements. The model relies on Finite-Volume Cubed-Sphere (FV3) dynamical core, Global Forecast System (GFS) physics and the aerosol parameterization based on the Goddard Chemistry Aerosol Radiance and Transport (GOCART). The assimilation tools are from the Joint Effort for Data assimilation Integration (JEDI): the forward operator uses aerosol scattering tables from NASA/GMAO and the assimilation approach combines the variational solver and the Local Ensemble Transform Kalman Filter (LETKF). Reanalysis for year 2016 is compared to NASA�s Modern-Era Retrospective analysis for Research and Applications Two (MERRA-2) and ECMWF�s Copernicus Atmosphere Monitoring Service interim Reanalysis (CAMSiRA) and evaluated against multiwavelength observations from AERONET and other independent satellite AOD retrievals.

Mariusz Pagowski

CIRES, CU Boulder and NOAA/OAR/GSL

Global historical reanalyses are a valuable database for the weather reconstruction community. Using regional weather forecasting models and data assimilation systems, these datasets can be further refined to a regional-to-local scale, allowing a more detailed analysis of past atmospheric processes and surface effects. In this regard, the newest version of the 20th century reanalysis (20CRv3) opens a new chapter, allowing us to go further back in time to the early 19th century. Furthermore, data rescue and digitization activities in the last decades brought to light a vast amount of meteorological station records covering this period. In our work, we combine these sources of information for a high-resolution weather reconstruction of the summer 1816 in Europe - a particularly interesting episode due to the catastrophic aftereffects of the Tambora eruption - using the Weather Research and Forecasting (WRF) model for downscaling 20CRv3 data and a 3DVAR approach for data assimilation. However, dynamical downscaling and data assimilation for the early instrumental period are challenging because of a coarser station network, larger uncertainties of the measurements and a different past land surface, among others. Our poster demonstrates the issues that need to be addressed to obtain accurate weather reconstructions.

Lucas Pfister

University of Bern

Sea surface temperature (SST) observations are a critical data set for long-term climate reconstruction. However, its assimilation with an ensemble-based data assimilation method can degrade performance in the ocean interior due to spurious covariance. Assimilation in isopycnal coordinate can delay the degradation, but it remains problematic for long-term reanalysis. We introduce vertical localisation for SST assimilation in isopycnal coordinate. The tapering functions are formulated empirically from a large pre-industrial ensemble. We propose three schemes: 1) a step function with a small localisation radius that updates layers from the surface down to the first layer for which insignificant correlation with SST is found, 2) a step function with large localisation radius that updates layers down to the last layer for which significant correlation with SST is found, and 3) a Gaussian-like smooth tapering function. These tapering functions vary spatially and with calendar month and are applied to isopycnal temperature and salinity. The impact of vertical localisation on reanalysis performance is tested in identical twin experiments within the Norwegian Climate Prediction Model (NorCPM) with SST assimilation over the period 1980-2010. The SST assimilation without vertical localisation dramatically enhances performance in the whole water column but introduces a weak degradation at intermediate depths (e.g., 2000-4000 m). Vertical localisation greatly reduces the degradation and improves the overall accuracy of the reanalysis, in particular in the North Pacific and the North Atlantic. A weak degradation remains in some regions below 2000 m in the Southern Ocean. Among the three schemes, scheme 2) outperforms schemes 1) and 3) for temperature and salinity.

Yiguo Wang

Nansen Environmental and Remote Sensing Center

We will introduce the research project CoRea working to produce a coupled reanalysis of high importance to the climate research community. The new reanalysis will provide an estimate of the climate from 1850 to the present with uncertainty. The project CoRea is led by three early-career researchers, fostering the development of a new generation of climate researchers. In CoRea, we only use the ocean observations, and as a consequence the reanalysis produced by CoRea will be of great use in understanding the role of the ocean among the rest of the climate system (e.g., the atmosphere and sea ice). The produced reanalysis will be the first one to use the advanced data assimilation method known as the Ensemble Kalman Smoother, which combines dynamical propagation of background uncertainty with forecasting and hindcasting of observational data. A theoretical methodology will be revisited to achieve the updates on several characteristic time scales of the climate. The new method will maximise the influence of the rich contemporary observations backwards in time for several decades.

Yiguo Wang

Nansen Environmental and Remote Sensing Center

Ice sheets lost mass over the last three decades and contribute to global mean sea level rise. The ice-sheet mass loss is also accelerating. Including ice melt in ocean state estimation will help improve projection of future sea level rise. Here, we describe the latest ocean and sea-ice state estimate (Version 4, Release 5) of the "Estimating the Circulation and Climate of the Ocean" consortium (ECCO; https://ecco-group.org). The estimate includes an adjointable thermodynamic ice shelf model around Antarctic. Ice-shelf heat transfer coefficient and other model controls are adjusted so the model ice shelf melt rate is consistent with observation-based estimates. Including thermodynamic ice shelve around Antarctic is a starting point for applying it to Greenland, including dynamic ice shelves and more complicated physics, and adding new data constraints from ICESat-2 and other satellite missions.

Ou Wang

NASA, Jet Propulsion Laboratory