ESR3 - TRUTHS

for Climate Workshop

27-28 June 2024 | ESA-ECSAT



SESSION: ASSIMILATION AND RE-ANALYSIS 

IN NWP & AC/AQ MODELS

The discussions on the expected benefits from TRUTHS and the way forward to prepare these user communities were preceded by four introductory talks from: 

Salient points from the presentations include:

  • NWP centres plan to assimilate “optical” channels in the next 5-10 years, to further constrain e.g. the cloud modelling. 2 to 5 broadband channels are considered, e.g. from traditional meteorological imagery (AVHRR, FCI, METimage, MODIS, VIIRS…). This is a new scientific development line, with important potential benefits for NWP, bringing in high resolution information on aerosol distribution and cloud position and microphysics, in particular boundary layer and low-level cloud, almost invisible to thermal instruments, and clouds under cirrus as the latter are semi-transparent in visible wavelengths, which is not available in TIR and MW data. 
  • This comes with a number of scientific challenges, including understanding the associated CALC and OBS uncertainties, on a 5 to 10 year horizon. 
  • Different total solar irradiance (TSI) sources are used by different models around the world, whose TSI currently have a spread of 5 W/m2, which is about the climate sensitivity. It was clarified that the TSI providers recently explained the departures and converged on a unified value based on SI-traceable measurements. See M. Haberreiter talk in the Climate session. 
  • TRUTHS absolute calibration, spectral and spatial resolution are expected to benefit in ECV directly derived from satellite observations, in seasonal forecasting, NWP and re-analysis 
  • NWP operational paradigm is to assimilate radiances (not L2 products). An effective RTM is required to enable operational OBS-CALC computations, to learn about and improve models errors 
  • Utilisation in re-analysis requires experience in NWP context. Non-NRT data feed much complicates this process, latency requirements varying among centres and global/regional models. 
  • Climate change signal is one of the key uncertainties in operational seasonal forecasting 
  • Earth radiation data are not yet assimilated in ECMWF system 
  • The Earth Imbalance is asymmetrical, the heat intake is higher in the Southern hemisphere. CMIP6 reveals a stronger N/S difference than previously modelled, related to air quality improvements and aerosol effect, whose modelling needs to be further improved. 
  • SST is one of the geophysical parameters where biases between climate models are important and must be understood. ERA-5 albedo is usually darker than GERB because of WV modelling. Error on optical depths have stronger effect on thin clouds, with consequently large uncertainties on the role of cirrus. 
  • CERES-ERA-5 agree well until 2015, then tend to diverge, presumably because ERA-5 is not accounting for Chinese clean air policies and anthropogenic aerosol reduction. 
  • TRUTHS’ radiometric, spectral and spatial performances are expected to help better understand and improve model’s errors, in particular the role of (thin) clouds in the Earth radiation budget and the interaction cloud-aerosol. 
  • TRUTHS will help unequivocally resolve discrepancies between models and between models and observations. 
  • TRUTHS (SW), in combination with other (LW) missions, will constrain global energy and water-vapour cycles 
  • ERA-6 spatial resolution will increase to about 14km, commensurate with e.g. GERB observations (unlike ERA-5 whose resolution was too coarse) 
Discussions outcome: 
  • TRUTHS will fill a critical gap in term of radiometrically stable reference record. 
  • Ideally TRUTHS is the precursor of a longer series, but the limited mission lifetime of up to 8 years is deemed sufficient to characterise potential mission aging and resolve ambiguities in trends (whether climate or sensor degradation). Experience made with Aeolus showed that as long as short-timer missions do not shift the mean model state, it is benefitial to assimilate as much good quality data as possible, whether in NWP or re-analysis. Longer time series will also be achievable thanks to similar international missions, e.g., CLAREO and LIBRA. 
  • The spatio-temporal sampling with TRUHS, e.g. upon RAAN, will condition the absolute value of the (annual) average ERB quantities. Hence TRUTHS vs e.g. CERES biases will depend on the orbit injection. However, the trends with TRUTHS will be extremely accurate and used to verify/calibrate other datasets.
  • Timeliness recommendations on input data in NWP models are of 3h/30min for the Global/Regional models at UKMO and most Met Services, and of 4 to 6h at ECMWF. The data assimilation windows are ±3h and ±30min around the analysis time in the Met Office global and regional models, respectively. This means that some data could be used in the global model if arriving in <6h, but they would be concentrated at the start of the assimilation window and bring very little benefits (see, e.g, https://doi.org/10.1002/qj.3596). 
  • In their current operational implementations, varBC schemes cannot proceed in a delayed mode with late incoming data, but coefficients could be spinup off-line over short period of time, typically 3-month assimilation experiments. With a current latency of several days, TRUTHS data cannot be used for passive monitoring and routinely anchoring the optical missions whose observations will be assimilated as this would require either continuous assimilation (of NRT data) or frequent off-line update of the varbc coefficients obtained in delayed mode (e.g., running a delayed experiment every 3/4 months) which is beyond the available resources. 
  • Direct broadcast (e.g. via DBnet) has proven a successful mechanism to downlink regional subsets to NWP centres, where otherwise missions’ data from central production were coming in with too strong delays, e.g. with Chinese missions at UKMO. The entire spectrum would not be necessary for this, a handful of channels would be sufficient to serve NWP purposes and later re-analysis. ESA will investigate the possibilities for direct broadcast, noting that the Space and Ground segments were not designed for this purpose. ESA and UKMO will meet to discuss the idea and requirements further. 
  • Alternatively, experience of TRUTHS in NWP could be made in off-line dedicated assimilation experiments. Such studies would already provide feedback to understand model and observation biases and improve modelling capabilities in view of re-analysis campaigns. The NWP-SAF will hold their first preparatory workshop of the CDOP-5 (Continuous Development and Operations Phase, 2027-2032) in February. Potential interests from the User communities could be conveyed there. F. Carminati is a PoC. This includes, e.g., RTTOV coefficients, need for data pre-processor (spatial and/or spectral thinning, BUFR encoding, ...). An ISSI workshop will be help at ESRIN in March to discuss the future requirement for the assimilation of VIS radiances in NWP systems. 
  • The optical channel selection for assimilation in NWP is not yet settled, but it will be broadband channels. In that perspective, the TRUTHS spectral resolution is an asset to convolve the information with the other missions’ ISRFs. Hyperspectrally resolved measurements are unlikely to be assimilated mid-term. DWD is the first centre assimilating a visible channel. UKMO has plans of assimilating 2-3 channels in the next 5 years, ECMWF is planning 2 and then up to half-a-dozen in the next 5 and 10 years, having already started passive monitoring of 2 channels. It is unlikely that scientific maturity will be rapidly reached such that the channels selection is redefined because of model’s sensitivities. The process will be partly based on tries and learning. 
  • TRUTHS monthly/annual average radiance products could be used as benchmark datasets for verification of future ERA generations, the question of spatial sampling is an important aspect that needs further preparatory studies. 
  • AC/AQ modellers were not represented in this expert meeting. The TRUTHS project should reach out to them as they may have specific requirements and TRUTHS may bring additional specific benefits. In particular, it was noted that e.g. in CAMS the assimilation of L2 products from EO missions is the current operational baseline (instead of radiances in NWP) and that TRUTHS has relevant spectral information for a number of species.