Remedy 1: Adoption of an international approach to implement recommendations for addressing existing gaps in MWR operational products for climate monitoring utilization
TRL 5-7
The characterization of the total uncertainty budget for MWR retrievals requires quantification of contributions from the instrument hardware and the retrieval method. These contributions have been quantified in the open literature (e.g. Han and Westwater 2000; Hewison, 2006; Maschwitz et al., 2013; Stähli et al., 2013), but they often refer to one particular instrument and/or set of environmental conditions, and thus should not be generalized.
A proper uncertainty quantification for MWR retrievals shall result from the propagation of the uncertainty in calibration (transfer from raw voltages to the primary observable, the brightness temperature Tb) and the uncertainty in the retrieval method (transfer from Tb to atmospheric variables). As the uncertainty depends on the instrument and environmental conditions, the quantification shall be made dynamically, such that each measurement will be associated with one, generally different, uncertainty. The estimated uncertainty is thus time- and, for profiles, height-dependent. For a MWR network, the estimated uncertainty is also space-dependent, as it will depend on the instrument types deployed at various sites.
A systematic approach that dynamically evaluates the total uncertainty budget of MWR at the network level is lacking. In the following, the contributions to the total uncertainty are divided into four aspects: calibration and instrument characterization, retrieval method, radiative transfer and absorption model uncertainty, quality control.
Calibration and instrument characterization
Calibration and instrument characterization of MWR are to be performed regularly as they are time-dependent. Common procedures are applied by the operators to perform MWR calibration and instrument characterization. Currently, these procedures are usually provided by the manufacturers, and thus they are instrument-specific, or are based on user experience, and thus may be site-specific. Therefore, there is currently a lack of standardization in calibration procedures and uncertainty characterization. This in turn impacts negatively on the uniformity of products provided by a heterogeneous MWR network. This gap shall need to be addressed at both manufacturer and network levels.
Retrieval method
Different methods are currently applied for the retrieval of atmospheric variables from MWR observations. Different retrieval methods are adopted by different MWR manufacturers, operators, and users. Common retrieval methods include, but are not limited to, multivariate regression, neural networks and optimal estimation. This situation holds true for heterogeneous networks, such as the one currently establishing in Europe. The uncertainty of MWR retrievals depends partially on the used retrieval method. Documentation, versioning, and settings are usually not accessible nor maintained. Information on retrieval uncertainty is often completely missing. The traceability of software documentation and versioning is also not guaranteed. This lack of coordination impacts negatively on the harmonization and spatio-temporal consistency of products from a heterogeneous MWR network. This gap shall need to be addressed at the network level.
Radiative transfer and absorption model uncertainty
Most common MWR retrieval methods are based on radiative transfer simulations through the atmospheric medium. Thus, uncertainties in modelling the absorption/emission of microwave (MW) radiation by atmospheric gases and hydrometeors affect all the retrieval methods based on simulated MW radiances. Only retrieval methods based on historical datasets of MWR observations and simultaneous atmospheric soundings are not affected by absorption model uncertainties. Currently, the information on MW absorption model uncertainties are dispersed and not easily accessible. Most operational MWR operate in the 20-60 GHz range, where relevant absorption comes from water vapour, oxygen, and liquid water. A variety of models are available which combine the absorption of water vapour, oxygen, and liquid water, as well as other minor contributions. Absorption model uncertainties are currently estimated from the output difference of different models, while a more rigorous estimate is lacking. An attempt to mitigate this gap is currently being carried out within GAIA-CLIM.
Quality control
Quality control (QC) procedures are fundamental for providing users with tools for judging and eventually screening MWR data and products. Most operational MWRs apply QC procedures that are developed by either the MWR manufacturer or by the operators based on their experience. There are different levels of QC procedures, going from sanity checks of the system electronics, to monitoring the presence of rain/dew on the instrument window, to radio frequency interference detection, to monitoring calibration against independent reference measurements (usually by radiosondes). The nature of the QC procedures varies, as these may be applicable to all instruments or conversely be instrument and/or site specific. Therefore, there is currently a lack of harmonization and automation of MWR QC procedures. This impacts on the quantity and quality of the data delivered, as poor QC may result in either delivery of faulty data, or screening out of good data. This gap shall need to be addressed at both manufacturer and network levels.
Temperature and humidity sounders in general
Attempts to mitigate this gap are currently being carried out within and outside of GAIA-CLIM. Within GAIA-CLIM, a review of state-of-the-art MW absorption models and associated uncertainty has started (Cimini et al., 2017a). The absorption model uncertainties need to be propagated through radiative transfer and inverse operator to estimate the total uncertainties affecting the simulated brightness temperatures and the retrieval methods. A review paper shall collect the outcome of this analysis.
Outside of GAIA-CLIM, attempts to mitigate this gap are currently being carried out in the framework of the EU COST Action TOPROF, specifically by the Microwave Radiometer Working Group (WG3). WG3 is actively tackling the above challenges by interacting with manufacturers and users. WG3 produced a report on calibration best practices. New developments on calibration target design have been stimulated through the interactions with manufacturers. Network-suitable retrieval methods are currently under development within TOPROF WG3 (De Angelis et al. 2016; 2017). The role of GAIA-CLIM is to follow the developments at TOPROF and report to GAIA-CLIM as well as MWR users/manufacturers.
The present overarching MWR gap will be considered closed when procedures for MWR calibration and instrument characterization and a unified retrieval method will be performed uniformly across the network.
Argument: The remedy of G2.13, i.e. the development of MW standards maintained at national/international measurement institutes and the availability of transfer standards, will set the basis for SI-traceability of MWR observations and retrievals. However, tools for evaluating the MWR total uncertainty budget can be developed independently of the solution of G2.13.
- Cimini D., P. Rosenkranz, M. Tratyakov, and F. Romano, Sensitivity of microwave downwelling brightness temperatures to spectroscopic parameter uncertainty, ITSC21, Darmstadt, Nov. 2017a.
- Cimini D., P. Martinet, F. De Angelis, Network 1DVAR temperature and humidity profile retrievals from ground-based microwave radiometers in Europe, European Meteorological Society Annual Meeting, Dublin, 4-8 September, 2017b.
- De Angelis, F., Cimini, D., Löhnert, U., Caumont, O., Haefele, A., Pospichal, B., Martinet, P., Navas-Guzmán, F., Klein-Baltink, H., Dupont, J.-C., and Hocking, J.: Long-term observations minus background monitoring of ground-based brightness temperatures from a microwave radiometer network, Atmos. Meas. Tech., 10, 3947-3961, https://doi.org/10.5194/amt-10-3947-2017, 2017.
- De Angelis, F., Cimini, D., Hocking, J., Martinet, P., and Kneifel, S.: RTTOV-gb – adapting the fast radiative transfer model RTTOV for the assimilation of ground-based microwave radiometer observations, Geosci. Model Dev., 9, 2721-2739, https://doi.org/10.5194/gmd-9-2721-2016, 2016.
- Han Y. and E. R. Westwater: Analysis and Improvement of Tipping Calibration for Ground-based Microwave Radiometers. IEEE Trans. Geosci. Remote Sens., 38(3), 1260–127, 2000.
- Hewison T., Profiling Temperature and Humidity by Ground-based Microwave Radiometers, PhD Thesis, Department of Meteorology, University of Reading, 2006.
- Maschwitz G., U. Löhnert, S. Crewell, T. Rose, and D.D. Turner, 2013: Investigation of Ground-Based Microwave Radiometer Calibration Techniques at 530 hPa, Atmos. Meas. Tech., 6, 2641–2658, doi:10.5194/amt-6-2641-2013
- Pospichal B., N. Küchler, U. Löhnert, J. Güldner, Recommendations for operation and calibration of Microwave Radiometers (MWR) within a network, Online: http://tinyurl.com/TOPROF-MWR-recommend-2016 , 2016.
- Stähli, O., Murk, A., Kämpfer, N., Mätzler, C., and Eriksson, P.: Microwave radiometer to retrieve temperature profiles from the surface to the stratopause, Atmos. Meas. Tech., 6, 2477-2494, doi:10.5194/amt-6-2477-2013, 2013.
Ground-based microwave radiometers (MWR) provide continuous and unattended retrievals of atmospheric temperature and humidity profiles, as well as of vertically-integrated total column water vapour (TCWV) and cloud liquid water (TCLW). Despite the significant scientific advancements allowed by MWR observations over the last forty years, current operational MWR retrievals are still lacking a traceable uncertainty estimate. The characterization of the total uncertainty budget for MWR retrievals requires quantification of the contributions from the instrument hardware (including absolute calibration) and the retrieval method (including the radiative transfer model). These contributions have been quantified in open literature, but they often refer to one particular instrument and/or environmental condition, and thus are not able to be generalized. A systematic approach that dynamically evaluates the total uncertainty budget of MWR (i.e. as function of instrument/environment conditions) at the network level is lacking. Initiatives for mitigating this gap are being undertaken in Europe as well as in the United States.