The NOAA Quantitative Observing System Assessment Program (QOSAP) coordinates the assessment of the impact of current and new observation technologies across the different NOAA Line Offices. QOSAP includes representatives from all NOAA Line Offices and, more recently, NASA/JPL. These studies help NOAA management prioritize mission designs in a cost-effective way by analyzing tradeoffs in the design of proposed observing systems, leading to a more effective observing system portfolio.

More specifically, QOSAP’s main focuses are (1) increase NOAA’s capacity to conduct quantitative observing system assessments, (2) develop and use appropriate quantitative assessment methodologies, and (3) inform major decisions on the design and implementation of optimal composite observing systems. Studies under the auspices of QOSAP identify technologies that could replace or improve current observing capabilities, fill existing gaps and enhance the balance in the existing observation portfolio. QOSAP uses observing system experiments (OSEs), observing system simulation experiments (OSSEs), and ensemble forecast sensitivity to observation impacts (EFSOI) as effective techniques to evaluate the impact of the different observation types.

In this talk, we will present an overview of QOSAP and a few of the emerging technologies being evaluated at NOAA under this program, which include atmospheric, oceanic, terrestrial and space-based observations. In particular, studies with CYGNSS data to evaluate the impact of derived sea surface winds to improve hurricane track and intensity analysis and forecasts at NOAA will be discussed. Emphasis will also be provided to other innovative and complementarity satellite technologies, including the Time-Resolved Observations of Precipitation structure and storm Intensity with a Constellation of Smallsats (TROPICS), the polar and equatorial components of the originally planned Constellation Observing Satellites for Meteorology, Ionosphere, and Climate-2 (COSMIC-2) and Geo-Hyperspectral sounders. In addition, improvements in hurricane prediction from the assimilation of manned and unmanned aircraft observations will presented. Finally, during this presentation a summary of these main areas or research and quantitative assessments aimed at improving operational global and hurricane prediction at NOAA, and associated challenges, will be discussed.

Global Navigation Satellite System-Reflectometry (GNSS-R) is an innovative and a rapidly developing approach to Earth Observation that makes use of signals of opportunity from Global Navigation Satellite Systems after they have been reflected off the Earth’s surface. The National Oceanography Centre (NOC) has worked extensively in recent years on inversion methods to retrieve ocean surface winds from spaceborne GNSS-R data. The Calibrated Bistatic Radar Equation (C-BRE) is NOC’s most advanced GNSS-R wind speed retrieval algorithm designed to account for the bistatic nature of the measurements and important calibrations and corrections for radiometry, geometry and GNSS transmitter power. The UK TechDemoSat-1 (TDS-1) is a technology demonstration satellite launched in July 2014 which has been providing calibrated GNSS-R data for more than three years. In this paper, we assess the ability of GNSS-R to accurately retrieve ocean wind speeds by determining the performance and stability of the NOC C-BRE wind speed products from TDS-1 against ECMWF ERA-5 re-analysis model output.

Results indicate that the best wind speed retrieval performance for TDS-1 is found in early phases of the mission, achieving an RMSE as low as ~ 2m/s relative to ERA-5. The performance is, however, not constant over time, and is characterised by marked deterioration throughout 2016, reaching its worse performance towards the end of 2016. The changes in performance can be linked to uncertainties in the knowledge of platform attitude, which appears to vary in time and has a strong first-order impact on wind speed retrieval through the knowledge of the antenna gain at the specular point.

Application of radiometric calibration and other corrections in the NOC C-BRE processor made it possible to assess secondary effects of other geophysical variables on GNSS-R signals. Effects of wind direction, sea state, sea surface temperature (SST) and sea surface salinity (SSS) are all examined using spaceborne TDS-1 data. These variables are expected to affect ocean wind speed retrieval through changes in the roughness and dielectric properties of the ocean surface. Our analyses reveal a small, but significant, effect due to SST and sea state that becomes more prevalent at lower wind speeds. A small effect of SSS can also be observed once other effects are isolated.

The NOC C-BRE wind speed retrieval algorithm demonstrates the potential of GNSS-R for accurate determination of wind speed and to understand other geophysical influences at low wind speed. GNSS-R offers promising opportunities for small satellites and constellations equipped with accurate platform attitude knowledge to contribute to improved ocean wind monitoring and operational weather forecasting by offering frequent high-quality wind speed observations at low relative cost.

Keywords: small satellites, constellations, ocean winds, GNSS

The SGR-ReSI instrument has been flying on UK TechDemoSat-1 since 2014, with exploitation of the instrument supported by ESA. It has demonstrated that GNSS signals can be used as a bistatic radar source for sensing the Earth's surface from space, and has also been used as the payload on the NASA CYGNSS small satellite hurricane sensing constellation. TDS-1 has been given a life extension, and in February 2018, the SGR-ReSI started collecting data continuously, seven days a week, which is four times higher than previously. To permit this increase, a form of on-board compression was implemented on-board the receiver to reduce the data rate by a factor of six. An automatic processing scheme was implemented that makes data directly available via FTP site. Ocean wind measurements from TDS-1 are being released from the web-site MERRByS with a short delay to demonstrate the feasibility of the use of GNSS reflectometry for an operational wind and wave sensing service. Recent results from an ESA-sponsored study called TGSCATT showed an agreement with ASCAT of better than 2 m/s for lower wind speeds, which TDS-1 and ASCAT were collocated.

Furthermore, data from TDS-1 has allowed the development of new potential GNSS-Reflectometry applications for ice, snow, soil moisture and flood sensing, amongst others. Preliminary work is being undertaken into a constellation of small satellites called ORORO with an instrument that can collect both reflectometry and radio-occultation measurements simultaneously to provide new measurements for numerical weather prediction assimilation purposes with unprecedented spatial and temporal coverage of the ocean surface. Recent new raw data collections from TDS-1 have included samples that are shown to contain Galileo signals, and an attempt has been made to collect reflections containing L1/L2C signals.

This presentation will show the latest results from TDS-1, and look to the potential for a future small satellite constellation dedicated to an operational service.

Spaceborne altimetry using signals of the Global Navigation Satellite Systems (GNSS) reflected off the Oceans (GNSS-Reflectometry or GNSS-R) has poor precision when conducted with small receivers using the publicly available GNSS codes. The reason is the narrow bandwidth of these codes, that limit the resolution in the domain of the time/range. On the other hand, using the wider bandwidth and encrypted codes that are transmitted for military applications requires more complex hardware solutions (e.g. the interferometric GNSS-R approach, iGNSS-R) which are not suitable for very small and cheap spaceborne platforms. A way to increase one to two orders of magnitude the altimetric precision, even when using small and cheap receivers (e.g. in nano-satellites), is by tracking the phase of the carrier signals, but this is only possible when coherent scattering occurs. Unfortunately, most reflections off the ocean surface follow a diffuse scattering mechanism hindering the tracking of the carrier signals. Exceptions are observations over very calm ocean surfaces or under very slant geometries. The latter condition holds because the actual roughness observed at very high incidence angles is effectively reduced. We call the latter type of observations 'grazing angle phase-delay altimetry'. Given that GNSS works at L-band, around 20-cm wavelength, tracking the carrier phase means that the altimetric delay can be measured at a few cm level in several-millisecond sampling intervals.

The concept was proved from airborne platforms at 3km altitude over Mediterranean open rough waters, and coherent signals were acquired for elevation angles of up to 30⁰ (incidence angles larger than 60⁰). In this new study we present the first spaceborne samples of grazing angle GNSS-R data analysed in terms of phase-delay altimetry. Some of the data streams acquired by the CyGNSS mission in raw data mode at grazing angles of observation present coherence, and their carrier phases can be tracked. Other segments of data do not present coherence. See examples of CyGNSS GNSS-R carrier phases along two segments over ocean waters, one for which the phase can be tracked and another one for which it cannot be tracked (Figure 1).

The study will present the existing CyGNSS grazing angle raw data cases and their derived altimetric performance. It will also investigate the reasons for different coherence behaviours (antenna gain pattern, effective roughness, system dynamics...), the expected frequency of occurrence in the ocean and the impact on designing very cost-efficient constellations of grazing angle GNSS-R altimeters. This innovative approach to ocean altimetry would complement the current and planned monostatic radar altimeters by providing very dense and cheap data to fill gaps in radar altimeter spatio-temporal coverage. A recent OSSE study indicated possible improvements for typhoon monitoring when these type of observations were included in ocean circulation models. The work presented here is also relevant to ESA's PRETTY mission, currently under Phase-B. This is an iGNSS-R altimetric mission that will work under similar geometries and using the same measurement concept.

Figure 1: Examples of CyGNSS carrier phase as obtained from raw acquisition at grazing angles of observation, at 50 ms sampling. In the upper case the phase can be tracked, and this results in delay-precisions of a few centimeter at several millisecond sampling. In the bottom case the signals are not sufficiently coherent and the carrier phase delay cannot be obtained. Both cases are segments of a CyGNSS track along the Java Sea acquired in raw data mode and satellite titled ~20⁰ on May 15, 2018.

Thanks to the successful operation of the spaceborne GNSS-R instrument, such as TechDemoSat-1 and CYGNSS, many studies are being conducted to expand the GNSS-R technique for the Earth observation. There are many reports for the ocean applications, e.g., measuring the sea surface roughness, wind speed, and sea surface height. In addition to them, land applications are also explored to measure land geophysical parameters. For the ground and airborne experiment, it has been reported that the GNSS-R technique is capable of providing the information of soil moisture, vegetation thickness, and biomass. However, the feasibility from spaceborne measurements are still questionable, i.e., is the sensitivity of GNSS-R observation to the soil moisture enough to produce informative results?

The European Space Agency is funding a project in order to study the feasibility of spaceborne GNSS-R to land applications. The main objective is to study that 1) the spaceborne GNSS-R is still applicable to retrieval soil moisture and vegetation; 2) which parameters should be included in the model and how they impact; 3) the retrieval algorithm and the products using TDS-1 data. The first step for the project is showing the sensitivity of GNSS-R measurements to the variation of soil moisture and vegetation. The results of this study have shown that the variation of soil moisture and vegetation influence the GNSS reflected signal, and it can be seen with proper descriptors of the DDM, e.g., SNR of DDM, the peak power of DDM, and the amount of cross-polarization.
Although it has been shown that the descriptors of DDM can be used for geophysical parameter retrieval, the quality of sensitivity is not very good due to several effects in the spaceborne GNSS-R observation scenario and instrument. The elevation (incidence) angles, the antenna pattern, thermal noise level vary the DDM descriptors. The topography and surface roughness of the reflected area also affects the GNSS-R measurements. Therefore, a further study inside all these effects is necessary to demonstrate the feasibility of spaceborne GNSS-R land applications. In order to do that, an end-to-end (E2E) simulator for GNSS-R Land is really helpful. In this study, a generic E2E simulator is presented for the spaceborne GNSS-R land applications, titled “GARCA SIM 4 Land”.

The simulator has been developed and implemented efficiently by using the heritage of spaceborne GNSS-R ocean simulator. The CommSensLab in UPC-BarcelonaTech developed the E2E simulator, “GARCA/GEROS-SIM M2” as participating the PARIS-IoD and GEROS-ISS projects. The land simulator uses many parts of the predecessor, e.g., the geometry, the DDM generation, and the instrument modules. The main module to be replaced is the scene generation module which computes the land reflection coefficients corresponding to the observed scene geophysical parameters. The computed reflection coefficients are further processed to generate the reflection intensities which are observed from the scene by the GNSS-R instrument. One of the important phenomena in the GNSS signal reflection over the land is modeling the coherent and incoherent components. In the case of land reflection, the coherent reflection is quite significant, and the extended “Zavorotny-Voronivich” model is used in “GARCA SIM 4 Land” to model it.

The developed simulator gives the following benefits to the study of GNSS-R land applications from a satellite. First, it simulates the DDMs with respect to the geophysical parameters such as variation of soil moisture, and vegetation optical depth. Many influential parameters are also efficiently analyzed for their impacts on the DDM, e.g., land cover, topography, incidence angle, etc. Furthermore, the GNSS signal reflection/scattering model is tested with the simulator. Finally, the simulator can be used to devise the tailored retrieval algorithm with error budget. At the conference, the structure and process of simulation will be explained and the simulation results for different scenarios will be presented.

First, a systematic evaluation of GNSS Reflectometry (GNSS-R) wind speeds is presented. After developing a wind speed retrieval algorithm, derived winds from measurements of UK TechDemoSat-1 (TDS-1), from May 2015 to July 2017, are evaluated answering the question: “How accurate are the GNSS-R winds?”. Wind speeds obtained with this relatively new technique are compared to the ones from the well-established Advanced Scatterometer, which can contribute to providing insights into the reliability of GNSS-R as an ocean wind scatterometer. European Reanalysis Interim (ERA-Interim) wind fields of European Centre for Medium-range Weather Forecasts (ECMWF) are taken as reference. The average global winds of TDS-1 and ECMWF are compared and discussed. The comparisons result in an RMSE and bias of 2.77 and -0.33 m/s by TDS-1, while the RMSE and bias derived by ASCAT winds are as large as 2.33 and 0.25 m/s, respectively. Using rain microwave-IR estimates of the Tropical Rainfall Measuring Mission (TRMM), the rain-affected observation of ASCAT and TDS-1 are collected for a performance evaluation during precipitation. Although ASCAT winds show a significant performance degradation resulting in RMSE and bias of 3.16 and 1.03 m/s during rainy conditions, TDS-1 winds are still reliable with an RMSE and bias of 2.94 and -0.21 m/s, respectively, which demonstrates the promising capability of GNSS-R technique for wind retrievals during rain events.
To improve the wind speed quality, a machine learning technique is implemented for wind speed inversion as a geophysical model function. More specifically, a feedforward neural network is trained for the TDS-1 wind speed inversion. The fitted network shows a significant improvement of 17% in the wind speed RMSE compared to the least-squares-based approach. It also demonstrates the ability to model a variety of effects degrading the retrieval accuracy such as the different transmitted level of the effective isotropic radiated power of GPS satellites. The derived Mean Absolute Error of the satellite with the SVN 34 is 30% decreased through our machine learning approach.
Furthermore, a rain signature in GNSS-R observations is demonstrated for the first time. This study presents evidence that the bistatic radar cross section derived from TechDemoSat-1 data is decreased due to altered roughness by the raindrops impinging on the ocean surface induced by winds lower than almost 6 m/s. The reduction is almost 0.7 dB at the wind speed of 3 m/s due to a precipitation of 0-2 mm/h. The simulations based on the recently published scattering theory provide a plausible explanation for this phenomenon which potentially enables the GNSS-R technique to detect precipitation over oceans at low winds.