New sampling strategy mitigates a solar-geometry-induced bias in sub-kilometre vapour scaling statistics derived from imaging spectroscopy

Upcoming spaceborne imaging spectrometers will retrieve clear-sky total column water vapour (TCWV) over land at a horizontal resolution of 30–80 m. Here we show how to obtain, from these retrievals, exponents describing the power-law scaling of sub-kilometre horizontal variability in clear-sky bulk planetary boundary layer (PBL) water vapour (q) accounting for realistic non-vertical sunlight paths. We trace direct solar beam paths through large eddy simulations (LES) of shallow convective PBLs and show that retrieved 2-D water vapour fields are “smeared” in the direction of the solar azimuth. This changes the horizontal spatial scaling of the field primarily in that direction, and we address this by calculating exponents perpendicular to the solar azimuth, that is to say flying “across” the sunlight path rather than “towards” or “away” from the Sun. Across 23 LES snapshots, at solar zenith angle SZA = 60∘ the mean bias in calculated exponent is 38 ± 12 % (95 % range) along the solar azimuth, while following our strategy it is 3 ± 9 % and no longer significant. Both bias and root-mean-square error decrease with lower SZA. We include retrieval errors from several sources, including (1) the Earth Surface Mineral Dust Source Investigation (EMIT) instrument noise model, (2) requisite assumptions about the atmospheric thermodynamic profile, and (3) spatially nonuniform aerosol distributions. By only considering the direct beam, we neglect 3-D radiative effects such as light scattered into the field of view by nearby clouds. However, our proposed technique is necessary to counteract the direct-path effect of solar geometries and obtain unique information about sub-kilometre PBL q scaling from upcoming spaceborne spectrometer missions.

New sampling strategy removes imaging spectroscopy solar-smearing bias in sub-km vapour scaling statistics

Upcoming spaceborne imaging spectrometers will allow retrieval of total column water vapour (TCWV) over land at horizontal resolution of 30–80 m. Here we show how to obtain, from these retrievals, exponents describing the power-law scaling of sub-km horizontal variability in clear-sky bulk planetary boundary layer (PBL) water vapour (q). Using large-eddy simulations (LES) of shallow convective PBLs we show how sunlight entering the PBL up to several km away from the footprint location degrades estimates of these exponents. We address this by calculating exponents perpendicular to the solar azimuth, that is to say flying “across” the sunlight path rather than “towards” or “away” from the Sun. Across 23 LES snapshots, at SZA = 60° the mean bias in calculated exponent is 38 ± 12 % (95 % range) along the solar azimuth, while following our strategy it is 3 ± 9 % and no longer significant. Both bias and root-mean-square error RMSE decrease with lower SZA. We include retrieval errors from several sources including: (1) the Earth Surface Mineral Dust Source Investigation (EMIT) instrument noise model, (2) requisite assumptions about the atmospheric thermodynamic profile, and (3) spatially nonuniform aerosol distributions. This technique can be used to obtain unique information about sub-km PBL q scaling from upcoming spaceborne spectrometer missions, while mitigating errors due to challenging solar geometries.

Boosting Automated Sleep Staging Performance in Big Datasets using Population Sub-grouping

Current approaches to automated sleep staging from the electroencephalogram (EEG) rely on constructing a large labeled training and test corpora by aggregating data from different individuals. However, many of the subjects in the training set may exhibit changes in the EEG that are very different from the subjects in the test set. Training an algorithm on such data without accounting for this diversity can cause underperformance. Moreover, test data may have unexpected sensor misplacement or different instrument noise and spectral responses. This work proposes a novel method to learn relevant individuals based on their similarities effectively. The proposed method embeds all training patients into a shared and robust feature space. Individuals that share strong statistical relationships and are similar based on their EEG signals are clustered in this feature space before being passed to a deep learning framework for classification. Using 994 patient EEGs from the 2018 Physionet Challenge (≈ 6,561 hours of recording), we demonstrate that the clustering approach significantly boosts performance compared to state-of-the-art deep learning approaches. The proposed method improves, on average, a precision score from 0.72 to 0.81, a sensitivity score from 0.74 to 0.82, and a Cohen’s Kappa coefficient from 0.64 to 0.75 under 10-fold cross-validation.

The re-analysis on the raw data processing of KBR and LRI on GRACE-FO

<div> <p>The KBR (K-Band ranging instrument) and LRI (Laser Interferometer) are used to measure the distance variations between the twin spacecraft, which is one of the most important observations used for temporal gravity field recovery. The data pre-processing from raw or so-called Level-1A into the Level-1B format, which is suited for gravity field recovery, is a key step. Although Level-1B files are made publicly available by the GRACE-FO Science Data System (SDS), it has been shown that alternative Level-1B datasets may yield improved the results of gravity field<sup>[1]</sup>. Investigations of the pre-processing may allow us to improve the gravity recovery strategy and are essential to support developments of gravimetric satellite missions in China, such as TianQin-2 project. The pre-processing normally includes the time-tag synchronization, filtering and resampling, and other corrections, e.g. light-time correction for both instruments and antenna offset correction for KBR. We re-processed the Level-1A data of KBR and LRI to the Level1B using code developed at IGG/Wuhan. The results show good agreement in case of the RL04 KBR data, i.e. the differences between IGG-KBR1B and SDS-KBR1B are about three orders of magnitude lower than the instrument noise level for KBR. For the LRI, we found that phase jumps are not removed completely in the SDS-LRI1B products. As shown by Abich<sup>[2]</sup>, these phase jumps in the LRI phase observations are mainly coincident with thruster activations. Our work will analyze the impacts of different processing methods of the raw data on post-fit residuals and the gravity field recovery based on IGG-KBR1B and IGG-LRI1B datasets.</p> <p> </p> <div> <p>[1] Wiese, D.: SDS Level-2/-3 JPL, GRACE/GRACE-FO Science Team Meeting 2020, online, 27 October–29 Oct 2020, GSTM2020-75, https://doi.org/10.5194/gstm2020-75, 2020.</p> <p>[2]    Abich K, Abramovici A, Amparan B, et al. In-Orbit Performance of the GRACE Follow-on Laser Ranging Interferometer [J]. Phys Rev Lett, 2019, 123(3): 031101, https://doi.org/10.1103/PhysRevLett.123.031101.</p> </div> </div><p> </p>

Adaptive Variational Mode Decomposition Method for Eliminating Instrument Noise in Turbulence Detection

Noise removal is a key issue in the retrieval of turbulence from meteorological radiosonde data using the method proposed by Thorpe. Only by reducing as much as possible the influence of noise in the potential temperature fluctuations can the retrieval results reflect the turbulence characteristics of the real atmosphere. In this paper, an adaptive variational mode decomposition (VMD) method is proposed that is used to remove noise fluctuations from the potential temperature profile, and particle swarm optimization and mutual information are used to optimize the preset VMD parameters. The Thorpe method is applied to the denoised potential temperature profile to identify and characterize turbulent regions. The results show that the adaptive VMD method is very effective for denoising the potential temperature profile in both simulation experiments and actual detection data. The real turbulence overturn can be selected from the inversions by optimal smoothing and statistical tests. This method is an improvement on the Wilson method and allows the Thorpe method to be applied to daytime sounding data, avoiding the confusion between noise and turbulence that results in the distortion of the turbulence scale.

HIR4: cosmological signatures imprinted on the cross-correlation between a 21-cm map and galaxy clustering

ABSTRACT We explore the cosmological multitracer synergies between an emission-line galaxy distribution from the Dark Energy Spectroscopic Instrument and a Tianlai Project 21-cm intensity map. We use simulated maps generated from a particle simulation in the light-cone volume (Horizon Run 4), sky-trimmed and including the effects of foreground contamination, its removal and instrument noise. We first validate how the foreground residual affects the recovered 21-cm signal by putting different levels of foreground contamination into the 21-cm maps. We find that the contamination cannot be ignored in the angular autocorrelation power spectra of H i even when it is small, but it has no influence on the accuracy of the angular cross-correlation power spectra between H i and galaxies. In the foreground-cleaned map case, as information is lost in the cleaning procedure, there is also a bias in the cross-correlation power spectrum. However, we found that the bias from the cross-correlation power spectrum is scale-independent, which is easily parametrized as part of the model, while the offset in the H i autocorrelation power spectrum is non-linear. In particular, we tested that the cross-correlation power also benefits from the cancellation of the bias in the power spectrum measurement that is induced by the instrument noise, which changes the shape of the autocorrelation power spectra but leaves the cross-correlation power spectra unaffected. We then modelled the angular cross-correlation power spectra to fit the baryon acoustic oscillation feature in the broad-band shape of the angular cross-correlation power spectrum, including contamination from the residual foreground and the effect of instrument noise. We forecast a constraint on the angular diameter distance DA for the Tianlai Pathfinder redshift 0.775 < z < 1.03, giving a distance measurement with a precision of 2.7 per cent at that redshift.

Eigenspectra: a framework for identifying spectra from 3D eclipse mapping

ABSTRACT Planetary atmospheres are inherently 3D objects that can have strong gradients in latitude, longitude, and altitude. Secondary eclipse mapping is a powerful way to map the 3D distribution of the atmosphere, but the data can have large correlations and errors in the presence of photon and instrument noise. We develop a technique to mitigate the large uncertainties of eclipse maps by identifying a small number of dominant spectra to make them more tractable for individual analysis via atmospheric retrieval. We use the eigencurves method to infer a multiwavelength map of a planet from spectroscopic secondary eclipse light curves. We then apply a clustering algorithm to the planet map to identify several regions with similar emergent spectra. We combine the similar spectra together to construct an ‘eigenspectrum’ for each distinct region on the planetary map. We demonstrate how this approach could be used to isolate hot from cold regions and/or regions with different chemical compositions in observations of hot Jupiters with the James Webb Space Telescope (JWST). We find that our method struggles to identify sharp edges in maps with sudden discontinuities, but generally can be used as a first step before a more physically motivated modelling approach to determine the primary features observed on the planet.