ALBERT

Description: Recent years have seen the increasing inclusion of per-retrieval prognostic (predictive) uncertainty estimates within satellite aerosol optical depth (AOD) data sets, providing users with quantitative tools to assist in optimal use of these data. Prognostic estimates contrast with diagnostic (i.e. relative to some external truth) ones, which are typically obtained using sensitivity and/or validation analyses. Up to now, however, the quality of these uncertainty estimates has not been routinely assessed. This study presents a review of existing prognostic and diagnostic approaches for quantifying uncertainty in satellite AOD retrievals, and presents a general framework to evaluate them, based on the expected statistical properties of ensembles of estimated uncertainties and actual retrieval errors. It is hoped that this framework will be adopted as a complement to existing AOD validation exercises; it is not restricted to AOD and can in principle be applied to other quantities for which a reference validation data set is available. This framework is then applied to assess the uncertainties provided by several satellite data sets (seven over land, five over water), which draw on methods from the empirical to sensitivity analyses to formal error propagation, at 12 Aerosol Robotic Network (AERONET) sites. The AERONET sites are divided into those where it is expected that the techniques will perform well, and those for which some complexity about the site may provide a more severe test. Overall all techniques show some skill in that larger estimated uncertainties are generally associated with larger observed errors, although they are sometimes poorly calibrated (i.e. too small/large in magnitude). No technique uniformly performs best. For powerful formal uncertainty propagation approaches such as Optimal Estimation the results illustrate some of the difficulties in appropriate population of the covariance matrices required by the technique. When the data sets are confronted by a situation strongly counter to the retrieval forward model (e.g. potential mixed land/water surfaces, or aerosol optical properties outside of the family of assumptions), some algorithms fail to provide a retrieval, while others do but with a quantitatively unreliable uncertainty estimate. The discussion suggests paths forward for refinement of these techniques.

Description: We developed a method for examining ice formation on solid materials under cloud-like conditions. Our experimental approach couples video-rate optical microscopy of ice formation with high-resolution atomic force microscopy (AFM) of the initial mineral surface. We demonstrate how colocating stitched AFM images with video microscopy can be used to relate the likelihood of ice formation to nanoscale properties of a mineral substrate, e.g., the abundance of surface steps of a certain height. We also discuss the potential of this setup for future iterative investigations of the properties of ice nucleation sites on materials.

Description: This study presents and evaluates several candidate approaches for downscaling observations from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) in order to increase the horizontal resolution of subsequent cloud optical thickness (τ) and effective droplet radius (reff) retrievals from the native 3 × 3 km2 spatial resolution of the narrowband channels to 1 × 1 km2. These methods make use of SEVIRI’s coincident broadband high–resolution visible (HRV) channel. For four example cloud fields, the reliability of each downscaling algorithm is evaluated by means of collocated 1 × 1 km2 MODIS radiances, which are re-projected to the horizontal grid of the HRV channel, and serve as reference for the evaluation. By using these radiances smoothed with the spatial response function of the native SEVIRI channels as retrieval input, the accuracy at the SEVIRI standard resolution can be evaluated and an objective comparison of the accuracy of the different downscaling algorithms can be made. For the example scenes considered in this study, it is shown that neglecting high-frequency variations below the SEVIRI standard resolution results in significant random absolute deviations of the retrieved τ and reff of up to ≈ 14 and ≈ 6 μm, respectively, as well as biases. By error propagation, this also negatively impacts the reliability of the subsequent calculation of liquid water path (WL) and cloud droplet number concentration (ND), which exhibit deviations of up to ≈ 89 g m−2 and ≈ 177 cm−3, respectively. For τ, these deviations can be almost completely mitigated by the use of the HRV channel as a physical constraint, and by applying most of the presented downscaling schemes. For the accuracy of reff,the choice of downscaling scheme however is important: deviations are generally of similar magnitude or larger than those for retrievals at the SEVIRI standard resolution, indicative of their limited skill at predicting high–frequency spatial variability in reff. A strong degradation of accuracy of reff is observed for some of the approaches, which also affects subsequent WL and ND estimates. As a result, an approach which constrains the reff to the lower–resolution results is recommended. Overall, this study demonstrates that an increase in horizontal resolution of SEVIRI cloud property retrievals can be reliably achieved by use of its HRV channel, yielding cloud properties which are preferable in terms of accuracy to those obtained from SEVIRI’s standard-resolution. This work advances efforts to mitigate impacts of scale mismatches among channels of multi–resolution instruments on cloud retrievals.

Description: A new method using the Micro Rain Radar (MRR) to determine the melting layer height is presented. The MRR is a small vertically pointing frequency modulated continuous wave radar which measures Doppler spectra of precipitation. From these Doppler spectra, various variables such as Doppler velocity or spectral width can be derived. The melting layer is visible through a higher reflectivity and an acceleration of the falling particles, among others. These characteristics are fed to a neural network to determine the melting layer height. For the training of the neural network, the melting layer height is determined manually. The neural network is trained and tested using data from two sites covering all seasons. For most cases, it is well able to detect the correct melting layer height. Sensitivity studies show that the neural network is able to handle different settings of the MRR. Comparisons to radiosonde data and cloud radar data show a good agreement in melting layer heights.

Description: This study assesses the performance of SKYNET in comparison to AERONET (Aerosol Robotic Network) for retrieving aerosol optical properties (AOPs) in Beijing, China. The results obtained from simultaneous measurements show high correlation coefficients (〉 0.994) for aerosol optical depth (AOD) at each wavelength. The highest correlation coefficient for Ångström exponent is 0.825, at 500–870 nm. The single scattering albedo (SSA) of SKYNET is systematically larger than that of AERONET at each wavelength, and adjusting the SVA (solid view angle) and SA (surface albedo) input values can easily affect the value of SKYNET SSA. The volume size distribution patterns derived from the two networks’ instruments are both bimodal, which is typical, while the coarse-mode volume of SKYNET is larger than that of AERONET on average. According to the frequency distribution of aerosol particles, coarser aerosol particles often present in autumn and finer particles usually exist in winter, and there are more absorbent aerosol particles in winter. SKYNET data, combined with meteorological data, CALIPSO (Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations) data, backward trajectories, and WPSCF (weighted potential source contribution function) and WCWT (weighted concentrated weighted trajectory) analyses are used to analyze a serious pollution event in winter over Beijing. The results suggest that it was not only affected by local emissions but also by regional transport. The AOPs under three weather conditions (clean, dusty, haze) in Beijing are discussed. The values of AOD on haze days are about 10.3, 10.0, 8.7, 6.3 and 6.2 times larger than those on clean days at 400, 500, 670, 870 and 1020 nm, respectively; and under haze conditions, the PM2.5 (fine particulate matter) is about 7.6 times larger than that under clean conditions. The values of AOD on dusty days are about 7.1, 7.4, 7.0, 5.3 and 5.2 times larger than those on clean days at 400, 500, 670, 870 and 1020 nm, respectively; and under haze conditions, the PM2.5 is about 5.2 times larger than that under clean conditions.

Description: The annual variation of δD in the tropical lower stratosphere is a critical indicator for the relative importance of different processes contributing to the transport of water vapour through the cold tropical tropopause region into the stratosphere. Distinct observational discrepancies of the δD annual variation were visible in the works of Steinwagner et al. (2010) and Randel et al. (2012), focusing on MIPAS (Michelson Interferometer for Passive Atmospheric Sounding) and ACE-FTS (Atmospheric Chemistry Experiment-Fourier Transform Spectrometer) data, respectively. Here we reassess the discrepancies based on newer MIPAS and ACE-FTS data sets, showing for completeness also results from SMR (Sub-Millimetre Radiometer) observations and a ECHAM/MESSy (European Centre for Medium-Range Weather Forecasts Hamburg/Modular Earth Submodel System) Atmospheric Chemistry (EMAC) simulation (Eichinger et al., 2015b). Similar to the old analyses, the MIPAS data sets yield a pronounced annual variation (maximum about 75 ‰) while that derived from the ACE-FTS data sets is rather weak (maximum about 25 ‰). While all data sets exhibit the phase progression typical for the tape recorder the annual maximum in the ACE-FTS data set precedes that in the MIPAS data set by 2 to 3 months. We critically consider several possible reasons for the observed discrepancies, focusing primarily on the MIPAS data set. We show that the δD annual variation in the MIPAS data is up to an altitude of 40 hPa substantially impacted by a start altitude effect, i.e. dependency between the lowermost altitude where MIPAS retrievals are possible and retrieved data at higher altitudes. In addition, there is a mismatch in the vertical resolution of the MIPAS HDO and H2O data (being consistently better for HDO), which actually results in an artificial tape recorder-like signal in δD. Considering these MIPAS characteristics largely removes any discrepancies between the MIPAS and ACE-FTS data sets and confirms a δD tape recorder signal with an amplitude of about 25 ‰ in the lowermost stratosphere.

Description: In order to obtain better performance of meteorological applications, it is necessary to distinguish radar echoes from meteorological and non-meteorological targets. After the comprehensive analysis of the computational efficiency and radar system characteristics, a fuzzy logic method similar to the MetSignal algorithm is adopted, but its performance is improved significantly in weak signal regions where polarimetric variables are severely affected by noise. In addition, post-processing is adjusted to prevent anomalous propagation at far range to be misclassified as meteorological echo. Moreover, an additional fuzzy logic echo classifier is introduced into post-processing to suppress misclassification in the melting layer. An independent test set is selected to evaluate algorithm performance, and the statistical results show that the performance of the algorithm has been significantly improved, especially with respect to the classification of meteorological echoes in weak signal regions.

Description: Stratospheric sulfate aerosols play an important role for the physics and chemistry of the atmosphere. Key fundamental properties of the aerosols are their size and particle size distribution. Despite extensive research spanning several decades, the scientific understanding of these properties of stratospheric aerosols is incomplete. The particle sizes reported in different studies cover a wide range – even under background stratospheric conditions – and particle sizes retrieved from satellite solar occultation measurements in the optical spectral range show a tendency to be systematically larger than retrievals based on other optical methods. In this contribution we suggest a potential reason for these systematic differences. Differences between the actual aerosol particle size distribution and the size distribution assumed for aerosol size retrievals may lead to systematic differences in retrieved aerosol size. We demonstrate that these systematic differences may differ significantly for different measurement techniques, which is related to the different sensitivities of these measurement techniques to specific parts of the aerosol particle population. In particular, stratospheric aerosol size retrievals based on solar occultation observations may yield systematically larger particle size estimates compared to, e.g., lidar backscatter measurements. Aerosol concentration – on the other hand – may be systematically smaller in retrievals based on occultation measurements compared to lidar measurements. The results question the overall significance of stratospheric aerosol size retrievals based on optical satellite or lidar measurements, as long as the actual aerosol particle size distribution is not well known.

Description: This manuscript presents the calibration and validation studies for the Radio Occultations and Heavy Precipitation experiment aboard the PAZ satellite. These studies, necessary to assess and characterize the noise level and robustness of the ΔΦ observable of Polarimetric Radio Occultations (PRO), confirm the good performance of the experiment and the capability of this technique in sensing precipitation. It is shown how all the predicted effects that could have an impact into the PRO observables (e.g. effect of metallic structures nearby the antenna, the Faraday Rotation at the ionosphere, signal impurities in the transmission, altered cross polarization isolation, etc.) are effectively calibrated and corrected, and they have a negligible effect into the final observable. The on-orbit calibration, performed using an extensive dataset of free-of-rain and low ionospheric activity observations, is successfully used to correct all the collected observations, which are further validated against independent precipitation observations confirming the sensitivity of the observables to the presence of hydrometeors. The validation results also show how vertically averaged ΔΦ can be used as a proxy for precipitation.

Description: In this study we developed a neural network (NN) that can be used to relate a large dataset of multi-angular and multi-spectral polarimetric remote sensing observations to retrievals of cloud microphysical properties. This effort builds upon our previous work, which explored the sensitivity of neural network input, architecture, and other design requirements for this type of remote sensing problem. In particular this work introduces a framework for appropriately weighting total and polarized reflectances, which have vastly different magnitudes and measurement uncertainties. The NN is trained using an artificial training set and applied to Research Scanning Polarimeter (RSP) data obtained during the ORACLES field campaign (Observations of Aerosols above Clouds and their Interactions). The polarimetric RSP observations are unique in that they observe the same cloud from a very large number of angles within a variety of spectral bands resulting in a large dataset that can be explored rapidly with a NN approach. The usefulness applying a NN to a dataset such as this one stems from the possibility of rapidly obtaining a retrieval that could be subsequently applied as a first-guess for slower but more rigorous physical-based retrieval algorithms. This approach could be particularly advantageous for more complicated atmospheric retrievals – such as when an aerosol layer lies above clouds like in ORACLES. For the ORACLES 2016 dataset comparisons of the NN and standard parametric polarimetric (PP) cloud retrieval give reasonable results for droplet effective radius (re : R = 0.756, RMSE = 1.74 μm) and cloud optical thickness (τ : R = 0.950, RMSE = 1.82). This level of statistical agreement is shown to be similar to comparisons between the two most well-established cloud retrievals, namely the the polarimetric cloud retrieval and the bispectral total reflectance cloud retrieval. The NN retrievals from the ORACLES 2017 dataset result in retrievals of re (R = 0.54, RMSE = 4.77 μm) and τ (R = 0.785, RMSE = 5.61) that behave much more poorly. In particular we found that our NN retrieval approach does not perform well for thin (τ  〈3), inhomogeneous, or broken clouds. We also found that correction for above-cloud atmospheric absorption improved the NN retrievals moderately – but retrievals without this correction still behaved similarly to existing cloud retrievals with a slight systematic offset.