This work presents and evaluates an algorithm that automatically conducts the steps of fitting peaks and identifying formulas – necessary but time-consuming steps for most applications of mass spectrometry in atmospheric science. The aim of the algorithm is to save researchers working on these tasks significant amounts of time and allow them to proceed with their analysis. The work demonstrates that this algorithm can achieve the goal of speeding up analysis and provide accurate formulas.

We developed an advanced algorithm for global retrieval of TROPOspheric Monitoring Instrument (TROPOMI) HCHO and NO₂ vertical column densities with much improved consistency. Sensitivity tests demonstrate the complexity and nonlinear interactions of auxiliary parameters in the air mass factor calculation. An improved agreement is found with measurements from a global ground-based instrument network. The scientific retrieval provides a useful source of information for studies combining HCHO and NO₂.

Information on the energy transported by atmospheric gravity waves (GWs) is crucial for improving atmosphere models. Most space-based studies report the potential energy. We use Aeolus wind data to estimate the kinetic energy (density). However, the data quality is a challenge for such analyses, as the accuracy of the data is in the range of typical GW amplitudes. We find a temporal coincidence between enhanced or breaking planetary waves and enhanced gravity wave kinetic energy density.

We developed an automated, flexible atmospheric sampling device for various platforms. During a 5 d field campaign in the Mount Qomolangma region, we performed 15 flights using the device mounted on a hexacopter unoccupied aerial vehicle (UAV). A total of 139 samples were analyzed using an Agilent 7890A gas chromatograph. Vertical profiles of four greenhouse gases (carbon dioxide (CO₂), methane (CH₄), nitrous oxide (N₂O), and sulfur hexafluoride (SF₆)) were analyzed and discussed.

The study examines the use of the spheroidal shape approximation for calculating cloud radar observables in rain and identifies some limitations. To address these, it introduces the empirical scattering model (ESM) based on high-quality Doppler spectra from a 94 GHz radar. The ESM offers improved accuracy and directly incorporates natural rain's microphysical effects. This new model can enhance retrieval and calibration methods, benefiting cloud radar polarimetry experts and scattering modelers.

A relative calibration technique is developed for the cloud radar by monitoring the intercept of the wet-radome attenuation log-linear behavior as a function of rainfall rates in light and moderate rain conditions. This resulting reflectivity offset during the recent field campaign is compared favorably with the traditional disdrometer comparison near the rain onset, while it also demonstrates similar trends with respect to collocated and independently calibrated reference radars.

This paper presents our latest development in tomographic reconstruction algorithms that use multi-angle (sub)millimeter-wave brightness temperature to reconstruct the spatial distribution of ice clouds. Compared to nadir-only retrievals, the tomography technique provides a detailed reconstruction of ice clouds’ inner structure with high spatial resolution and significantly improves retrieval accuracy. Also, the technique effectively increases detection sensitivity for small ice cloud particles.

Currently, measurement of methane from the TROPOMI satellite is biased with respect to surface reflectance. This study demonstrates a new method of correcting for this bias on a seasonal timescale to allow for differences in surface reflectance in areas of intense agriculture where growing seasons may introduce a reflectance bias. We have successfully implemented this technique in the Denver–Julesburg basin, where agriculture and methane extraction infrastructure is often co-located.

The large fleet of Earth observation satellites in orbit currently generate huge volumes of data, requiring significant computational resources to process these data in a timely manner. We present a method for predicting poor-quality measurements using machine learning. We find that machine learning methods can accurately predict poor-quality measurements and remove them from the processing chain, saving time and computational resources.