Daily seamless dataset of HCHO concentrations: Vertical relationship between surface and column HCHO in China in 2019–2022

Great efforts have been made to observe atmospheric volatile organic compound (VOC) concentrations. However, observation-based VOC datasets at surface level with broad temporal and spatial coverage are still lacking. This gap poses a barrier to understanding the vertical relationship between surface and column VOC concentrations, a critical scientific question limiting the application of satellite observations to infer surface VOCs. To bridge this gap, we developed a multistep deep learning (DL) framework. This framework integrates Tropospheric Monitoring Instrument (TROPOMI) satellite observations, GEOS-Chem model simulations, and meteorological and geographical data to generate a seamless, daily-resolution dataset of formaldehyde (HCHO) vertical column densities (VCDs) and surface concentrations over eastern China for the period 2019–2022. HCHO was chosen as a high-yield intermediate product of VOC oxidation and a good proxy for VOC sources. The dataset quality was assessed through comparisons with independent TROPOMI observations and ground-based in situ measurements. Our analysis revealed good consistency in the variability between surface and column HCHO concentrations at the annual scale, with correlation coefficients of 0.90 (spatial variability), 0.91 (temporal variability) and 0.81 (interannual trend). Nevertheless, this vertical relationship significantly weakened during summer, a period of peak VOC and ozone activity, due to the reduced sensitivity of HCHO VCDs to meteorological factors. Our analysis further provides useful insights into the seasonality and interannual trends in HCHO concentrations in the studied period. The observation-constrained dataset and its analysis is helpful for improving our understanding of HCHO variability in China and is useful for better applications of space-based HCHO observations.