Cold pools are mesoscale bodies of cool and dense air that form underneath precipitating clouds and spread horizontally on the Earth’s surface. Lifting of environmental air at their leading edge constitutes the importance of cold pools for the development and organization of atmospheric convection. Although being explicitly represented in hectometer-scale numerical simulations, cold pools and especially their spatial properties are hardly captured by existing observation systems. This dissertation presents a comprehensive characterization of cold pools in Germany, discusses factors controlling their strength, and describes the Field Experiment on Sub-mesoscale Spatio Temporal Variability in Lindenberg (FESSTVaL) that determines the morphology of cold pools (i. e., their size, shape, and structure) for the first time ever using in situ observations. The first part presents a preparatory study on the analysis of nearly 500 cold pool passages at a 280-m high tower in Hamburg (northern Germany) observed over 14 years. The sampled cold pools most likely occur in July and in the afternoon. The near-surface temperature perturbation ∆T exhibits a median of -3.3 K and can be as strong as -10.8 K. ∆T is more strongly correlated with the saturation deficit of the pre-event air mass (r = −0.71 ) than with the event-accumulated rainfall amount (r = −0.35), emphasizing that the cold pool strength is mainly controlled by evaporative cooling by precipitation. This process becomes occasionally superposed by the downward transport of upper-level air masses in convective downdrafts. The second part describes the realization of FESSTVaL that took place at the Meteorological Observatory Lindenberg – Richard Aßmann Observatory (MOL- RAO) in eastern Germany from May to August 2021. The experiment featured a dense network of 99 measurement stations at sub-mesoscale resolution (O(100) m – O(10) km) that covered a circular area of about 30 km in diameter. For this purpose, novel, custom-designed Autonomous cold POoL LOgger (APOLLO) stations equipped with fast-response air temperature and pressure sensors as well as supplementary Weather Transmitter (WXT) stations were developed, which proved to operate reliably. The impacts of the COVID-19 pandemic delayed the main experiment by one year and motivated the implementation of the precursor experiment FESST@HH in Hamburg during summer 2020, which was facilitated by the support of volunteers. The last part focuses on the evaluation of the FESSTVaL and FESST@HH data sets. The spatial interpolation of the ∆T signals allows to define individual cold pool objects. The derived sample of about 1300 cold pool objects during 39 events exhibits a median size of 8.4 km, whereas the size of cold pools increases with their strength. Contrarily to common assumption, the cold pools are not circularly shaped, which is independent of their size and strength. The analysis of four selected events suggests that cold pools grow linearly with the spatially-integrated rainfall accumulation and cool most efficiently early in their life cycle, confirming the cold air production by precipitation. This dissertation demonstrates that dense station networks illuminate the previously obscured morphology of cold pools and allow to more precisely describe the underlying processes required for the validation and improvement of weather and climate simulations. Furthermore, the work provides valuable guidance for the implementation of future measurement campaigns.