Numerous studies have shown that lakes are sensitive worldwide to climate change, but these studies focus almost exclusively on deep (mono- and dimictic) lakes and reservoirs. Changing meteorological forcings affect the temperature structure and energy balance of lakes. In Lake Balaton, the record algal bloom in the late summer of 2019 has also been shown to be closely related to the diurnal temperature stratification of the lake. Unfortunately, long-term water temperature profile observations are not available for large lakes in Hungary, allowing us to accurately and quickly assess how their stratification has changed. Furthermore, we do not have a tool yet to investigate what changes may be expected in the future due to climate change. The aim of the TDK research is to set up a robust, low-computational-demand neural network-based model that can be used to efficiently study and predict the diurnal stratification of our shallow lakes (e.g., Lake Balaton). The development of the model requires the determination of the model input hydrometeorological variables based on detailed measurements. After successful model validation, the model should be used to estimate stratification conditions for the past and future decades. A further objective is to test ECMWF ERA5 reanalysis data. The question is whether they are accurate enough to model weak temperature stratifications.