TreeNet is a well-established project, running since 2011, that monitors the changes in stem radius and environmental conditions of a few hundred trees across Switzerland (https://treenet.info). It runs a fully automated data acquisition system with a central database and is focused on the development of new methods that will accelerate data processing to help us obtain adequate tree physiological answers faster, e.g., in the case of a drought events. To achieve a longer battery lifetime, the sensors continuously emit their data wirelessly over LoRaWAN without using an onsite logger. This means that very often there are transmission errors, resulting in many gaps in the time series. The presence of missing or corrupt values is a common and very frequent problem with time-series data regardless of the sensor type and especially if it is obtained through an automated process such as the TreeNet project. To tackle this problem, we developed an application for gap-filling multi-channel time series data. Although the code was designed around the TreeNet dendrometer sensors, it can be used with any time series data. The core of the application is a deep neural network trained on segments of multi-channel time series (1 hour resolution), composed of tree stem radius changes, temperature, relative humidity, vapor pressure deficit, solar radiation, soil water potential and total precipitation. We therefore provide a pre-trained machine learning model for gap-filling time series data from the TreeNet project.