PASTEL is a dataset for LiDAR-based low-altitude autonomous navigation of unmanned aerial systems in complex environments specifically devised to research on SLAM tuning performance and robustness. PASTEL includes sequences with high diversity in the main factors that affect SLAM tuning: type of environment, aerial robot velocity, and 3D LiDAR characteristics. The dataset consists of data captured from three different LiDARS (Velodyne VLP16, Ouster OS0-128 and Ouster OS1-16) simultaneously in distinct types of environments: wide-open spaces with distant obstacles, horizontally confined spaces with open sky, confined GNSS-denied spaces, and also includes sequences where the aerial robot transitions between two or more of these distinct environments. The environments combine both GNSS-denied and GNSS-enabled areas with both confined and wide-open spaces. The dataset consists of measurements from the three LiDAR and Inertial Measurement Units (IMUs) and a total station trajectory combined in a ROS bag format, ground truth maps of the environments obtained by merging multiple multi-station maps, and calibration files. The dataset has been validated in terms of trajectory and map accuracies with well-known SLAM methods. A journal article describing this dataset is published in IEEE Access, an interdisciplinary open access journal under DOI: 10.1109/ACCESS.2025.3603733.