Aircraft Type Recognition (ATR) from Remote Sensing Images (RSIs) is a critical application in military and civil domains. However, existing datasets in this domain possess challenges such as class imbalance, resulting in insufficient samples representing certain classes, and a lack of diversity, which is necessary to accurately depict real-world situations. This paper explores the impact of performing data augmentation in an attempt to address such challenges associated with the MTARSIFAIR1M dataset, which has been created by combining the benchmark MTARSI2 and FAIR1M datasets. Various data augmentation techniques were applied to obtain a balanced representation of 50 aircraft types under diverse conditions, including different backgrounds, nighttime imagery, and occlusions. The impact of these augmentations on model performance was evaluated using state-of-the-art deep learning models. Although the augmented dataset improved the robustness of ATR models allowing them to recognize more diverse images, results indicate a need for further research to improve performance under challenging conditions. This work highlights the potential of data augmentation in enhancing ATR from RSIs and lays a foundation for future advancements in the field.