Low Cloud Type Classification System Using Convolutional Neural Network Algorithm

Clouds are one of the most important parameters in doing weather observations. Meteorologist and atmospheric experts are able to determine short-term and long-term weather conditions by observing atmospheric parameters, one of which is cloud types. The goal of this research is to develop a system which is able to recognize cloud patterns through imagery to determine the types so that the subjective factor in cloud observations can be decreased and help observers to observe cloud types more easily. This system was built by implementing the Python programming language and one of the deep learning models, named as Convolutional Neural Network (CNN). The data used in the system are low cloud image data which are divided into 5 (five) classes, namely Cumulunimbus, Cumulus, Stratus, Stratocumulus, and Nombostratus. The result shows that the accuracy achieved the number of 92.22% in the training process with the total loss of 0.20%.