Classification of Cumulonimbus Cloud Formation based on Himawari Images using Convolutional Neural Network model Googlenet

Mohammad Rizal Abidin (UIN Sunan Ampel Surabaya)
Dian candra Rini Novitasari (UIN Sunan Ampel Surabaya)
Hani Khaulasari (UIN Sunan Ampel Surabaya)
Fajar Setiawan (UIN Sunan Ampel Surabaya)

Jurnal Buana Informatika
Vol. 14 No. 02 (2023): Jurnal Buana Informatika, Volume 14, Nomor 2, Oktober 2023

Cumulonimbus clouds (Cb) are dangerous for many human activities. To reduce this effect, a system to classify formations is needed. The formation of Cb clouds can be seen in the Himawari-8 IR image. This research aimed to create a Cb cloud classification system with Himawari-8 IR Enhanced imagery using the GoogleNet model CNN method. The total data used was 2026 image data. Parameter testing was carried out on the CNN GoogleNet model in this study, namely a data distribution ratio of 90:10 and 80:20. The probability of dropout is 0.6, 0.7, and 0.8. and batch sizes of 8, 16, 32, and 64. The trials conducted in this study yielded a sensitivity value of 100.00%, an accuracy of 99.00%, and a specificity of 99.60% obtained from the experimental data distribution of 90:10, probability 0.8, and batch size 8.