Designing A Rainfall Potential Prediction Model Based on Atmospheric Stability in Airport Areas

Abstract

Rainfall forecasting in airport areas plays a crucial role in ensuring flight safety and operational reliability, yet local characteristics effect on rainfall intensity are overlooked. This study aims to develop a rainfall occurrence and intensity prediction model using Artificial Neural Network (ANN) based on atmospheric stability indices including Showalter Index (SI), Lifted Index (LI), K Index (KI), Total Totals (TT), Convective Available Potential Energy (CAPE), Convective Inhibition (CIN), and Precipitable Water (PW). Data were obtained from radiosonde observations, Automated Weather Observing System (AWOS), and Meteorological Aerodrome Report (METAR) in Soekarno-Hatta International Airport January 2020-December 2024 at 00 and 12 UTC. We used two stage analysis (1) binary rainfall occurrence prediction and (2) three-class rainfall intensity prediction based on selected feature. Our results reveal that ANN is capable to simulate rainfall occurrence with high accuracy (>0.78), outperforming the minimal-feature model (0.738) and all other configurations. Physically, the atmospheric indices can be grouped into three categories: stability-related indices (SI, LI, KI, TT), energy-related indices (CAPE, CIN), and moisture-related indices (PW), representing key factors influencing convective rainfall in tropical regions. However, the model’s applicability may be limited in study site due to local climatic characteristics and temporal constraints of the dataset. These findings highlight the importance of selecting physically relevant atmospheric parameters and implementing robust data preprocessing to enhance rainfall prediction accuracy in operational aviation contexts.