Machine Learning ARIMA for Modelling and Forecasting Variations in Earth’s Surface Phenomenon using Sparse Time Series Satellite Data - A Case Study of Sea Surface Salinity in a Tropical Coast

• Author(s): Opeyemi Ajibola-James ; Francis I. Okeke
• Paper ID: 1710037
• Page: 268-277
• Published Date: 14-08-2025
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 9 Issue 2 August-2025

Abstract

The contemporary all-weather satellites observations of Earth’s surface phenomena (ESP) are characterized by relatively sparse time series data that discourage their utilization in building efficient machine learning (ML) models for exploratory and predictive purposes. Additionally, data-poor areas usually have difficulties in meeting the multiple predictor variables requirement of building appropriate multivariate ML regression models. We utilized a relatively sparse sea surface salinity (SSS) dataset from the Soil Moisture Active Passive Mission (SMAP) satellite for this study. We determined the accuracy and variability of the relatively sparse SSS data. We built ML autoregressive integrated moving average (ARIMA) models; determined and validated the best model for modelling and forecasting ESP using the relatively sparse data as a case study. We show root mean squared differences, RMSDs (0.1279 and 0.1162 psu) for the modelling and forecasting data accuracy respectively. We show a standard deviation, SD (0.2528 psu) for the interannual SSS variability (iSSSv). We show the modelling accuracy with an R-squared, R2 (0.8345 psu) and its validation with a mean absolute percentage error, MAPE (0.7779%) for the best model. We show the best variant of the traditional SSS forecasts (“Lo”) accuracy with root mean squared error, RMSE (0.5435 psu) and its validation with MAPE (1.5038%) for the best model. The results suggest relatively high modelling and prediction accuracy. The results imply that relatively sparse satellite time series data of at least 60 epochs can be integrated with a ML ARIMA model for modelling and forecasting variations in any ESP, regardless of the location.

Keywords

Earth’s Surface Phenomenon, Sea Surface Salinity, Machine Learning Arima, VariationsModelling, Time Series Forecasting

Citations

IRE Journals:
Opeyemi Ajibola-James , Francis I. Okeke "Machine Learning ARIMA for Modelling and Forecasting Variations in Earth’s Surface Phenomenon using Sparse Time Series Satellite Data - A Case Study of Sea Surface Salinity in a Tropical Coast" Iconic Research And Engineering Journals Volume 9 Issue 2 2025 Page 268-277

IEEE:
Opeyemi Ajibola-James , Francis I. Okeke "Machine Learning ARIMA for Modelling and Forecasting Variations in Earth’s Surface Phenomenon using Sparse Time Series Satellite Data - A Case Study of Sea Surface Salinity in a Tropical Coast" Iconic Research And Engineering Journals, 9(2)