SII-NowNet - A machine learning tool for nowcasting convection initiation and intensification in the Tropics

Aug 5, 2026

Nowcasting developing convection is a crucial component of early warning systems in the Tropics. While machine learning has proven effective for radar-based nowcasting, limited radar coverage across much of the region creates a capability gap. The study presents SII-NowNet, a neural-network tool that uses satellite brightness temperatures to generate probabilistic nowcasts for intensifying and initiating convection. Results over Sumatra show skill for 1–6 hours for intensification and up to 2 hours for initiation. Sensitivity tests indicate reduced training needs before climatology becomes superior. The model generalises across multiple Tropical regions without retraining.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages37

Reading Time1 h 33 min

AccessFree to download

Abstract

Significance Statement

1. Introduction

Tropical convection and forecasting challenges

Nowcasting and its role in early warning

Manuscript (non-LaTeX) SII-NowNet: A machine learning tool for nowcasting convection initiation and intensification in the Tropics Joseph Smith1, Cathryn E. Birch 1, John Marsham 1,3, David Moffat2 1 School of Earth and Environment, University of Leeds, Leeds, LS2 9JT, UK 2Plymouth Marine Laboratory, Prospect Place, PL1 3DH, Plymouth, UK 3UK Met Office, Exeter, EX1 3PB, UK Corresponding author: Joseph Smith, [eejasm@leeds.ac.uk.com](eejasm@leeds.ac.uk.com) [File generated with AMS Word template 2.0](File generated with AMS Word template 2.0) 1 Early Online Release: This preliminary version has been accepted for publication in Artificial Intelligence for the Earth Systems, may be fully cited, and has been assigned DOI 10. 1175/AIES-D-25-0043.1. The final typeset copyedited article will replace the EOR at the above DOI when it is published. © 2025 The Author(s) . Published by the American Meteorological Society. This is an Author Accepted Manuscript distributed under the terms of the Creative Commons Attribution 4.0 International (CC BY 4 .0) License. Unauthenticated | Downloaded 01/14/26 04:36 PM UTC ABSTRACT Nowcasting developing convection is a crucial component of early warning systems in the Tropics. While machine learning has proven effective for radar-based nowcasting, the lack of radar coverage across much of the Tropics creates a significant capability gap. This study presents Simple Initiation and Intensification Nowcasting neural Network (SII-NowNet), a machine learning tool that uses satellite brightness temperatures to produce probabilistic nowcasts of intensifying and initiating convection in the Tropics. SII-NowNet is first demonstrated over Sumatra, Indonesia—a densely populated tropical island with frequent convective activity. For nowcasts of intensifying convection, SII-NowNet outperforms an optical flow model for lead times of 1–6 hours but begins to over-predict events beyond 3 hours, indicating its limit of capability. For nowcasts of initiating convection, SII-NowNet’s limit of capability is reached at 2 hours, beyond which it over-predicts events and is outperformed by climatology. SII-NowNet is trained on 8,661 samples (12 months of data), but sensitivity testing shows that the number of samples can be reduced to three weeks for intensification and three months for initiation, before its outperformed by climatology. This has practical implications for the implementation and further development of SII-NowNet in resource-constrained settings. To exemplify generalisability in other Tropical regions, SIINowNet is tested over New Guinea, Zambia, Congo and West Africa. Without retraining or region-specific tuning, SII-NowNet achieves skill scores comparable to those over Sumatra. Overall, SII-NowNet’s promising results, combined with ease of applicability across the Tropics, make it a valuable tool for future operational nowcasting. SIGNIFICANCE STATEMENT In the Tropics, short-term forecasts of destructive, rapidly developing storms are critical components of early warning systems—essential for protecting lives and minimizing impacts. However, many regions face challenges in accurate forecasting due to limited access to groundbased meteorological observations. This study presents a machine learning tool that uses freely and continuously available satellite data—covering the entire Tropics—to forecast storm development several hours in advance. The tool performs well in five distinct Tropical regions and requires minimal computational resources, showing its suitability for implementation in resource-constrained settings. Furthermore, the results highlight the challenges in capturing 2 File generated with AMS Word template 2.0 Unauthenticated | Downloaded 01/14/26 04:36 PM UTC Accepted for publication in Artificial Intelligence for the Earth Systems. DOI 10. 1175/AIES-D-25-0043 .1. newly initiating storms – a focus for future studies. Continued work on the tool will focus

FAQs

What capability gap does SII-NowNet address for Tropical nowcasting?

SII-NowNet targets the lack of radar coverage across much of the Tropics, which limits radar-based nowcasting systems. It provides probabilistic nowcasts using freely available satellite brightness temperatures.

How does SII-NowNet perform for intensifying convection lead times?

Over Sumatra, SII-NowNet outperforms an optical flow model for lead times of 1–6 hours. It begins to over-predict beyond about 3 hours, indicating a practical limit.

At what lead time does SII-NowNet’s initiation forecasting become less reliable?

For initiating convection, capability is reached at around 2 hours. Beyond this, it over-predicts events and is outperformed by climatology.

Aug 5, 2026

Upload by Theodora