CloudSense - A Model for Cloud Type Identification using Machine Learning from Radar data

Aug 4, 2026

Upload by Oliver

CloudSense is a machine-learning model designed to identify precipitating cloud types for improved radar-based quantitative precipitation estimates over India’s Western Ghats complex terrain. The method classifies clouds into four categories—stratiform, mixed stratiform-convective, convective, and shallow—using vertical reflectivity profiles collected in July–August 2018 from an X-band Doppler radar. Training uses a SMOTE-balanced dataset with physically meaningful features. Evaluations show LightGBM achieves strong performance, outperforming conventional radar algorithms on tested samples, supporting more accurate precipitation detection and classification.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages39

Reading Time1 h 38 min

AccessFree to download

Abstract

Introduction

Importance of cloud-type knowledge for QPE

Limitations of previous satellite and threshold-based methods

Radar-based precipitation and cloud classification approaches

Doppler radar data and conventional algorithms

arXiv:2405.05988v1 [[physics. ao-ph](physics. ao-ph)] 8 May 2024 CloudSense: A Model for Cloud Type Identification using Machine Learning from Radar data Mehzooz Nizar 1,2 , Jha K. Ambuj 1 , Manmeet Singh 1,3 , Vaisakh S.B 1 , and G. Pandithurai 1 1 Indian Institute of Tropical Meteorology, Ministry of Earth Sciences, Pune, India 2 Cochin University of Science and Technology, Kochi, India 3 Jackson School of Geosciences, The University of Texas at Austin, Austin, Texas, USA ABSTRACT The knowledge of type of precipitating cloud is crucial for radar based quantitative estimates of precipitation. We propose a novel model called CloudSense which uses machine learning to accurately identify the type of precipitating clouds over the complex terrain locations in the Western Ghats (WGs) of India. CloudSense uses vertical reflectivity profiles collected during July-August 2018 from an X-band radar to classify clouds into four categories namely stratiform,mixed stratiform-convective,convective and shallow clouds. The machine learning(ML) model used in CloudSense was trained using a dataset balanced by Synthetic Minority Oversampling Technique (SMOTE), with features selected based on physical characteristics relevant to different cloud types. Among various ML models evaluated Light Gradient Boosting Machine (LightGBM) demonstrate superior performance in classifying cloud types with a BAC of 0 .8 and F1-Score of 0 .82 . CloudSense generated results are also compared against conventional radar algorithms and we find that CloudSense performs better than radar algorithms. For 200 samples tested, the radar algorithm achieved a BAC of 0 .69 and F1-Score of 0.68, whereas CloudSense achieved a BAC and F1-Score of 0.77. Our results show that ML based approach can provide more accurate cloud detection and classification which would be useful to improve precipitation estimates over the complex terrain of the WG. Keywords: Machine learning, Precipitating clouds, Doppler weather radar, Western Ghats, LightGBM 1 INTRODUCTION Clouds are an integral component of convection and precipitation and playa vital role in modulating the global circulation, Earth’s radiative budget and hydrological cycle. Different cloud types are associated with different microphysical and radiative properties which influence the vertical distribution of heating in the atmosphere (Houze 1982; Houze 1997; Schumacher and Houze 2003) . Improved knowledge of cloud types can improve weather and climate predictions through better representation of clouds in atmospheric models. From a regional perspective, accurate detection and classification of precipitating clouds are crucial for accurate quantitative precipitation estimation (QPE) which is widely used for extreme weather forecasting, climate studies and hydrological applications (Steiner and Houze, 1997; Gourley and Vieux 2005; K¨uhnlein et al.,2014; Thompson et al. , 2015; Arulraj and Barros,2019) . QPE algorithms provide rain estimates based on relationships specific to precipitating cloud type (Rao et al. 2001; Auipong and Trivej 2018) and therefore, any misclassification of precipitating clouds will impact the accuracy of QPE. Earlier studies classified precipitating clouds either using space-borne active and passive sensors (Inoue 1987; Sassen and Wang 2008; Subrahmanyam and Kumar 2013; So and Shin 2018) or ground-based radars and gauges (Penide et al., 2013; Loh et al., 2020; Zuo et al.,2022) . Satellite-based classification though has wider spatial coverage but has limitations due to coarser spatial and temporal sampling. Broadly, precipitating clouds are categorized as convective and stratiform types (Houze 2014) . Convective clouds are characterised by strong vertical air currents, small areal coverage, and high rainfall intensities while the converse is associated with stratiform clouds. Most techniques apply a threshold value to distinguish between different precipitation types. For example

FAQs

What problem does CloudSense address?

CloudSense targets accurate identification of precipitating cloud types, which is crucial for improving radar-based quantitative precipitation estimates (QPE).

How does CloudSense classify cloud types?

It uses vertical reflectivity profiles from an X-band radar and a machine-learning approach to classify clouds into four categories: stratiform, mixed stratiform-convective, convective, and shallow.

Which model shows the best performance and how does it compare to conventional algorithms?

LightGBM performs best among the evaluated ML models, achieving higher BAC and F1-scores than conventional radar algorithms on the tested samples.

Aug 4, 2026

Upload by Oliver