Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Mehzooz Nizar, Jha K. Ambuj, Manmeet Singh, Vaisakh S. B, G. Pandithurai

TL;DR

CloudSense tackles the problem of cloud-type classification for radar-based precipitation estimates in the Western Ghats by leveraging vertical reflectivity profiles from an X-band radar and an SMOTE-balanced feature set derived from three height zones. The approach experiments with seven ML models and identifies LightGBM as the best performer, achieving BAC ~0.80 and F1 ~0.82, and it outperforms a conventional radar-threshold method (BAC ~0.69, F1 ~0.68) on a 200-sample test. The study demonstrates strong per-class performance, notably for shallow clouds, while highlighting some confusions between mixed stratiform-convective and other types. The findings suggest ML-based cloud classification can meaningfully improve QPE in mountainous regions and motivate broader application to additional sites and radar datasets.

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.

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

TL;DR

CloudSense tackles the problem of cloud-type classification for radar-based precipitation estimates in the Western Ghats by leveraging vertical reflectivity profiles from an X-band radar and an SMOTE-balanced feature set derived from three height zones. The approach experiments with seven ML models and identifies LightGBM as the best performer, achieving BAC ~0.80 and F1 ~0.82, and it outperforms a conventional radar-threshold method (BAC ~0.69, F1 ~0.68) on a 200-sample test. The study demonstrates strong per-class performance, notably for shallow clouds, while highlighting some confusions between mixed stratiform-convective and other types. The findings suggest ML-based cloud classification can meaningfully improve QPE in mountainous regions and motivate broader application to additional sites and radar datasets.

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.
Paper Structure (13 sections, 1 equation, 15 figures, 6 tables)

This paper contains 13 sections, 1 equation, 15 figures, 6 tables.

Table of Contents

  1. INTRODUCTION
  2. DATA AND METHODOLOGY
  3. Data
  4. Methodology
  5. Processing of Z data
  6. Labelling
  7. CloudSense
  8. Tuning and Evaluating the model
  9. RESULTS AND DISCUSSIONS
  10. Evolution case of a convective system
  11. CloudSense Results
  12. CloudSense versus radar algorithm output
  13. SUMMARY AND CONCLUSIONS

Figures (15)

  • Figure 1: Topography map of the Western Ghats. Black dots show the locations of radar site (MDV) and HACPL. Dotted circle represents the X-band radar's maximum range (125 km). RHI scan is taken along the dash-dotted line
  • Figure 2: a) Typical RHI scan observed on X July, 2018. Vertical black line is drawn at 26 km to show VPRs considered over HACPL. b) Vertical profile of Doppler velocity from a PPI scan at 85º elevation
  • Figure 3: Logical flow diagram for classifying clouds
  • Figure 4: Architecture of CloudSense
  • Figure 5: CFADs of Z for a) Stratiform b) Mixed Stratiform-Convective c) Convective and d)Shallow clouds
  • ...and 10 more figures