Scholarlink

<p>Atmospheric Top of the atmosphere (TOA) radiation flux is often calculated using satellite observations, reanalysis data, in the radiative transfer models, however cloud feedback, aerosol interactions, etc. may cause inaccuracies in the calculation due to the complex dynamics and nonlinear interdependencies. Machine learning (ML) offers a promising approach by capturing complex, non-linear atmospheric process relationships. This study develop two separate eXtreme Gradient Boosting (XGBoost) models to compute TOA window region longwave flux (WLW) and reflected shortwave flux (RSW) over Central India from cloud and geophysical data. The data used in the study is Clouds and the Earth’s Radient Energy System (CERES) Single Scanner Footprint Level 3 (SSF1deg) Terra observations for the 2018–2023 period. For both WLW and RSW, models present the performance with R<sup>2</sup> &gt;0.95. And Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) in case of WLW are 2.70% and 2.71&#xa0;W/m<sup>2</sup> whereas for RSW, 14.13% and 6.00&#xa0;W/m<sup>2</sup> respectively. SHAP (SHapley Additive exPlanations) analysis reveals that Liquid Water Path (LWP) and cloud optical depth (COD) act as key drivers of RSW, while cloud top Pressure (CTP) and surface Temperature (ST) are the primary influencers of WLW. Seasonal and cloud type analyses are carried out to understand cloud and atmospheric thermodynamics drive TOA flux. This study highlights XGBoost’s effectiveness in cloud-radiation interaction studies and demonstrates in understanding TOA flux variability and dependence through a ML-driven SHAP analysis. Results also provide an insight on how each cloud property modifies the outgoing longwave and shortwave flux in a statistical approach along with their linear and non-linear dependence.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A cloud radiative flux computation using machine learning approach: influence of cloud properties in modifying longwave and shortwave outgoing radiation

  • 期刊论文

  • 发表日期：

    2026年5月26日

查看全文
  • R. B. Krishnaveni,
  • Archita M. Hari,
  • Ijas Mytheen ,
  • Jyotirmayee Satapathy

摘要

Atmospheric Top of the atmosphere (TOA) radiation flux is often calculated using satellite observations, reanalysis data, in the radiative transfer models, however cloud feedback, aerosol interactions, etc. may cause inaccuracies in the calculation due to the complex dynamics and nonlinear interdependencies. Machine learning (ML) offers a promising approach by capturing complex, non-linear atmospheric process relationships. This study develop two separate eXtreme Gradient Boosting (XGBoost) models to compute TOA window region longwave flux (WLW) and reflected shortwave flux (RSW) over Central India from cloud and geophysical data. The data used in the study is Clouds and the Earth’s Radient Energy System (CERES) Single Scanner Footprint Level 3 (SSF1deg) Terra observations for the 2018–2023 period. For both WLW and RSW, models present the performance with R2 >0.95. And Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) in case of WLW are 2.70% and 2.71 W/m2 whereas for RSW, 14.13% and 6.00 W/m2 respectively. SHAP (SHapley Additive exPlanations) analysis reveals that Liquid Water Path (LWP) and cloud optical depth (COD) act as key drivers of RSW, while cloud top Pressure (CTP) and surface Temperature (ST) are the primary influencers of WLW. Seasonal and cloud type analyses are carried out to understand cloud and atmospheric thermodynamics drive TOA flux. This study highlights XGBoost’s effectiveness in cloud-radiation interaction studies and demonstrates in understanding TOA flux variability and dependence through a ML-driven SHAP analysis. Results also provide an insight on how each cloud property modifies the outgoing longwave and shortwave flux in a statistical approach along with their linear and non-linear dependence.