hariprasath-v / Solafune_Aerosol_Optical_Depth_Estimation

Estimate AOD using Sentinel-2 data, leveraging its spectral bands and spatial resolution to improve accuracy.

Solafune Aerosol Optical Depth Estimation

Competition hosted on Solafune

Overview

Aerosol Optical Depth (AOD) is a critical parameter for understanding air quality and climate change, as it quantifies the amount of aerosols present in the atmosphere. The Sentinel-2 mission, part of the European Space Agency's Copernicus program, provides high-resolution multispectral imagery, offering a valuable resource for remote sensing applications. In this competition, participants are challenged to estimate AOD using Sentinel-2 data, leveraging its spectral bands and spatial resolution to improve accuracy. This endeavor aims to enhance the precision of AOD measurements, facilitating better environmental monitoring and decision-making. Accurate AOD estimation will have significant implications for public health, weather forecasting, and climate research. The ultimate goal is to advance the methodologies for AOD estimation, contributing to global efforts in environmental protection and climate change mitigation. Participants' contributions will pave the way for more accurate and reliable AOD datasets, benefiting researchers, policymakers, and society at large.

Data Overview

Sentinel-2 images

The Sentinel-2 image contains information on 12 bands and includes the following bands 'B1', 'B2', 'B3', 'B4', 'B5', 'B6', 'B7', 'B8', 'B8A', 'B9', 'B10', 'B11', 'B12'

For more information, please visit this site

The images have been processed to mask clouds for Sentinel-2 images from 2016/1/1 to 2024/5/1, and each image that sentinel-2 captured is based on the availability of the satellite images at the time combined with the AERONET dataset.

Competition Policy

cf. @solafune(https://solafune.com) Use for any purpose other than participation in the competition or commercial use is prohibited. If you would like to use them for any of the above purposes, please contact us.

Evaluation metric is Pearson R Coefficient

My Approach

Exploratory Data Analysis

• Target distribution analysis.
• Visualize a sample image with 13 bands based on minimum and maximum aerosol optical depth.
• Histogram analysis for each band.
• Visualize popular Sentinel-2 RGB composites based on minimum and maximum aerosol optical depth.
• Visualize remote sensing indices based on minimum and maximum aerosol optical depth:
  • NDVI - Normalized Difference Vegetation Index
  • NDWI - Normalized Difference Water Index
  • FMI - Ferrous Mineral Index
  • MSI - Moisture Stress Index
  • BSI - Bare Soil Index
  • NBR - Normalized Burn Ratio
• Band-wise reflectance and correlation analysis.
• Gray-Level Co-occurrence Matrix (GLCM) texture features and correlation analysis.

Model-1

Data Preparation

Data was split into training and validation sets randomly.

Band-wise pixels were chosen, and embeddings were created using the timm efficientnet_b0.ra_in1k model.

An XGBoost regressor model was created for 13 bands and evaluated on the validation dataset using the Pearson correlation coefficient score.

Band-wise validation results,

Band	Validation_score
band_6	82
band_2	81
band_5	79
band_7	78
band_9	77
band_1	75
band_3	75
band_8	74
band_4	67
band_12	62
band_10	58
band_13	42
band_11	3

Model-2

Data Preparation

Data was split into training and validation sets randomly.

Embeddings were created using the timm efficientnet_b0.ra_in1k model.

An XGBoost regressor model was created for the following popular RGB composites and evaluated on the validation dataset using the Pearson correlation coefficient score.

• Natural Color
• False Color Infrared
• False Color Urban
• Agriculture
• Atmospheric Penetration
• Healthy Vegetation
• Land/Water
• Natural Colors with Atmospheric Removal
• Shortwave Infrared
• Vegetation Analysis

RGB composites validation results,

RGB_Composites	Validation_score
False_color_Infrared	81
Shortwave_Infrared	80
Healthy_Vegetation	78
Natural_Colors	76
False_color_Urban	72
Vegetation_Analysis	72
Land_Water	71
Natural_Colors_with_Atmospheric_Removal	70
Agriculture	67
Atmospheric_Penetration	67

Model-3

Created statistical features for extracted spectral indices, texture, and slope features.

Trained a CatBoost model with a validation score of 0.96.