Scholarlink

<p>The remote sensing data products from space-borne LiDAR missions, such as Global Ecosystem Dynamics Investigation (GEDI) are expected to help in measuring forest canopy heights. The study area has 53% forest area and rugged topography with an average slope of 27°. This study attempts to integrate GEDI, optical and backscatter data using random forest regression to retrieve a canopy height model in the Western Himalaya. Comparison of relative height metrics with field data confirmed RH95 as the most suitable metric for canopy height estimation. Seven footprint selection criteria based on acquisition parameters viz. beam type, acquisition time, sensitivity and land surface characteristics slope and land cover were evaluated for modelling. Signal-to-noise ratio (SNR) for each selection criteria was calculated. The power beam type had a higher SNR of 15.14 compared to the coverage beam type, which had an SNR of 11.82. However, a similar difference in SNR was not observed between night-time versus daytime acquisitions. Prediction accuracy improved when low-slope (&lt;30°) and power-beam footprints were used; however, excessive filtering reduced model performance despite its intent to minimize noise. Canopy height modeling combined backscatter (Sentinel-1, PALSAR) and optical data (Sentinel-2, Landsat). Long-term optical variables showed high variable importance. The optimum model performance (MAE = 3.91&#xa0;m, RMSE = 4.93&#xa0;m, R² = 0.63) was achieved using power-beam footprints across all LULC classes with slopes &lt;30°. Models restricted to forest-only and additional night-time footprints selection criteria performed poorly (MAE = 4.99&#xa0;m, RMSE = 6.02&#xa0;m, R² = 0.43). This study uniquely demonstrates the integration of optimized GEDI footprint selection with long-term optical and backscatter data to improve canopy height estimation in complex mountainous terrains.</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

Optimizing GEDI Footprint Selection for Forest Canopy Height Estimation in the Western Himalaya Using Multi-sensor Data

  • 期刊论文

  • 发表日期：

    2025年11月12日

查看全文
  • Akshay Paygude,
  • Hina Pande,
  • Poonam Seth Tiwari,
  • Praveen Kumar Varma,
  • Manoj Kumar

摘要

The remote sensing data products from space-borne LiDAR missions, such as Global Ecosystem Dynamics Investigation (GEDI) are expected to help in measuring forest canopy heights. The study area has 53% forest area and rugged topography with an average slope of 27°. This study attempts to integrate GEDI, optical and backscatter data using random forest regression to retrieve a canopy height model in the Western Himalaya. Comparison of relative height metrics with field data confirmed RH95 as the most suitable metric for canopy height estimation. Seven footprint selection criteria based on acquisition parameters viz. beam type, acquisition time, sensitivity and land surface characteristics slope and land cover were evaluated for modelling. Signal-to-noise ratio (SNR) for each selection criteria was calculated. The power beam type had a higher SNR of 15.14 compared to the coverage beam type, which had an SNR of 11.82. However, a similar difference in SNR was not observed between night-time versus daytime acquisitions. Prediction accuracy improved when low-slope (<30°) and power-beam footprints were used; however, excessive filtering reduced model performance despite its intent to minimize noise. Canopy height modeling combined backscatter (Sentinel-1, PALSAR) and optical data (Sentinel-2, Landsat). Long-term optical variables showed high variable importance. The optimum model performance (MAE = 3.91 m, RMSE = 4.93 m, R² = 0.63) was achieved using power-beam footprints across all LULC classes with slopes <30°. Models restricted to forest-only and additional night-time footprints selection criteria performed poorly (MAE = 4.99 m, RMSE = 6.02 m, R² = 0.43). This study uniquely demonstrates the integration of optimized GEDI footprint selection with long-term optical and backscatter data to improve canopy height estimation in complex mountainous terrains.