| This article has been：Browse 207Times Download 124Times |
scan it! |
|
|
| DOI：10.13522/j.cnki.ggps.2025397 |
|
| Estimating leaf area index and chlorophyll content of Gynura procumbens (Lour.) Merr. using UAV multispectral remote sensing |
|
Song Yixin, Pei Qingbao, Wu Lifeng, Li Liang, Han Yu, Luo Taiyou, Wu Runnan
|
|
1. Jiangxi University of Water Resources and Electric Power, Nanchang 330099, China;
2. East China University of Technology, Nanchang 330013, China
|
| Abstract: |
| 【Objective】Leaf area index (LAI) and chlorophyll content are key physiological traits, and their traditional measurement methods are destructive and labor-intensive. The advances in UAV multispectral remote sensing provide a non-destructive alternative for monitoring them in the field. This paper studies the potential of UAV multispectral data for estimating LAI and chlorophyll content of Gynura procumbens (Lour.) Merr.【Method】Multispectral imagery was acquired using a UAV flying at three heights: 20, 40 and 60 m. We selected 12 empirical vegetation indices derived from imagery acquired at different flying heights and analyzed their correlations with field-measured LAI and chlorophyll content. Based on the correlation results, the three most sensitive vegetation indices were used to construct a Random Forest, a Gradient Boosting Decision Tree, and a Ridge Regression model to improve the estimation of LAI and chlorophyll content of Gynura procumbens (Lour.) Merr.【Result】①There were no significant correlations between LAI and vegetation indices derived from imageries acquired at flying height of 20 m. At flying height of 40 m, the chlorophyll absorption ratio index (CARI) derived from the imageries correlated strongly with both chlorophyll content and LAI, with correlation coefficients of 0.797 and 0.570, respectively. At flying height of 60 m, the triangular vegetation index derived from the imageries had a strong correlation with LAI, with a correlation coefficient of 0.707. ②Among all three models we compared, the Random Forest inversion model was most accurate for estimating chlorophyll content and LAI, with their associated coefficient of determination being 0.837 and 0.659, and root mean square error (RMSE) being 0.041 and 4.046, respectively.【Conclusion】The Random Forest model using UAV imagery acquired at a flying height of 40 m is most accurate for estimating canopy chlorophyll content and LAI of Gynura procumbens (Lour.) Merr. |
| Key words: UAV; multispectral; Gynura procumbens (Lour.) Merr; leaf area index; chlorophyll content |
|
|
