Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
5784548 | Physics and Chemistry of the Earth, Parts A/B/C | 2017 | 8 Pages |
Abstract
Reliable and accurate quantification of plant Leaf Area Index (LAI) is critical in understanding its role in reducing runoff. The main aim of the present study was to evaluate the ability of the Red Edge (RE) band derived from RapidEye in estimating LAI for applications in quantifying canopy interception at landscape scale. To achieve this objective, the study also compares the predictive power of two machine learning algorithms (Random Forest-RF and Stochastic Gradient Boosting-SGB) in estimating LAI. Comparatively, the results of the study have demonstrated that the inclusion of spectral information derived from the Red Edge band yields high accurate LAI estimates, when compared to the use of traditional traditional Red, Green, Blue and Near Infra-Red (traditional RGBNIR) spectral information. The results indicate that the use of the four traditional RGBNIR bands yielded comparatively lower R2 values and high Root Mean Squares, Mean Absolute Error (Pinus taeda: R2 of 0.60; the lowest RMSE (0.35Â m2/m2) and MAE of 28); whereas the use of integration of traditional RGBNIRÂ +Â RE in more accurate LAI estimates (Pinus taeda: R2Â =Â 0.65; RMSEÂ =Â 0.30Â m2/m2) and the lowest MAE of 0.23). These findings therefore underscores the importance of new generation multispectral sensors with strategically-position bands and machine learning algorithms in estimating LAI for quantifying canopy interception, especially in resource-poor areas.
Related Topics
Physical Sciences and Engineering
Earth and Planetary Sciences
Geochemistry and Petrology
Authors
Timothy Dube, Onisimo Mutanga, Mbulisi Sibanda, Cletah Shoko, Abel Chemura,