Leaf Area Index Retrieval Using High Resolution Remote Sensing Data

• Main Authors: Michele Rinaldi, Sergio Ruggieri, Pasquale Garofalo, Alessandro Vittorio Vonella, Giuseppe Satalino, Pietro Soldo
• Format: Article
• Language: English
• Published: PAGEPress Publications 2010-06-01
• Series: Italian Journal of Agronomy
• Online Access: https://www.agronomy.it/index.php/agro/article/view/105

Vegetation indices obtained from remote sensed data can be used to characterize crop canopy on a large scale using a non-destructive method. With the recent launch of the IKONOS satellite, very high spatial resolution (1 meter) images are available for the detailed monitoring of ecosystems as well as for precision agriculture. The aim of this study is to evaluate the accuracy of leaf area index (LAI) retrieval over agricultural area that can be obtained by empirical relationships between different spectral vegetation indices (VI) and LAI measured on three different dates over the spring-summer period of 2008, in the Capitanata plain (Southern Italy). All the VIs used (NDVI, RDVI, WDVI, MSAVI and GEMI) were related to the LAI through exponential regression functions, either global or crop-dependent. In the first case, LAI was estimated with comparable accuracies for all VIs employed, with a slightly higher accuracy for GEMI, which determination coefficient achieved the value of 0.697. Whereas the LAI regression functions were calculated separately for each crop, the WDVI, GEMI and RDVI vegetation indices provided the highest determination coefficients with values close to 0.90 for wheat and sugar beet, and with values close to 0.70 for tomatoes. A validation of the models was carried out with a selection of independent sampling data. The validation confirmed that WDVI and GEMI were the VIs that provided the highest LAI retrieval accuracies, with RMSE values of about to 1.1 m2 m-2. The exponential functions, calibrated and validated to calculate LAI from GEMI, were used to derive LAI maps from IKONOS high-resolution remote sensing images with good accuracy. These maps can be used as input variables for crop growth models, obtaining relevant information that can be useful in agricultural management strategies (in particular irrigation and fertilization), as well as in the application of precision farming.