Remote Sensing Leaf Chlorophyll Content Using a Visible Band Index

Published in Agron. J. 103:1090–1099 (2011) Posted online 2 May 2011 doi:10.2134/agronj2010.0395 Copyright © 2011 by the American Society of Agronomy, 5585 Guilford Road, Madison, WI 53711. All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. I to estimate leaf chlorophyll content (μg cm–2), such as leaf chlorophyll meters, can be used to manage the amount of N fertilizer applied to crops based on site-specifi c requirements (Schepers et al., 1992; Varvel et al., 2007). Remote sensing of chlorophyll content in crop canopies may help provide a low-cost alternative to plant or soil sampling (Scharf et al., 2002; Gitelson et al., 2005; Hatfi eld et al., 2008). Vegetation indices are a simple method to reduce large data volumes from remote sensing to information useful for management. Besides leaf chlorophyll content, vegetation indices are also sensitive to diff erences of soil refl ectance, LAI, canopy cover, and canopy architecture (Eitel et al., 2009). Various vegetation indices are sensitive to canopy variables to diff erent degrees (Daughtry et al., 2000), leading to combination indices, in which one index adjusts for diff erences of LAI, so that the combination index is more sensitive to leaf chlorophyll content (Haboudane et al., 2002, 2004; Eitel et al., 2007, 2008, 2009). Newer indices are based on narrow-band imaging spectrometers (also called hyperspectral sensors), which are currently expensive and create very large data volumes. Th e research forefront of imaging spectroscopy is the estimation of leaf chlorophyll content and other variables by model inversion, including atmospheric and topographic corrections (Botha et al., 2007; Houborg and Anderson, 2009; Houborg et al., 2009; Jacquemoud et al., 2009). However, agricultural management generally requires information within short windows of time, and it is uncertain that more detailed information will lead to better decisions for N management. Th e human eye is very sensitive to changes in the green color of leaves caused by changes in chlorophyll content (unless the person is colorblind), but without some sort of aid, it is very diffi cult to quantify these changes on a consistent basis through space and time (Singh et al., 2002). Color aerial photography is useful for determining areas with N defi ciency (Blackmer et al., 1996; Scharf et al., 2002). Digital cameras are widely available and are sensitive to chlorophyll diff erences of leaves (Adamsen et al., 1999; Karcher and Robinson, 2003; Dani et al., 2005; Jia et al., 2007; Meyer and Carmargo Neto, 2008; Pagola et al., 2009). One major diff erence between digital camera bands and the broad bands of multispectral sensors (such as the Landsat Th ematic Mapper [TM]) is the considerable overlap of wavelength ranges for the three bands of a digital camera (Hunt et al., 2005). Furthermore, in most digital color cameras, the red, green, and blue pixels are arranged in a Bayer color fi lter array, with twice as many green as red or blue pixels. Demosaicing and interpolation are used to estimate the digital numbers of the two colors not sensed by a given pixel. Vegetation indices calculated from combinations of red, green, and blue bands have been studied previously. Usually indices that perform well at a leaf scale do not perform well at the canopy scale. For example, the normalized green red diff erence index (NGRDI):