LAI Measurement with Hemispherical Photographs at Variable Conditions for Assessment of Remotely Sensed Estimations
Alexandridis, Thomas; Stavridou, Domna; Strati, Stavroula; Silleos, Nikolaos
Aristotle University of Thessaloniki, School of Agriculture, GREECE
The leaf area index (LAI) of a plant canopy is defined as its leaf area per unit of ground area. It is an important input parameter for several studies related to water, gases and energy exchange on the Earth surface, and several algorithms have been developed to estimate LAI from remotely sensed images. Nevertheless, in-situ estimations are required for the development and validation of these algorithms, which is often the most costly part of a project. There are two general methods for in-situ measurement of LAI: direct and indirect methods. Direct methods are based on actual leaf area measurements, which are very accurate but extremely laborious. Indirect methods are mostly optical observations based on the measurement of light transmission through canopies, which are faster and thus more widely applied. The source of errors during acquisition or analysis of digital hemispherical canopy photography are position, exposure, evenness of sky lighting, distinguishing foliage from canopy openings, leaf angle variability and consideration of clumping. Specifically for exposure settings (shutter speed, lens aperture and ISO sensitivity) its influence has been shown to be important for the selection of the threshold between vegetation and background sky. Several instructions have been found in literature, however some seem arbitrarily related to the automatic exposure of the camera, and further investigation is needed to achieve standards. The aim of this work is to identify a procedure for LAI in-situ estimations with hemispherical lens photographs that guarantees comparable measurements across variable acquisition conditions. The experimental design is based on the assumption that LAI should be constant in a site throughout the day, and in various cloud conditions. The effect of variable illumination conditions has been examined with taking hourly time-series of hemispherical photos at the exact same locations, under the same canopy conditions. Several methods for processing the hemispherical photos and estimating LAI were tested, along with the correction methods presented in the literature. Moreover, two calibration techniques were tested to minimize the effect of clear vs. cloudy and completely overcast sky. Finally, two computer software specialized in hemispherical photos processing were used. Results suggest that calibration using the sky lighting is a good means for repeatability of observations under closed canopies. Generally, the LAI-2000 method is accepted to be closer to a standard reference, closely followed by the Lang method. The correction methods did not provide any consistent conclusions. Finally, a protocol was put together to facilitate surveyors. The resulting LAI measurements were compared with validated LAI maps from various sources and resolutions: SPOT XS, Resourcesat-2 LISS-III, MODIS Terra, and SPOT VEGETATION. The results of the statistical analyses demonstrate the use of the in-situ measurements for assessing the accuracy of each map in the local conditions. The work presented is part of the FP7-EU project "Merging hydrological models and Earth observation data for reliable information on water - MyWater". ESA is acknowledged for providing the SPOT XS and Resourcesat-2 LISS-III data (C1P.9275). USGS provided the MODIS Terra data and geoland2 project the SPOT VEGETATION LAI.