Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/9/2004
Publication Date: 9/2/2004
Citation: Walthall, C., Dulaney, W., Anderson, M., Norman, J., Fang, H., Liang, S. 2004. A comparison of empirical and neural network approaches for estimating corn and soybean leaf area index from Landsat ETM+ imagery. Remote Sensing of the Environment. 92:465-474. Interpretive Summary: Plant foliage density expressed as leaf area index (LAI)is used in many ecological, meteorological, and agronomic models, and is a means of quantifying crop spatial variability for precision farming. LAI can be measured with multispectral remote sensing imagery using several methods. A problem with many methods is that LAI measurements made on the ground are needed to calibrate the methods. Recently, several methods that do not require calibration methods have been proposed. Empirical methods requiring calibration data (using a traditional spectral vegetation index, the NDVI and a new index that uses green wavelength reflectance, the GI) and two new methods that do not require calibration methods (a scaled NDVI, and a hybrid neural network procedure) were evaluated using two dates of Landsat ETM+ imagery of the Iowa watershed where the Soil Moisture Experiments 2002 (SMEX02) was conducted. Validation measurements of LAI were collected on the ground. The two methods requiring calibration measurements and the scaled NDVI method produced comparable results. The hybrid neural network performed well, however; did not account for as much variability as the other methods. A greater amount of atmospheric haze present in one of the ETM+ scenes is believed to have contributed to the poorer peformance for one of the dates. The results show that reasonable estimates of LAI are possible from remote sensing imagery using algorithms that do not require ground calibration measurements. Algorithms not requiring calibration measurements have the potential for being used by operational satellite systems for global monitoring of LAI.
Technical Abstract: Plant foliage density expressed as leaf area index (LAI) is used in many ecological, meteorological, and agronomic models, and is a means of quantifying crop spatial variability for precision farming. LAI retrieval from optical wavelength remote sensing using spectral vegetation indices (SVI) usually requires calibration values from the surface, or the use of within-scene image information without surface calibrations to invert radiative transfer models. An evaluation of methods was conducted using: 1) empirical methods employing the Normalized Difference Vegetation Index (NDVI), and a new SVI that uses green wavelength reflectance, 2) a scaled NDVI approach that uses no calibration measurements, and 3) a hybrid approach that uses a neural network (NN) and a radiative transfer model without site-specific calibration measurements. Landsat-7 ETM+ data for July 1 and July 8 from the Soil Moisture EXperiment 2002 (SMEX-02) field campaigns in the Walnut Creek watershed south of Ames, Iowa were used for the analysis. Sun photometer data collected from a site within the watershed was used to atmospherically correct the imagery to surface reflectance. LAI validation measurements of corn and soybeans were collected close to the date of the Landsat 7 overpasses. Comparable results were obtained with the empirical SVI methods and the scaled SVI method within each date. The hybrid method, although promising, did not account for as much of the variability as the SVI methods. Higher atmospheric optical depths for July 8 are believed to have resulted in overall poorer performance for this date. Use of SVIs employing green wavelengths, improved methods for definition of image minimum and maximum clusters to avoid water pixels and outliers used by the scaled NDVI method, and further development of a soil reflectance index used by the hybrid NN approach are warranted. The results demonstrate that reasonable LAI estimates are possible using optical remote sensing methods without in situ calibration measurements.