PubAg

Main content area

Near-Ground Remote Sensing of Green Area Index on the Shortgrass Prairie

Author:
Przeszlowska, Agnieszka, Trlica, Milton J., Weltz, Mark A.
Source:
Rangeland ecology & management 2006 v.59 no.4 pp. 422
ISSN:
1551-5028
Subject:
prairies, grasses, remote sensing, leaf area index, vegetation cover, spectral analysis, multispectral imagery, digital images, data collection, simulation models, new methods, Colorado
Abstract:
Accurate and efficient leaf area measurements of shortgrass prairie vegetation are difficult to obtain. Few studies have considered the green area index (GAI) as an approximation of the total area of photosynthetically active tissue per unit of ground area. The main objective of this study was to evaluate several near-ground remote sensing methods as reliable and cost efficient measures of GAI on the shortgrass prairie. GAI measured with a standard leaf area meter was compared to 1) spectral vegetation indices calculated from multispectral radiometer data, 2) GAI obtained from laser point-frame measurements, and 3) green cover estimates derived from digital camera images. All methods were assessed for accuracy, time, and cost efficiency. Data were collected in 2001 at the Central Plains Experimental Range in northern Colorado. The standard leaf area meter method was neither time nor cost efficient in comparison with the other methods evaluated in this study. The cost of GAI measurement with the traditional leaf area meter method ($$225 per plot) was 20 times greater than GAI estimation with the multispectral radiometer ($$11 per plot). Comparison of GAI obtained with the standard leaf area meter method with red-band reflectance index values (0.63––0.69 μμm) obtained with a portable multispectral radiometer resulted in the best model predictions (R² == 0.76, Akaike's information corrected criterion [[AICC]] == 182.9) and the most cost efficient method for GAI estimation. Green cover estimates from digital image analysis resulted in a good correlation with the leaf area meter GAI (R² == 0.72, AICC == 178.1). However, classification accuracies of digital images were decreased by limited spectral separability between green vegetation, brown vegetation, and soil background. Further calibration and refinement of near-ground remote sensing techniques for vegetation might establish these methods as efficient ground-truth alternatives to satellite-based remote-sensing applications of rangelands such as the shortgrass prairie.
Agid:
1158
Handle:
10113/1158