Item

Abstract

Normalized Difference Vegetation Indices (NDVIs) are typically determined using satellite or airborne remote sensing, or field portable spectrometers, which give an averaged signal on centimetre to metre scale plots. Biological soil crust (BSC) patches may have smaller sizes, and ecophysiological, hydrological as well as pedological processes may be heterogeneously distributed within this level of resolution. A ground-based NDVI imaging procedure using low-cost equipment (Olympus Camedia 5000z digital camera equipped with a Hoya R72 infrared filter) was developed in this study to fill this gap at the level of field research, where carrying costly and bulky equipment to remote locations is often the limiting factor for data collection. Method principle and field data are presented, and the field experiment was deepened comparing NDVI measurements and CO2 turnover of soil crust samples in the laboratory, backing the reliability of the approach. A commercially available colour rendition chart with known red (600–700 nm) and NIR (800–900 nm) reflectances was placed into each scene and used for calibration purposes on a per-image basis. Generation of NDVI images involved (i) determination of red and NIR reflectances from the pixel values of the red and NIR channels, respectively, and (ii) calculation and imaging of the NDVI, where NDVI values of −1 to +1 were mapped to grey values of 0 to 255. The correlation between NDVI values retrieved from these images and NDVI values determined using field spectrometry was close (r2 = 0.91), the 95% confidence interval amounted to 0.10 NDVI units. The pixel resolution was 0.8 mm in the field and 0.2 mm in the laboratory, but can still be improved significantly with closer distance to the crust or with higher camera resolution. NDVI values obtained using the new method were related to the net CO2 uptake of BSCs, where both slope and correlation coefficient of the respective regression function conformed with literature data. Geostatistical analysis revealed that both spatial variability of net CO2 uptake as well as size of individual hot spots of this parameter increased with crust development. The latter never exceeded 4 mm in the investigated crusts, which points to the necessity of high resolution imaging for linking remote sensing with ecophysiology. Perspectively, the new method could be used for field monitoring of both biological soil crusts and vascular vegetation.