Search

In the Syrian Steppe we carried out a field experiment using repeated hyperspectral measurements throughout the 2001 growing season to investigate the separability of vegetation types based on their temporal-spectral signatures. We studied two different perennial shrubs and annual grasses showing differences in the length of their growing period. These differences cause sea- sonal variations in spectral plant reflectance, also giving hints to functional vegetation characteristics such as palatability, soil stabilization or certain hydro- meteorological characteristics. By submitting biophysical and spectral data to statistical analyses we identified the dry season period as the most suitable to discriminate between the studied plants or their functional characteristics respectively. NDVI time series using MODIS or SPOT VEGETATION NDVI bands perform nearly as well as an optimum narrow banded index. For a balanced assessment of the biomass from different functional vegetation groups, our study recommends a narrow banded (9 nm width) vegetation index that uses the 630 nm band together with the 755 nm band. Its particular advantage, compared with indices calculated from MODIS and SPOT VEGETATION bands, is the better correlation with biomass from annual grasses. Conclusions from the field experiment were tested for their transferability to remote sensing conditions. Using the noise-filtered shapes of NDVI cycles as a primary classifier, we identify the distribution and the fractional cover of species with an extended growing period. Pixels with a dominant cover of species with an extended growing period were classified by correlating a reference NDVI cycle with all image pixels, where the calculated correlation coefficient is the measure for shape similarity, and the corresponding slope value the measure for vegetation cover relative to the reference. [ABSTRACT FROM AUTHOR]

This study demonstrates a unique matrix approach to determine within-field variability in wheat yields using fine spatial resolution 4m IKONOS data. The matrix approach involves solving a system of simultaneous equations based on IKONOS data and post-harvest yields available at entire field scale.This approach was compared with a regression-based modelling approach involving field-sensor measured yields and the corresponding IKONOS measured indices and wavebands. The IKONOS data explained 74-78%variability in wheat yield. This is a significant result since the finer spatial resolution leads to capturing greater spatial variability and detail in landscape relative to coarser spatial resolution data. A pixel-by-pixel mapping of wheat yield variability highlights the fine spatial detail provided by IKONOS data for precision farming applications. [ABSTRACT FROM AUTHOR]