Soil salinity is a huge problem negatively affecting physiological and metabolic processes in plant life, ultimately
diminishing growth and yield. An area with more than 70,000 ha sugarcane farming and its by-products are the major
agricultural activities in the Khuzestan province, in the southwest of Iran. Therefore, mapping and identification of soil
salinity is the most important issue to improve management of large scale crop production in this area. Besides labour
intensive fieldwork, remote sensing is the most suitable technique to assess soil salinity for large areas. This study was
carried out to investigate the capability of Hyperion spaceborne hyperspecteral data for mapping the salinity stress in the
sugarcane fields and determine the best method to classify soil salinity into 3 classes (low, moderate and high salinity).
For this purpose the capability of different classification methods like support Vector Machine (SVM), Spectral Angle
Mapper (SAM), Minimum Distance (MD) and Maximum Likelihood (ML) in conjunction with different band
combinations (all bands, principle component analysis (PCA), Vegetation Indices) as an input data was performed.
Results indicated that best method for classification is SVM classifier when we use all bands or PCA(1-5) as an input
data for classification with an overall accuracy and kappa coefficient of 78.7% and 0.68 respectively. Therefore, salinity
stress can be classified in agricultural fields using Hyperion satellite imagery with good accuracy and salinity map can be
very useful for management of agricultural activity and increase the crop production.
In this paper, the effect of dimensionality reduction of hyperspectral data on 10 subpixel target detectors is investigated.
The genetic algorithm (GA) and wavelet feature extraction methods are used for dimensionality reduction as they
maintain physically meaningful bands and physical structure of the spectra, respectively. In the former case, the
wrapper method is used to improve subpixel target detectors' results in terms of the area under the curve (AUC) of the
receiver operating characteristic (ROC) curve. Meanwhile, in the latter case, the AUC is used as a criterion to choose the
optimum level of wavelet decomposition. Experimental results obtained from a real-world hyperspectral data and a
challenging synthetic dataset approved that band selection with the wrapper method is more efficient than using target
detection methods without dimensionality reduction, especially in the presence of difficult targets at subpixel level.