Assessment of remote sensing-based classification methods for change detection of salt-affected areas (Biskra area, Algeria)

Abstract

In the Wadi Biskra arid and semiarid areas, sustainable development is restricted by land degradation processes such as secondary salinization of soils. Being an important highquality date production region of Algeria, this area needs continuous monitoring of desertification indicators, hence highly exposed to climate-related risks. Given the limited access to field data, appropriate methods were assessed for the identification and change detection of salt-affected areas, involving image interpretation and automated classifications employing Landsat imagery, ancillary and multisource ground truth data. First, a visual photointerpretation study of the land cover and land use classes was undergone according to acknowledged methodologies. Second, two automated classification approaches were developed: a customized decision tree classification (DTC) and an unsupervised one applied to the principal components of Knepper ratios composite. Five indices were employed in the DTC construction, among which also is a salinity index. The diachronic analysis was undergone for the 1984 to 2015 images (including seasonal approach), being supported by the interpreted land cover/land use map for error estimation. Considering also biophysical and socioeconomic data, comprehensive results are discussed. One of the most important aspects that emerged was that the accelerated expansion of agricultural land in the last three decades has led and continues to contribute to a secondary salinization of soils

Classification methods for detecting and evaluating changes in desertification-related features in arid and semiarid environments

“…Specific red-green-blue (RGB) composites have been defined through the combination of various ratios, among which Knepper ratios. Upto-date literature show that Knepper ratios have been used only in geological remote sensing until now (Mia and Fujimitsu 2012;Langford 2015;Afrasinei et al 2017b). In this study, we propose its employment for salt and sand features extraction.…”

“…Out of a total of 11 indices employed in the decision tree classifier (Table 3), five new indices were constructed by us for this particular type of study, through broad spectral analysis, band transformation techniques, image statistics, and expert-knowledge (Rao et al 2006;Afrasinei et al 2017b;Matthew 2012;Srimani and Prasad 2012).…”

Influence Characteristics of Laser Transmission Amplitude Fluctuation Based on Turbulent Medium

“…The noise generated by atmosphere [8] [9] [10] can reduce the reception performance of system and can disturb the accuracy of acquired signal. The reception perturbation of signal [11] [12] is more obvious in remote sensing application, thus it should be studied deeply using related theory.…”

Land use/land cover change detection combining automatic processing and visual interpretation

“…Different approaches can be used to elaborate and update existing LUC maps. Visual interpretation -often computer aided -has been widely used to elaborate LUC cartography (Afrasinei et al, 2017;Asner, Keller, Pereira, & Zweede, 2002;Gurgel, Farias, & Oliveira, 2017;Sunar, 1998), including cartography of large areas, such as Europe (Feranec, Hazeu, Christensen, & Jaffrain, 2007), Africa (Disperati & Virdis, 2015), China (Zhang et al, 2014), and India (Roy et al, 2015). Visual interpretation enables map producers to include many classification criteria, such as texture, shape, pattern, size, and proximity between objects, and the interpreter's knowledge about the study area (Lu et al, 2004;NRSC/ISRO, 2010).…”