Geological mapping is the process of physically going to the field observation and recording the geological information from the rocks that protrudes from the surface of the earth. The information that usually the scientists attempt to find are the boundaries between different structures and rock types, such as fault-lines and evidence of the rocks undergoing deformation (Davis et al., 2011). Geologic mapping is a scientific field that aims to produce usable maps for various applications, such as quality assessment of ground waters and pollution hazards; land-use planning and land management; forecasting volcano, landslide, and earthquake; describing energy and mineral resources as well as their extraction costs; waste repository location; and general education (Compton, 1985; Soller, 2002). A powerful tool that can be implemented to improve the process of geological mapping is the technology of remote sensing (Varnes, 1974; Bernknopf, 1993; Pour and Hashim, 2015; Yang et al., 2018).

Remote sensing technology is useful for the explorations of minerals and geothermal energy, geological investigations, and assessment of geotechnical engineering and environmental geology. Remote sensing is also an essential tool for understanding the significant natural hazards pertinent to geology such as earthquakes, floods, avalanches, river channel migration and avulsion, liquefaction, landslides and debris flows, sinkholes, subsidence, tsunamis, and volcanoes (Bhan and Krishnanunni, 1983). The modern and applicable remote sensing tools include optical images, thermal data, LiDAR and digital elevation models (DEM), microwave and SAR data, hyperspectral remote sensing, and archived aerial photos and satellite images. Therefore, it is possible to perceive the Earth beyond our visual capability and transact the temporal and spatial limitations of earth observations (Prost, 2013).

Many image processing techniques have been presented in recent decades for purposes of geological mapping using remote sensing technology, such as band ratio (Inzana et al., 2003), correlation coefficient (Kühn et al., 2009), principal component analysis (Loughlin, 1991), optimum index factor (Fal et al., 2019), decorrelation stretch (Kenea, 1997), and log residual (Hook et al., 1992), etc. In this study, the applicability of different image processing techniques including the band ratio, decorrelation stretch, principal components analysis, and minimum noise fraction, as well as false color composition ASTER RGB:468 was investigated for discrimination of different geological units from ASTER Level1T VNIR+SWIR data.

The effectiveness of Terra Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) satellite data for lithological mapping and discrimination of geological units was demonstrated in many studies during the recent decades (Hewson et al., 2005; Pournamdari et al., 2014; Abdelouhed et al., 2022). The suitability of ASTER image for geological applications is mainly due to the spectral characteristics of the ASTER visible/near-infrared (VNIR), shortwave infrared (SWIR), and thermal infrared (TIR) bands, consequently, possibilities to perform different image processing techniques for mapping geological units (Rokni et al., 2011; Hewson et al., 2017; Rezaei et al., 2020).

At first, an evaluation on false color composition (RGB: 468) of ASTER image was implemented. With reference to the geology map of the study area (Figure 2), our inspection indicated that some of the geological units such as Basalt, Tuff, Conglomerate, and Dolomite can be discriminated from the ASTER RGB: 468 color composition, as shown in Figures 3.

Subsequently, the band ratio, decorrelation stretch, principal components analysis, and minimum noise fraction techniques were applied to find out their applicability in highlighting different geological units from ASTER imagery. Several band rations that have previously been verified for geological purposes were examined and consequently the ASTER band ratios (4/1, 4/5, 4/6) was found more suitable for detailed lithological mapping in the study area. As displayed in Figure 4, the ASTER band ratios (4/1, 4/5, 4/6) clearly discriminated Dolomite, Tuff, Sandstone, Conglomerate, and specially Basalt from the ASTER data. Besides, the outcome of decorrelation stretch demonstrated greatly discrimination of Limestone, Basalt, Tuff, Conglomerate, and Dolomite from ASTER image (Figure 5). This technique was able to highlight all the geological units in the study area except for Sandstone.

PCA technique also was applied on ASTER image (figure 6). Based on statistical analysis and also percentage of data variation, the results showed that the first PC involved the highest variance with positive loadings at each band. PC1 can extract information about lithological and mineralogical rock units through visual interpretations. The last two PCs include small variance with mostly noise. Therefore, the band combination PC1PC2PC3 that contains about 97.78 percent of total variance was selected. Our investigation revealed that the PCA technique is suitable to discriminate Tuff, Limestone, Basalt, Sandstone, Dolomite, and Conglomerate. However, some Basalt units that were adjacent to Sandstone were not differentiated by this technique.

MNF technique was implemented to highlight lithological units from ASTER imagery, as illustrated in Figure 7. The result show that MNF approach is suitable to highlight Basalt, Tuff, Dolomite, and some areas containing Conglomerate, while it is difficult to discriminate Sandstone and Limestone using this technique.

Finally, the BRs-PCs method was performed on ASTER imagery of the study area and the output image was evaluated to assess the applicability of this method to highlight geological units. In doing so, the PCA technique was applied on the selected band ratios in this study (4/1, 4/5, 4/6 as RGB), and the obtained PCs (BRs-PC1, BRs-PC2, BRs-PC3 as RGB) were analyzed. The result demonstrated the effectiveness of BRs-PCs method for geological mapping. As displayed in Figure 8, this approach provided an enhanced output compared to methods PCA and BR alone, and clearly discriminated the available geological units in the study area.

Overall, as presented in Table 2, the findings of this study indicate that discrimination of Conglomerate, Tuff, Dolomite, and Basalt units from ASTER image at the region of Alborz Mountain was easy, so that all the applied techniques were able to clearly extract these geological units. On the other hand, discrimination of Limestone and Sandstone was difficult and these geological units were not identified by some of the applied techniques. The ASTER band combination 468 that is well known for geological mapping, was also not able to discriminate Limestone and Sandstone.

The specified band ratios could not extract Limestone from ASTER image. In contrast, although PCA technique highlighted all the geological units in the study area, but some errors was appeared in extraction of Basalt using this technique. But BRs-PCs approach, because considered the advantages of both PCA and BR techniques, provided a superior output compared to PCA and BR methods alone, and also better result compared to other applied techniques in this study. The BRs-PCs approach successfully discriminated all the geological units from ASTER imagery.

The findings of this study were confirmed by the results of previous studies in different regions. In the study of Aali et al. (2022), suitability of the PCA and BR techniques was demonstrated for detection of the geological units such as Sandstone, Dolomite, Conglomerate, Limestone and Marl in the Chakchak region, Yazd province, Iran. Pournamdari et al. (2014) also proved applicability of these techniques for lithological mapping of Soghan ophiolitic complex in south of Iran. Aboelkhair and Watanabe (2011) sucsessgully performed the PCA and DS techniques for mapping Basalt flow in Madinah area, Saudi Arabia. Moreover, Sekandari et al. (2020) executed the PCA, MNF and BR techniques for the extraction of lithological units such as Dolomite, Limestone, Basalt, Sandstone, Conglomerate and Tuff in the central part of the Kashmar–Kerman tectonic zone, the Central Iranian Terrane (CIT). Similarly, in the study of Othman and Gloaguen (2017), the applicability of PCA, BR and classification techniques was verified for mapping the geological units such as Conglomerate, Limestone, Tuff and reddish green Shales in the Bardi-Zard area in north-east Iraq, a part of the Zagros Fold – Thrust Belt. As shown in the above studies, the principal component analysis and band ratio were the most widely used techniques for geological mapping using satellite data.

In this study, the applicability of several image processing techniques, including the band ratio (BR), decorrelation stretch (DS), principal components analysis (PCA), minimum noise fraction (MNF), and the ASTER false color composition RGB: 468, was evaluated for the extraction of geological units from ASTER satellite imagery in southern margin of Alborz Mountain in Iran. In addition, a method based on Principal Components of Band Ratios (BRs-PCs) was proposed for the discrimination of geological units from ASTER imagery. The results indicated that the ASTER RGB:468 and MNF technique were not able to clearly highlight Limestone and Basalt from ASTER image, while the DS technique successfully discriminated all the geological units except for Sandstone. In addition, the results revealed that the specified band ratios could not extract Limestone from ASTER image. In contrast, PCA technique highlighted all the geological units in the study area, however a failure was appeared in discrimination of Basalt using this technique. Nonetheless, the BRs-PCs approach, because considered the advantages of both PCA and BR techniques, provided a superior output compared to these techniques alone, and successfully discriminated all the geological units from ASTER imagery. The study concluded that the BRs-PCs approach may be useful for geological mapping along the whole Alborz Mountain with similar lithological and geomorphological conditions.