Sabkha is an Arabic word for salt-flats found mainly in arid regions along the coastlines and inlands, within sand dunes. The sabkhas that form within the sand are relatively flat and very saline areas of sand or silt that forms just above the water-table where the sand is cemented together by evaporite salts from seasonal ponds. The UAE is home to some of the largest concentrations of both coastal and inland sabkhas. The coastal areas of Abu Dhabi includes several small shoals, islands, protected lagoons, channels and deltas, an inner zone of intertidal flats with algal mats and broad areas of supratidal sabkha salt flats. Sabkha surfaces can in certain situations be a geotechnical hazard due to its high salinity and with adverse effects on concrete, asphalt, steel and other structures, in addition to their sporadic heaves and collapses. As the UAE continue to develop major urban infrastructure identifying the location of such habitats is of utmost importance in proper urban planning processes. Identifying sabkha surfaces from remotely sensed data is a challenging process. Traditional remote sensing mapping techniques of multispectral data, usually fail to properly identify sabkha pixels or provide lower rates of mapping accuracy for sabkha surfaces. The primary objective of this research is to assess the feasibility of using multispectral salinity indices for properly identifying sabkha surfaces from remotely sensed data. Salinity indices have previously been developed for mapping saline soils within agricultural areas. These indices include the visible and near infrared parts of the spectral radiation in various mathematical combinations. This research applies a series of multispectral salinity indices, based on green, red and NIR parts of spectra, for identifying sabkha pixels on a DubaiSat2 multispectral image of western UAE. The preliminary results suggest that salinity indices based on green and red radiation, perform better in identifying sabkha pixels compared to a combination with NIR, as suggested in former literature. The index SI=√(R2 + G2) comprising of red and green bands of the spectra, in this study, was able to identify sabkha pixels with very high accuracy. The comparison between actual sabkha pixels and index predicted pixels reveal an accuracy of more than 90%.