Remote sensing has been used in support of many agricultural applications such as Soil studies, Land cover and soil mapping, and has some important advantages where decision-makers need detailed information in order to formulate well-designed land use plans.
Salinity has been identified as one of three main barriers to crop production in many parts of the world. In Thailand, for example, studies have revealed that salinity affects soil across some 2.58 million hectares or 17 per cent of the nation's land mass.^1,2

YOUSEF ALDAKHEEL EVALUATES TWO METHODS OF DETERMINING SOIL SALINITY USINGLANDSAT™ IMAGERY

Remote sensing has been used in support of many agricultural applications such as soil studies, land cover and soil mapping, and has some important advantages where decision-makers need detailed information in order to formulate well-designed land use plans.
Salinity has been identified as one of three main barriers to crop production in many parts of the world. In Thailand, for example, studies have revealed that salinity affects soil across some 2.58 million hectares or 17 per cent of the nation's land mass.^1,2

The Middle East in general and Al-Hassa in particular is no stranger to this problem. Al-Hassa, located in eastern Saudi Arabia (see keymap below), is one of the world's largest oases with its focal point being the town of Al-Hofuf. The area lies about 70 km from the Gulf coast at elevations of 130 to 160 meters above sea level.
The area is characterised by an Arid climate with an evaporation rate that exceeds precipitation by a sizeable amount. Temperatures range up to 45^oC in summer (June to August) and down to 2^oC in winter (December to February). Rainfall is scarce and variable, ranging from 3.5 mm to 146.4 mm per annum and falling mainly during the winter months.

Materials and methods
To test the usefulness of Landsat imagery in determining soil salinity, two Landsat TM5 Satellite images of Al-Hassa acquired in June 1993 and November 1998 were employed.
The following Salinity Indices (SI) and NDSI (Normalized Differential Salinity Index) as proposed by Tripathi et. al.³ were applied:(1) SI = (TMBand1 X TMBand3)^1/2
(2) NDSI = (TMBand3 - TMBand4)/(TMBand3 + TMBand4)

These yielded relatively good results in re-classifying the salt-affected land. NDSI is just the reverse of the NDVI index for vegetation.
The integration of thematic layers was achieved using the logical approach defined by Mongkolsawat and Thirangoondone.⁴ This involved identifying areas that were subject to specific conditions and classifying soils as extremely or moderately saline and then further categorising the latter into those that were potentially or slightly saline. The interpretation of salinity in respect of crop production was based on previously published salt tolerance guidelines.⁵

Results and discussion
Salinity Indices were applied on the basis of the known spectral response of salt-affected soils. Here, it is worth noting that the spectral response in terms of the Digital Number (DN) of salt-affected soils is relatively higher than other categories in TM band1 and TM band3. Fig.2 depicts the soil salinity maps derived from the salinity index (SI). This index shows the variation across the territorial extent of the image and the derived mapping reveals an increase in the salt-affected area from 92.7 km² in 1993 to 260.1km² in 1998.
The NDSI images (Fig.3) also show an increase in the extent of salt-affected soils across the study area from 43.0 km² in 1993 to 47.7 km² in 1998.
While these results based on the same imagery show a discrepancy, the SI-calculated area was relatively close to the statisticallyestimated area of Saline Soil, i.e., 213 km².⁶

Top: Fig 2: Salinity Index(SI) images of Al-Hassa Oasis 1988 (left) and 1993 (right). Purple = saline land Green = vegetation, Yellow = Sand Bottom: Fig 3: Normalized Differential Salinity Index(NDSI) images of Al-Hassa Oasis 1988(left) and 1993 (right). Purple = saline land, Green = vegetation, Yellow = sand

Because TM band 1 is useful for soil/vegetation discrimination⁷ this gave SI superiority over NDSI, which is just a reverse of the Normalized Differential Vegetation Index (NDVI). In another study using NDVI image analysis6, the estimated salt-affected area was 89.4 km² and 131.4 km² for 1993 and 1998, respectively. The variability of these figures could, of course, be attributed to categorisation of the scene phenomena of the satellite image.
In conclusion, this study emphasises that SI and NDSI may be used in place of individual or multi bands to enhance the presence of saline patches as they suppress vegetation. However, more research is needed to validate this over other conventional methods.

Acknowledgment: The author acknowledges the Space Research Institute, and GDRGP at King Abdulaziz City for Science and Technology (KACST) for providing satellite imagery and financially supporting the study. Thanks also to King Faisal University for partial funding. Particular thanks are extended to Dr. Masoud Abdel Atti, and Dr. Adel Hussain, of the Water Studies Center at KFU for their technical assistance.

Yousef Y. Aldakheel is with the Water Studies Center at King Faisal University in Hofuf, Saudi Arabia, and can be contacted by email at:

This article is based on a paper delivered by the author at the 1st International Remote Sensing Conference, held in Kuwait in 2005

REFERENCES
1] Mitsuchi, M., Wichaidit, P. and Jeungnijnirund, S.(1986) Outline on Soils of the Northeast Plateau, Thailand, TheirCharacteristics and Constraints, Technical Paper No. I, Agric. Dev. Res. Center in Northeast Thailand, Khon Kaen, pp:76.
[2] Arunin, S.(1992) Strategies for Utilizing Salt-affected Lands in Thailand, Proc. of the int. Symp. on Strategies for UtilizingSalt- affected Lands, Bangkok, Thailand, Feb. 17-25, 1992, pp:259-268.
[3] Tripathi, N.K., Rai, B.K., and Dwivedi, P.,1997. Spatial modeling of soil alkalinity in GIS environment using IRS data.18 thAsian conference on remote sensing, Kualalampur, pp.A.8.1-A.8.6.
[4] Mongkolsawat, C.and Thirangoon, P.(1990) A Practical Application of Remote Sensing and GIS for Soil SalinityPotential Mapping in Korat Basin, Northeast Thailand, Tech. Rep. Series No.8, Remote Sensing, Soil and Water Managt. inNortheast Thailand, Khonkaen.
[5] Mass, E. V. (1984) Salt Tolerance of Plants. In: the Handbook of Plant Science in Agriculture. B.R. Christie (ed). CRCPress, Boca, Raton, Florida.
[6] Aldakheel, Y., A.M. Elprince, and Al-Hussaini, A.I. 2005. Mapping of Salt-Affected Soils of irrigated lands in arid regionsusing remote sensing and GIS. 2nd International Conference on Recent Advances in Space Technologies (RAST 2005). June09-11, 2005. Istanbul, TURKEY.
[7] Lillesand, T. M. and R. W. Kiefer. 1994. Remote Sensing and Image Interpretation, John Wiley & Sons, New York.