One of the essential requirements in soil salinity management and reclamation is access to high accuracy salinity maps. This study was conducted to evaluate the accuracy of soil salinity mapping using remote sensing and geostatistical methods in a part of the alluvial lands of the Tigris River within the Al-Suwayrah farms in Iraq, located 30 km south of Baghdad. For this purpose, 100 soil samples were collected using a regular grid with 1,000 m spacing from a depth of 0–20 cm across an area of 10,000 hectares, and the electrical conductivity (EC) of the saturated paste extract was measured. Spectral bands and indices of ASTER and Landsat satellite images were prepared in three spectral modes: digital number (DN), radiance, and reflectance. Subsequently, using multiple linear regression, equations were developed between the derived spectral indices in each spectral mode and soil salinity, and a soil salinity map was also generated using the kriging method. The results of comparing the radiance, reflectance, and DN spectral modes for salinity modeling indicated that the Landsat satellite in reflectance mode achieved the highest accuracy (MBE = −1.24, RMSE = 4.58) among the remote sensing methods. Comparison of the remote sensing and geostatistical approaches showed that the geostatistical method, due to the use of measured field data, exhibited higher accuracy (MBE = −0.06, RMSE = 3.53) than the remote sensing methods; however, the maps produced by remote sensing also demonstrated acceptable accuracy, and the salinity map derived from this method closely corresponded to the salinity distribution map obtained from geostatistics. Considering the lower cost and time requirements, as well as the capability for temporal repetition, remote sensing methods can be effectively used to produce reconnaissance level soil salinity maps for the alluvial plain of the Tigris River with acceptable accuracy.
Introduction
Soil salinity is a major environmental challenge in arid and semi-arid regions, as it negatively affects agricultural productivity and can cause irreversible soil degradation. Accurate assessment of soil salinity is therefore essential for sustainable land management. The Tigris alluvial plain in Iraq an agriculturally important region is highly susceptible to salinization due to its arid climate and intensive irrigation practices. Providing precise information on soil properties, particularly salinity, is crucial for managing land resources and improving crop production.
Traditional soil mapping and sampling methods are time-consuming, costly, and spatially limited. In recent years, remote sensing techniques have emerged as effective tools for large-scale soil salinity assessment. Spectral indices derived from satellite imagery enable the detection of subtle variations in soil characteristics such as moisture, salinity, organic matter, and texture. Geostatistical methods, particularly kriging, also offer powerful capabilities for interpolating field-based soil salinity measurements. Although many studies worldwide have explored the application of remote sensing and geostatistical approaches for soil salinity mapping, each region possesses unique environmental and soil conditions that influence the accuracy of these techniques. The Tigris alluvial plain presents specific challenges due to its soil properties, vegetation cover, and hydrological conditions. Therefore, evaluating the performance of these techniques in this region is essential for identifying the most accurate method for soil salinity assessment and for supporting effective management of soil degradation.
Material and Methods
The study was conducted in the alluvial plain of the Tigris River, within the Al-Suweera (Alsouyreh) region, approximately 30 km south of Baghdad, Iraq. Soil samples were collected using a systematic grid sampling scheme with a spacing of 1000 m, covering an area of 10,000 ha at a depth of 0–20 cm. Laboratory analyses included electrical conductivity (EC), pH, and sodium adsorption ratio (SAR).
Satellite imagery from ASTER and Landsat was processed using three spectral modes Digital Number (DN), radiance, and reflectance to generate soil salinity maps. The kriging interpolation technique was applied to the field-measured salinity data to produce geostatistical salinity maps.
Accuracy assessment of the resulting maps was performed using Mean Bias Error (MBE) and Root Mean Square Error (RMSE), which are widely used metrics for evaluating differences between observed and estimated values.
Results and Discussion
The results indicated that the radiance mode of Landsat imagery and the reflectance mode of ASTER imagery provided the highest accuracy among remote sensing–based salinity estimates. However, the geostatistical approach (kriging) outperformed all remote sensing methods, producing the most accurate salinity maps. Validation results showed negative MBE values for all remote sensing and geostatistical models, with the lowest MBE (–0.06) obtained using kriging. Kriging also yielded the lowest RMSE value (3.53), confirming its superior performance.
Landsat reflectance demonstrated better accuracy than all spectral modes derived from ASTER imagery. The superior accuracy of kriging can be attributed to its direct reliance on dense field measurements, which is particularly advantageous for relatively small study areas with adequate sampling density.
Both the remote sensing and geostatistical maps showed similar spatial patterns of salinity distribution and exhibited strong agreement, highlighting the potential of remote sensing for large-scale soil salinity mapping. The spatial patterns revealed lower salinity levels in the southern part of the study area and higher levels in the northern part, nearer the Tigris River. This aligns with actual field conditions, where the southern region has benefited from surface drainage systems since 2000, reducing salinity. Conversely, the northern region relies on long-term irrigation with saline river water (ECw = 4 dS/m) and lacks an effective drainage system, resulting in persistent salinity accumulation.
Conclusion
This study demonstrates the effectiveness of remote sensing and geostatistical methods for soil salinity mapping in arid environments such as the Tigris alluvial plain in Iraq. Although geostatistical methods particularly kriging were found to provide the highest accuracy, remote sensing offers valuable spatial information and is well suited for broader scale assessment. Maps produced by both approaches provide essential data for monitoring soil degradation and guiding land management strategies. The results also underscore the importance of proper drainage infrastructure in preventing salinity buildup, especially in areas where irrigation depends on saline river water.
Acknowledgements
The authors gratefully acknowledge the financial support of Ferdowsi University of Mashhad (Project Code: 47644), which made this research possible.