Soil salinization refers to the excessive accumulation of soluble salts in soil to a degree that adversely influences vegetation and environmental health. Soil salinization poses an existential threat to ecosystem functioning, socioeconomic structure and food security. The excess accumulation of salt in soil is a global problem and is one of the main land-degrading threats influencing soil fertility and biodiversity. High salinity in the root zone severely impedes plant growth resulting in reduced crop productivity. Moreover, salt stress in salt-affected soils may cause a drastic reduction of plant transpiration influencing soil water budget and hydrologic cycle. Soil salinity also imposes nutritious imbalances in plants that may induce significant alterations in Earth’s organisms and ecosystems. Salt storms from saline soils, desertification, modification of soil organic carbon cycle and worsening of economic opportunity which may lead to human migration are other detrimental consequences of soil salinization. All this created an urgent need to understand the processes affecting soil salinization.

We employ a broad array of cutting-edge experimental and computational tools including artificial intelligence, pore- and continuum-scale numerical simulation, high resolution thermal imaging, and customized laboratory experiments to investigate the parameters and mechanisms controlling soil salinization at different scales (ranging from a pore with the size of a few microns to the global scale). One of our key objectives is to develop predictive models capable of describing soil salinization on a global scale under different climate scenarios. We employ a comprehensive set of climatic, topographic, soil, agroeconomic and remote sensing data to develop models trained by machine learning algorithms to predict past, present and future dynamics of soil salinization on a global scale (typical examples of our results are presented below). Additionally, we are interested in establishing the link between the physical mechanisms identified using small‐scale experimental and numerical analyses and the responses observed at the field-scale. Such efforts will contribute toward addressing the long-standing upscaling issue present in soil physics and hydrology community.

 
Figure caption. Using machine learning algorithms, we are developing quantitative tools capable of predicting soil salinity at the global-scale. This figure presents our typical results showing the average of annual predictions for soil surface salinity (represented by soil electrical conductivity ECe and exchangeable sodium percentage ESP) between 1980 and 2018. (After Hassani et al. (2020), Proc. Nat. Acad. Sci., doi.org/10.1073/pnas.2013771117).

Figure caption. Variations in the soil cell-level salinity/sodicity and country-level area of the salt-affected soils (p < 0.05). a and d: Cell-level variations in ECe and ESP between 1980 and 2018, respectively. Soil cell is any ~1 × 1 km stretch of the soil. b and c: Variations in the total area of soils with salinity of ECe ≥ 4 dS m-1 since 1980, at the country level. e and f: Variations in the total area of soils with sodicity of ESP ≥ 6% since 1980, at the country level. Countries are sorted based on the mean annual area of soils with an ECe ≥ 4 dS m-1 or ESP ≥ 6% between 1980 and 2018, largest to smallest (After Hassani et al. (2020), Proc. Nat. Acad. Sci., doi.org/10.1073/pnas.2013771117).