Resource information
Inefficient irrigation and the excessive use of water on agricultural land in the Aral Sea Basin over several decades have led to saline soils. The main objective of this paper is to identify the environmental predictors to model the spatial distribution of soil salinity in a highly irrigated landscape. Soil salinity at farm scale was measured in the topsoil (Total Dissolved Solids, TDS) and down to a depth of 1.5m by electromagnetic conductivity meter (CMv) over a regular grid covering an area of approximately 15km² in Khorezm Province, Uzbekistan. Six nested samplings within selected grids were conducted to reveal short-distance variation. Apart from widely-used terrain indices and those acquired from remote sensing, data on distance to drainage channels and long-term average groundwater observations were used to account for local parameters possibly influencing soil salinity. Topsoil salinity (TDS) was seen to be highly variable even at short distances (40m) compared to average bulk soil salinity (CMv). CMv readings were better correlated with factors obtained from remote sensing and distance to drains than TDS. This might be attributable to the fact that topsoil salts are dynamic in nature, and land management practices (e.g. leaching, cultivation, and irrigation) might have contributed considerably to spatial variation. The CMv shows the average amount of salt within a larger soil volume and to greater depth and is less affected by land management than topsoil salinity, which is reflected in the TDS. Most terrain indices showed a low correlation with topsoil and bulk salinity. There was a strong indication that the effects of water management are dominant and tend to outweigh the effects of environmental factors. The very low R² for relationship of TDS with environmental factors is evidence that taking TDS samples close to the soil surface is not a good way to assess salinity trends in irrigated land. These findings have important implications for salinity survey methods on flat irrigated terrain: CMv seems to be a more reliable predictor than environmental proxy factors, even if the latter are easier to determine.
