%0 Generic %1 sultanov2018remote %A Sultanov, Murodjon %A Conrad, Christopher %A Bauer, Christian %A Thiel, Michael %A Ibrakhimov, Mirzkayot %D 2018 %K conference poster Thiel LSFE %T Remote sensing of soil salinity levels in irrigated cotton production systems in the Aral Sea Basin %X Soil salinization is among the major constraints of sustainable agricultural production in the irrigated areas of the Aral Sea Basin (ASB). Existing field investigations on cotton production areas underlined that irrespectively of the salinity level, the salt moves downwards through irrigation events and re-enters upper soil layers when soils dry up. A good impression of the salinity level within a soil profile can thus be gathered after the last irrigation application. It is hypothesized that these salinity levels can be explained through multi-temporal optical remote sensing (RS) data that cover the entire cropping season. Time series of soil and vegetation indices calculated from Landsat imagery were used for indirect assessments of salinity levels in a cotton production area of 75.5 ha during 2008-2011. Accuracy of the RS-based maps was assessed against salinity maps obtained by dense points of EM-38 electromagnetic induction measurements (returns salinity in mS/cm). Several environmental variables such as topography, groundwater depth and salinity, were also included as predictors beside the spectral information. The application of random forest regression showed that RS allows for moderate precision of salinity prediction with coefficients of determination (r2) ranging between 0.44 and 0.58, while the inclusion of environmental variables did not improve the prediction quality. The estimated root mean square error (RMSE) ranged between 8.45 and 21.81. The variations of accuracy most likely depended on the different image acquisition dates among the years. The unexplained variance may result from the fact the indirect mapping of vegetation also shows effects of other growth-limiting factors, such as inappropriate timing of irrigation or reduced water supply. However, the results suggest that multi-temporal optical RS data can support the monitoring of soil salinity levels in the irrigated lowlands of the ASB and can hence be of practical use, e.g. for monitoring the success of maintenance operations in the drainage system.

@conference{sultanov2018remote, abstract = {Soil salinization is among the major constraints of sustainable agricultural production in the irrigated areas of the Aral Sea Basin (ASB). Existing field investigations on cotton production areas underlined that irrespectively of the salinity level, the salt moves downwards through irrigation events and re-enters upper soil layers when soils dry up. A good impression of the salinity level within a soil profile can thus be gathered after the last irrigation application. It is hypothesized that these salinity levels can be explained through multi-temporal optical remote sensing (RS) data that cover the entire cropping season. Time series of soil and vegetation indices calculated from Landsat imagery were used for indirect assessments of salinity levels in a cotton production area of 75.5 ha during 2008-2011. Accuracy of the RS-based maps was assessed against salinity maps obtained by dense points of EM-38 electromagnetic induction measurements (returns salinity in mS/cm). Several environmental variables such as topography, groundwater depth and salinity, were also included as predictors beside the spectral information. The application of random forest regression showed that RS allows for moderate precision of salinity prediction with coefficients of determination (r2) ranging between 0.44 and 0.58, while the inclusion of environmental variables did not improve the prediction quality. The estimated root mean square error (RMSE) ranged between 8.45 and 21.81. The variations of accuracy most likely depended on the different image acquisition dates among the years. The unexplained variance may result from the fact the indirect mapping of vegetation also shows effects of other growth-limiting factors, such as inappropriate timing of irrigation or reduced water supply. However, the results suggest that multi-temporal optical RS data can support the monitoring of soil salinity levels in the irrigated lowlands of the ASB and can hence be of practical use, e.g. for monitoring the success of maintenance operations in the drainage system.}, added-at = {2020-09-11T12:04:50.000+0200}, author = {Sultanov, Murodjon and Conrad, Christopher and Bauer, Christian and Thiel, Michael and Ibrakhimov, Mirzkayot}, biburl = {https://www.bibsonomy.org/bibtex/21fe52d0f30be028c0523c9390e7a7d34/earthobs_uniwue}, eventdate = {16.-19.04.2018}, eventtitle = {Between Europe and the Orient, Status symposium}, interhash = {16d4e9ccb43453922c5bda3aaf4e4e53}, intrahash = {1fe52d0f30be028c0523c9390e7a7d34}, keywords = {conference poster Thiel LSFE}, month = {April }, timestamp = {2020-11-18T22:08:33.000+0100}, title = {Remote sensing of soil salinity levels in irrigated cotton production systems in the Aral Sea Basin}, venue = {Almaty/Kasachstan}, year = 2018 }