In consequence of the rapid economic development in China, the cotton production in the Xinjiang Autonomous Region has risen sharply in recent decades. The farming industry and the natural vegetation are thus in conflict over the very limited water resources.

The study area for the water allocation model MIKE HYDRO begins in Alar and ends on Taitema Lake (see Fig. 1). MIKE HYDRO is a deterministic, semi-distributive and physically based model system that can be used for the simulation of water consumption, actual evapotranspiration, irrigation management, crop yields and plant growth. For large-scale modelling, the Tarim River has been divided into four main sections, which were marked by discharge gauges (see Fig. 3): Alar Xinqiman (A), Xinqiman-Yingbaza (B), Yingbaza-Qiala (C), and Qiala-Taitema Lake (D). In each of these sub-regions, the results are aggregated for water consumption and biomass production and passed to the DSS.

After calibration of MIKE HYDRO various water-saving scenarios were simulated (Yu et al., 2015). Important results are shown in Fig. 4 and Table 1. From Fig. 4 it is obvious that the plant available water, which can be determined from usable field capacity and effective root length for most agricultural crops must not be filled up to 70 %. For cotton a deficit irrigation up to about 40 % TAW is acceptable without having severe yield losses. Therefore, this type of irrigation should be promoted in Xinjiang. A second way to save water for agriculture is the irrigation technology “drip irrigation under plastic mulch” (DIUM). Table 1 shows a water saving potential of up to 40 %, mainly due to prevention of (unproductive) soil evaporation by covering the soil with plastic wrap (“mulch”).

The results of MIKE HYDRO – supplemented by scenarios of water losses along irrigation canals – were finally passed as simplified regression relationships to the DSS. In this way it could be ensured that the impact of land management scenarios is based on a solid model and data base. The loss of accuracy by the transfer of the model results to the DSS are negligible. If necessary, additional MIKE HYDRO scenario calculations can supplement the database of the DSS.

Since the effect of the respective scenarios has a direct impact on the groundwater situation along the Tarim River, the internal calculation of groundwater levels is indispensable (Keilholz et al., 2015). The DSS should be usable for subsequent users without a license, so that the software can be easily distributed and freely accessed to both public authorities and for educational institutions.

One common open-source code for simulating groundwater movement is the model MODFLOW. Due to its modular structure and the grid-based calculation, it was used for calculating the groundwater distribution. The resulting groundwater levels of MODFLOW due to artificial flooding of the Lower Tarim River and other management practices are coupled with the DSS with a spatial resolution of 500 m × 500 m (see Fig. 5).

Based on these two models and the contribution of the other work blocks within the project, the DSS could be designed and implemented.

The DSS is an indicator based tool that enables stakeholders and decision-makers to evaluate the consequences of their actions. Among selected scenarios of future climate and socio-economic development, the impact of planned management measures is determined by quantitative and semi-quantitative methods based on ESS indicators (Table 2). All indicators were conducted by SuMaRiO workshops (DSS workshops, stakeholder workshops) and has been classified as relevant. The following Tables 3–5 show examples of the required input data (i.e. management data) which formed the core for the evaluation of the management alternatives.

The impact on the ESS indicators and the available discharge along the Tarim River for each scenario year is calculated with the help of multivariate functions, lookup tables or fuzzy rules. The output is presented in tables and graphs. Based on these results and the objectives assigned to the ESS indicators by the user, a degree of target achievement between 0 and 1 is calculated, indicating the relative target fulfilment by each measure for each ESS indicator. Additionally, the DSS displays for each year to what extent there was an improvement/deterioration of the utility value compared to the first planning year. Based on this differentiated outputs the trade-offs between the ESS indicators can be analysed. Furthermore, it calculates the change in land use and soil salinization for each grid cell for the last year of the simulated time period (see Fig. 6).

The DSS with the coupled MODFLOW groundwater model will be delivered to the Chinese partners free of charge. In the coming 5-year plan (starting 2016) three pilot areas have been defined in Xinjiang. The DSS will be tested and further developed in these areas under the supervision of the Chinese Academy of Science (Xinjiang Institute of Ecology and Geography). Generally, the DSS is applicable for other regions, too, and is free available on request.