The WEAP model was used to simulate the Limay–Neuquén–Negro system. The construction of the model had two components: (1) a supply component represented by the WEAP hydrological module to estimate the runoff as a function of precipitation and other hydrological processes that occur in the upper-river basin; and (2) a requirement component representing agricultural and urban water demands, as well as artificial reservoirs/hydropower plants located downstream and operated under system restrictions. The supply component uses a set of algorithms to represent rainfall–runoff processes with the objective of simulating the observed river flows at a satisfactory level of approximation. The requirement component is also modeled in WEAP and considers agricultural, industrial, urban, and other uses. Sub-basin delineation was made based on contributions to the main river flows and based on the data availability of hydro meteorological stations supporting an adequate calibration of the model. This resulted in nine sub-basins with a total area of 53,236 km2
). Each of the sub-basins was discretized into 400-meter elevation bands constituting the hydrological units—or catchments—in which the WEAP two-layer soil moisture method was applied [3
]. This method is based on empirical functions that describe the processes of snow accumulation/melt, infiltration, evapotranspiration, surface and subsurface runoff, deep percolation, and base flow runoff.
The simulation was performed on a monthly scale covering the period between January 2000 and December 2010. Input data provided by the AIC corresponded to precipitation, temperature, relative humidity, wind velocity (available for 27 meteorological stations), snow accumulation, and albedo. Vegetation coverage data were also available for evapotranspiration estimation purposes within the model (forest and meadow/steppe). The model included five hydropower dams located in the Limay River, of which two have seasonal regulation capacity (Piedra del Aguila and Chocón); the model also included the Cerros Colorados Complex, a facility in the Neuquén River that consists of two reservoirs and a hydroelectric power station. The model simulated the demands of urban centers, agricultural areas under irrigation, exploitation of conventional and non-conventional hydrocarbons, transfers, and hydroelectric generation up to the mouth of the Negro River as shown in Figure 3
Seasonal reservoirs were geometrically represented by a water level–volume curve, and model parameters were related to their capacity of energy generation. The simulation of the reservoir operation assumed the current norms of operation and orders of priority for water storage and hydropower generation. Non-seasonal reservoirs were simply represented as passing power plants with no storage capacity. The goodness of fit of the calibrated model for the 2000–2010 period was evaluated using statistical standards. The comparison of simulated and observed flows ensures the goodness of fit at different control points in the watershed (Table 3
Once the WEAP model application of the Comahue region was calibrated to accurately represent the basin, a series of scenarios were developed and implemented. These scenarios were identified by local stakeholders in a Problem Formulation activity using the XLRM framework.
Participants included the provincial Water Department of the Rio Negro Province (DPA), the Under Secretary of Water Resources of Neuquén province (SSRH), the AIC, research team members from the local university, and local project partners.
One of the key sources of uncertainty in the system was climate. Working with the Centro de Investigaciones del Mar y la Atmósfera (CIMA), a set of climate projections were developed to represent the likely climate impacts to the system from 2016 to 2050 in WEAP (Figure 4
). The selection of the climate model was based on climate similarities with the Comahue region. A second uncertainty was the potential expansion of the agricultural region and the introduction of higher value crops. This was a concern for water managers in charge of supplying irrigation water during the summer months. Table 4
contains a list of the considered uncertainties and the various projections developed for each of those uncertainties.
Strategies were identified as potential actions that decision-makers could take. Because of the increasing agricultural irrigation requirements, these included the reduction in water losses in the canal system, the incorporation of new technologies related to irrigation efficiency, and the development of irrigation canals. While agriculture was considered a large consumptive user, urban conservation strategies were encouraged because of the large losses in the urban water infrastructure system. Multi-purpose reservoir development was considered for increasing water storage, as well as for increasing the production of hydropower. Table 5
contains a list of the strategies evaluated in the study and a brief description each strategy.
The outcomes of the various management actions were evaluated through performance metrics. These metrics were utilized to assess the impacts of the uncertainties and evaluate the effectiveness of the proposed strategies based on decision-making management objectives. Table 6
contains a list of performance metrics and their corresponding management objectives.
All scenario ensemble results can be accessed via a link to an interactive Tableau Public visualization platform, included in the captions for Figures 6 and 7. Figure 5
shows the main impacts of the uncertainties under the current management, i.e., if no strategy is considered in the short-term. Figure 6
show those impacts in the long-term. The impacts of the recognized future uncertainties on the performance metrics are represented as the frequency of failure of each of the performance metrics based on the thresholds described in Table 6
. The frequency of failure is estimated by calculating the percentage of times that specific metric results are below (or above, depending on the metric) the established threshold. This failure is shown as a value in percentage terms and also by the color dimension in the graphic. The cut-off point between red and green was not set arbitrarily; it was set based on a discussion with stakeholders of what they would consider an undesired percentage value of failures. The visualization is structured to look at the long-term impacts by selecting the data corresponding to specific decades of the time horizon of the results (e.g., 2011–2020, 2021–2030, 2031–2040, and 2041–2050). In addition, users can select the month of the year to analyze impacts in specific times of the year (e.g., summer months vs. winter months).
Under the current management, environmental flows (Metrics 1–7) are below acceptable levels during the summer months (November, December, January, February, and March). These impacts are more evident in the Neuquén River reach at San Patricio del Chañar, Confluencia, Canal Margen Norte and Desembocadura (Metrics 2, 4, 5, and 7). Figure 5
shows the short-term impacts and Figure 6
the long terms impacts. Long-term impacts of climate change are more significant on the ecological flows, primarily in the scenarios that examine the potential expansion of agricultural areas and climate model ESM2. In the long term, all environmental flows are negatively impacted under ESM2 climate projection. Changes in cropping patterns (shown as High Profitability) show no difference from current trends in the impacts on ecological flows.
The ESM2 climate projections for the long-term show that during the irrigation season (summer), water demand for agricultural areas (performance metrics 8 through 11) is not met, specifically in areas that depend on the Neuquén River between Portezuelo Grande and El Chañar. While in the short-term the agricultural expansion seems feasible, the failures are greatly magnified in the long-term. However, there is little difference in the water demand satisfaction between the High Profitability crop scenario and the Current Trend scenario. For the other climate projections, GFDL and MIROC, negative impacts only show up for the potential expansion of agricultural areas for Anelo, Cinco Saltos and Los Barreales in the long-term.
Under the ESM2 climate projections for the long-term, the Mari Menuco and Cerros Colorados reservoirs fail to have the minimum levels required for normal operations. In the scenarios of agricultural expansion, the failures increase slightly. Cropping pattern changes do not affect reservoir storages. Short-term results show no failures for all scenarios except ESM2 climate projections.
For the Canal metrics, the failures of performance are based on reaching the total capacity of the canals because of the higher water demands. The capacity of irrigation canals in Canal Margen Norte Valle Medio, Canal Conesa and Canal Principal Valley Inferior reached their maximum capacity in all scenarios in the short- and long-term.
contains an image of the visualization dashboard showing the changes in the frequency of failure when various strategies are implemented. As expected, the strategy that contemplates the combination of all strategies is the one that provides the greater magnitude of impacts. The combination of all strategies includes changes that are more integrated and reduces the failures in all metrics, although in small proportions for some metrics and minimally for Canal Matriz and Canal Valle Inferior. A series of strategies improved the outcome of the region when they were integrated (not evaluated in isolation). Reducing agricultural water losses and implementing irrigation efficiency strategies seem to reduce the negative impacts on environmental flow metrics in the Neuquén River and some of the canal flows, primarily in the scenarios where agricultural production is expanded.
Urban water demand is not shown in the visualization dashboard because it had the highest priority and, therefore, it was always met. Further detail on urban water demands and coverage is needed to assess the extent to which the current system is, and will be, able to meet water demands for domestic use.