*3.3. Evaluation Framework*

The simulated results are examined based on performance indicators such as water abundance, groundwater dependency, field IE, irrigation return, and hydrologic connectivity. These measures are described below.


The policy performance is analyzed to measure water sustainability while investigating water economics. To understand the tradeoffs between the selected scenarios from both economical and hydrological perspectives, the above-described measures were employed. The percentage deviation is given by Equation (1):

$$
\Delta y\_{ij} = \frac{\sum y\_{ijt} - \sum y\_{ojt}}{\sum y\_{ojt}} \times 100 \tag{1}
$$

The percentage deviation is given by Equation (1), where *yij* represents the value of measure *j* with scenario *i*. Note that *i* = 0 indicates the base case simulation whereas *i* = 1 represents irrigation efficiency. The values are then summed for *t* (time) during the periods 2017–2050 and 2051–2099. Policy notations are listed in Table 3. For each of these measures, we calculate their relative deviation from base run values. For each comparison, the climate scenario remains unchanged. For example, we do not compare a UKMO policy scenario (Ui) with a GFDL base scenario. These simulations provide three alternative future possibilities as benchmarks that the policy scenarios could be compared with.

**Table 3.** Notation of scenarios.


The <sup>Δ</sup>*yij*'s value of hydrology performance indicators are plotted against percentage change in agriculture income on two-dimensional quadrant grids. On the presented quadrants, the *x*-axis always represents agricultural income, while the *y*-axis represents one of the hydrologic measures (IE, water abundance, irrigation return, and hydrologic connectivity). The title of each row in the figure indicates which hydrologic measures should be on the *y*-axis for the graphs in that column. Each column of the graphs is for a specific climate scenario. The first column indicates the results for GFDL, the second for UKMO, and the third for NCAR. Each point on the grid represents the discrepancy between a policy simulation run and a corresponding base run (no policy applied). Therefore, for each row of graphs, the origin of a grid represents an average steady-state, which is equivalent to a comparison of a base case and itself. Because there is no discrepancy between a simulation run and itself, such comparisons yield *x* = 0 and *y* = 0. The analysis is broken into two periods being the years 2017–2050 and 2051–2099. The impact of policy implementation could be prolonged and yield unexpected consequences. The comparison between the two periods can illustrate the expected and unexpected performance.

A quadrant analysis is applied to interpret the hydrology performance indicators and agricultural income. Each quadrant represents a potential outcome: beneficial, unacceptable, unsustainable agriculture development, and unsustainable hydrology. The upper right quadrant (beneficial) represents policies that are beneficial both economically and hydrologically, which would be the most favorable state. Thus, results that are located in the lower-left quadrant (unacceptable) are neither economically beneficial nor sustainable, thus unacceptable. The upper left quadrant (unsustainable agriculture development) represents results with a positive effect on water system performance but a negative influence on agricultural income. The lower right quadrant (unsustainable hydrology) indicates results that would be beneficial economically but would need additional water managemen<sup>t</sup> policies to offset associated negative hydrologic impacts. The following sections detail each grid presented in Figure 3 and demonstrate different performance measures of irrigated agriculture and how they deviate from the base runs due to the policy applications.
