First of, you would not be able to infer actual data on marginal measures of social benefit or social cost from this prompt alone, as they do not indicate any such information (if they are constant or variable). You'd instead have to conjecture the explicit effects of the saving of water. In doing so, we can, more safely than not, assume that conserving water is a positive externality of this subsidy, and this marginal benefit is ..clears throat.. constant.

The subsidy being imposed, we notice that Supply (MPC) curve moves to the right; that is, for every same level of price, producers now supply more to the market.
Furthermore, we notice that the Demand (MPB) curve remains constant, as no measure that takes place should change the needs of the consumers for agricultural products.
On this we will add a constant Marginal Social Benefit; that is the positive externality of rationing on water usage. Mechanisms at work, classical economic theory suggests, would imply that the consumption of these subsidized products, saves more water; so people naturally want to purchase more of them. This being the case, the Marginal Social Benefit curve is established further to the right than the Demand curve.

One final point of equilibrium is that where $Q=Q_2$, and producers receive price $P_0$ as part of their subsidy, while costumers pay an actual price of $P_2$.
The cost of this subsidy is $Q_2 * (P_0-P_2)$.