In their article “Modeling Time-dependent Randomness in Stochastic Dual Dynamic Programming“ [1], Nils Löhndorf and Alexander Shapiro, compare the effectiveness of two prominent approaches on how to represent inflow uncertainty in models of long-term hydropower planning. The planning problem is formulated as a multistage stochastic program and solved with QUASAR®’s dual dynamic programming solver.
Numerical results indicate that discretizing the stochastic inflow process to a scenario lattice yields lower system cost than sample average approximation when evaluating decisions out-of-sample. The article also demonstrates how dynamic risk measures can make the solution of a stochastic programming model robust against the model risk of using a mis-specified stochastic process.
Interested to see the model in action and learn more about its benefits? 
Interested to learn more about how to model and solve a stochastic hydropower optimization problem with QUASAR®? 
[1] Löhndorf N, Shapiro A. 2019. Modeling time-dependent randomness in stochastic dual dynamic programming. European Journal of Operational Research 273(2), 650-661.