Probabilistic analysis and scenario analysis both require simulation of a range of possible futures. Ensembles of future hydrologic scenarios can be based on the statistical characteristics of the historical record. Reconstructed paleohydrology can be used to extend the record. (PDSI data can be used to derive sequences of hydrologic states that can be used along with historic flow values.) Climate change projections can generate futures that are not based on the past. The RiverSMART tools for hydrologic simulation can be used for any of these techniques. The library consists of R code and a graphical user interface through which the user specifies the reference sequence, the sampling parameters, the paleo sequence, the conditioning method (Homogeneous Markov or Non-homogeneous Markov), increase or decrease in mean and other parameters.
KNN Resampling
Ensembles of stochastic hydrologic sequences (traces) are typically based on the statistical characteristics of the historical record. The RiverSMART tool set includes a Hydrology Simulator that generates ensembles of traces based on a nonparametric K-nearest neighbor (KNN) technique (Prairie et al., 2006). This simple approach is akin to simulating from the conditional PDF without actually fitting it; thus it has the ability to capture any nonlinear or non-Gaussian features present in the data. This resampling can also be used in the paleo record and climate change projections.
Nonparametric Simulation via Conditioning
This resampling technique of high-fidelity historic streamflow values can be “conditioned” on the richer variety of sequences found in the paleo record or in climate change projections. We implement the method by Prairie et al. (2008), in which paleo reconstructions are mined for their sequences of states (“wet” and “dry”) by modeling the state transitions using Markov Chain while the magnitude of streamflow is modeled from the observed flow data. This model can generate a rich variety of wet and dry spells. They can also be combined with climate change projections to generate flow scenarios that can capture the nonstationarity in the flow variability.
Temporal and Spatial Disaggregation
These techniques are combined with a multisite streamflow simulation method (Prairie et al 2006; Nowak et al., 2010) to generate monthly flow scenarios at the spatially distributed nodes in the basin.
References
Prairie, J., B. Rajagopalan, T. Fulp and E. Zagona, (2006), “Modified K-NN Model for Stochastic Streamflow Simulation,” Journal of Hydrologic Engineering, 11(4), 371-378, 2006.
Prairie, J., K. Nowak, B. Rajagopalan, U. Lall, and T. Fulp, (2008), “A Stochastic Nonparametric Approach for Streamflow Generation Combining Observational and Paleo Reconstructed Data,” Water Resources Research, 44.
Prairie, J., B. Rajagopalan, and U. Lall (2006), “A Stochastic Nonparametric Technique for Space-time Disaggregation of Streamflows,” Water Resources Research, 43.
Nowak, K., J. Prairie, B. Rajagopalan, and U. Lall (2010), “A Non-parametric Stochastic Approach for Multisite Disaggregation of Annual to Daily Streamflow,” Water Resources Research, 46.