reservoir (version 1.1.5)

dirtyreps: Quick and dirty stochastic generation of seasonal streamflow replicates for a single site.

Description

Generates seasonal time series using either the kNN Bootstrap (non-parametric) or a numerically-fitted PARMA(1,1) (parametric) model. For the parametric model, the function automatically transforms the seasonal sub-series to normal and deseasonalizes prior to model fitting.

Usage

dirtyreps(Q, reps, years, k, d, adjust, parameters, method = "kNNboot")

Arguments

Q

time series object with seasonal resolution (e.g., frequency = 2, 3, 4, 6 or 12 for monthly data).

reps

integer. The number of replicates to be generated. The default is 100.

years

integer. The length of each replicate in years. The default is equal to the number of complete years given in Q.

k

integer. The k parameter of the kNN Bootstrap (i.e., number of nearest neighbors). If left blank k = n ^ 0.5., where n is the number of years in the input data.

d

integer. The d parameter of the kNN Bootstrap (i.e., number of previous time periods to inform the model). If left blank d = 1.

adjust

logical. If TRUE (the default) the final output time series X will be coerced for 0

parameters

logical. If TRUE the output will be given as a list including the replicate samples and relevant model parameters (k and d for kNNboot and phi, theta and standard deviation of residuals for PARMA). The default is FALSE.

method

character string giving the method used to generate the data. Defaults to "kNNboot" - the k Nearest Neighbour Bootstrap. See references for detail on the two methods available.

Value

Returns a multi time series object containing synthetic streamflow replicates.

References

kNN Bootstrap method: Lall, U. and Sharma, A., 1996. A nearest neighbor bootstrap for resampling hydrologic time series. Water Resources Research, 32(3), pp.679-693.

PARMA method: Salas, J.D. and Fernandez, B., 1993. Models for data generation in hydrology: univariate techniques. In Stochastic Hydrology and its Use in Water Resources Systems Simulation and Optimization (pp. 47-73). Springer Netherlands.

Examples

Q <- resX$Q_Mm3
replicates <- dirtyreps(Q, reps = 3)
mean(replicates); mean(Q)
sd(replicates); sd(Q)
plot(replicates)