Improved Weather Generator Algorithm for Multisite Simulation of Precipitation and Temperature

Abstract

The KnnCAD Version 4 weather generator algorithm for nonparametric, multisite simulations of temperature and precipitation data is presented. The K-nearest neighbour weather generator essentially reshuffles the historical data, with replacement. In KnnCAD Version 4, a block resampling scheme is introduced to preserve the temporal correlation structure in temperature data. Perturbation of the reshuffled variable data is also added to enhance the generation of extreme values. A case study of the Upper Thames River Basin in Ontario, Canada is performed and the model is shown to simulate effectively the historical characteristics at the site. The KnnCAD Version 4 approach offers a major advantage over parametric and semiparametric weather generators as it can be applied to multiple sites for simulation of temperatures and precipitation amounts without making assumptions regarding the spatial correlations and the probability distributions for each variable.

Uncertainty in Precipitation Projection under Changing Climate Conditions: A Regional Case Study

“…The total drainage area of this coastal watershed is approximately 1856 Km 2 ( Figure 2). The river basin is both snow and rain-fed, however mountain snowpack will likely decreases due to higher temperatures as a result of climate change [21]. As a result, there will be a shift towards the river being predominantly rain-fed, causing stream flow to be lower during the spring and summer months and higher in the fall and winter.…”

“…A non-parametric multisite weather generator named KnnCAD V4 [21] based on K-nearest neighbors (K-NN) is used in this study. The KnnCAD V4 is the updated version of KnnCAD V3 [4] …”

“…β is estimated based on maximum likelihood estimation (MLE). For generation of extreme precipitation outside of the observed range, a perturbation technique is used following King et al (2015) [21].…”

“…Six Downscaling methods applied in this study are as follows: 1) bias corrected spatial disaggregation (BCSD) [11] [15], 2) bias correction constructed analogues with quantile mapping reordering (BCCAQ) [16], 3) delta change method coupled with a non-parametric K-nearest neighbor weather generator [17], 4) delta change method coupled with maximum entropy based weather generator [18], 5) non-parametric statistical downscaling model based on the kernel regression [19], and 6) beta regression based statistical downscaling model [20]. BCSD and BCAAQ were successfully applied across Canada in the past, however these methods cannot explicitly capture changes in daily extremes [16] where other four downscaling methods can capture changes in daily extremes and can produce extreme values outside of the historical boundaries [18]- [21]. The above mentioned six downscaling methods are used to quantify the amount of uncertainty arising from different types of statistical downscaling methods and compare it with other sources of uncertainties.…”

Improving Stochastic Modelling of Daily Rainfall Using the ENSO Index: Model Development and Application in Chile

“…This model was implemented as WGEN (Weather GENerator) by Richardson and Wright (1984) [1], which used a simple Markov Chain for precipitation occurrence, a gamma distribution for simulation of rainfall amounts, and an autoregressive model for the remaining variables. A number of subsequent WGs, such as WXGEN [8], CLIGEN [9,10], LARS-WG [11][12][13], ClimGen [14], WeaGETS [15,16], Met and Roll [17], MOFRBC [18,19], WeatherMan [20], MarkSim [21], AAFC-WG [22,23], WM2 [24], KnnCAD [25][26][27], and the WG used by the UK Met Office (UKCP09) [28,29], all share the basic principles of stochastic simulation presented in WGEN. These WGs are station-scale generators, with time scales that range from daily (or even hourly in the case of rainfall) to annual, daily resolution being the most common.…”