WGEN Weather Generation Model

|

Description

WGEN is a stochastic weather generator originally developed in the 1980s in Fortran at the US Department of Agriculture Agricultural Research Service (Richardson and Wright 1984). It uses monthly and annual statistics to generate daily time series of precipitation, minimum temperature, maximum temperature, and solar radiation.

Precipitation is modeled as a two-state Markov process. The Markov probabilities are probabilities of transition to a wet state for the current day based on the previous day's classification as wet or dry. On wet days, a gamma distribution is used to model the amount of precipitation. Markov probabilities and gamma distribution parameters are supplied on a monthly basis in the Inputs_Dashboard. Note that a wet day is defined as a day with at least 0.01 inches of rainfall.

Minimum and maximum temperature and solar radiation are each modeled by creating a seasonal cycle for the mean value and a seasonal cycle for the coefficient of variation. The following data are inputs to create the seasonal cycle:

  1. Mean value of the parameter for the year (for example, annual mean daily solar radiation),
  2. Amplitude of the mean value annually (e.g. the difference between average solar radiation in June and December),
  3. Mean value for the coefficient of variation (this would be the average of the monthly coefficients of variation), and
  4. Amplitude of the annual variation in the coefficient of variation (the difference between summer and winter coefficients of variation).

Given the means and amplitudes, the seasonal cycle of the mean value and the seasonal cycle of the coefficient of variation are created with cosine functions and a fixed phase shift (one phase shift for solar radiation and a longer phase shift for temperature since temperature lags solar radiation). The phase shifts were found to vary little from site to site, so they are treated as constants. For maximum temperature and solar radiation, the four statistics listed above are supplied for dry days and wet days separately, so that, for example, rainy days have less solar radiation. Matrix multiplication is used to generate daily residuals for minimum temperature, maximum temperature, and solar radiation that are cross-correlated and autocorrelated, so that, for example the minimum and maximum temperatures increase at the same time. Temperature and solar radiation parameters are also entered in the Inputs_Dashboard, which has page numbers of relevant user manual sections for guidance.

Download the WGEN Manual here.

 

Making Better Decisions In An Uncertain World