berngamma: The Bernoulli-Gamma distribution

In qmap: Statistical Transformations for Post-Processing Climate Model Output

Description

Density, distribution function, quantile function and random generation for the Bernoulli-Gamma distribution with parameters prob, shape, and scale.

Usage

dberngamma(x, prob, scale, shape)
pberngamma(q, prob, scale, shape)
qberngamma(p, prob, scale, shape)
rberngamma(n, prob, scale, shape)

Arguments

x,q	vector of quantiles.
p	vector of probabilities.
prob	probability of non-zero event.
n	number of random samples.
scale,shape	shape and scale parameters of the gamma distribution.

Details

Mixture of the Bernoulli and the Gamma distribution. The Bernoulli distribution is used to model the occurrence of zero values with the probability of 1-prob. Non-zero values follow the Gamma distribution with shape and scale parameters.

The probability density function (PDF) is defined as:

g(x) = if(x > 0) pi*gamma(x) else 1-pi

where gamma(x) is the probability density function of the gamma distribution and π is probability of a non-zero event.

The cumulative distribution function (CDF) is defined as:

G(x) = if(x > 0) 1- pi + pi*Gamma(x) else 1-pi

where Gamma(x) is the cumulative distribution function of the gamma distribution.

The quantile function (inverse of the CDF) is defined as

G^{-1}(x) = if(p > 1-p) Gamma^{-1}((p-1+pi)/pi) else 0

where Gamma^{-1}(p) is the inverse CDF of the gamma distribution and p is a probability.

Value

dberngamma gives the density (pdf), pberngamma gives the distribution function (cdf), qberngamma gives the quantile function (inverse cdf), and rberngamma generates random deviates.

Note

The implementation is largely based on the bgamma family in the CaDENCE-package (Cannon, 2012) that was only available as test version at time of implementation (Mar. 2012). The CaDENCE-package is available at http://www.eos.ubc.ca/~acannon/CaDENCE/.

For further details and meteorological application of Bernoulli-Gamma distributions (also referred to as 'Mixed Gamma' distribution) see Burger et al. 2012, Cannon 2008, Li et al. 2010, Mooley 1973, Piani et al. 2010, Thom 1968, Sloughter et al. 2007.

Author(s)

Lukas Gudmundsson

References

Burger, G.; Murdock, T. Q.; Werner, A. T.; Sobie, S. R. & Cannon, A. J. Downscaling extremes - an intercomparison of multiple statistical methods for present climate. Journal of Climate, American Meteorological Society, early online release, 2012, doi:10.1175/JCLI-D-11-00408.1.

Cannon, A. J. Probabilistic Multisite Precipitation Downscaling by an Expanded Bernoulli-Gamma Density Network. Journal of Hydrometeorology, American Meteorological Society, 2008, 9, 1284-1300, doi:10.1175/2008JHM960.1.

Cannon, A. J. Neural networks for probabilistic environmental prediction: Conditional Density Estimation Network Creation and Evaluation (CaDENCE) in R. Computers & Geosciences, 2012, 41, 126 - 135, doi:10.1016/j.cageo.2011.08.023.

Li, H.; Sheffield, J. & Wood, E. F. Bias correction of monthly precipitation and temperature fields from Intergovernmental Panel on Climate Change AR4 models using equidistant quantile matching. J. Geophys. Res., AGU, 2010, 115, D10101, doi:10.1029/2009JD012882.

Mooley, D. A. Gamma Distribution Probability Model for Asian Summer Monsoon Monthly Rainfall. Monthly Weather Review, 1973, 101, 160-176, doi:10.1175/1520-0493(1973)101<0160:GDPMFA>2.3.CO;2.

Piani, C.; Haerter, J. & Coppola, E. Statistical bias correction for daily precipitation in regional climate models over Europe. Theoretical and Applied Climatology, 2010, 99, 187-192, doi:10.1007/s00704-009-0134-9.

Thom, H. C. S. Approximate convolution of the gamma and mixed gamma distributions. Monthly Weather Review, 1968, 96, 883-886, doi:10.1175/1520-0493(1968)096<0883:ACOTGA>2.0.CO;2.

Sloughter, J. M. L.; Raftery, A. E.; Gneiting, T. & Fraley, C. Probabilistic Quantitative Precipitation Forecasting Using Bayesian Model Averaging. Monthly Weather Review, 2007, 135, 3209-3220, doi:10.1175/MWR3441.1.

See Also

GammaDist

Examples

data(obsprecip)

(ts <- startberngamma(obsprecip[,1]))
hist(obsprecip[,1],freq=FALSE) 
lines(seq(0,20),dberngamma(0:20,
                           prob=ts$prob,
                           scale=ts$scale,
                           shape=ts$shape),
      col="red")

pp <- seq(0.01,0.99,by=0.01)
qq <-quantile(obsprecip[,1],probs=pp) 

plot(qq,pp)
lines(qberngamma(pp,
                 prob=ts$prob,
                 scale=ts$scale,
                 shape=ts$shape),
      pp,col="red")

plot(qq,pp)
lines(qq,
      pberngamma(qq,
                 prob=ts$prob,
                 scale=ts$scale,
                 shape=ts$shape),
      col="red")

hist(rberngamma(1000,
                prob=ts$prob,
                scale=ts$scale,
                shape=ts$shape),freq=FALSE)

Example output

Loading required package: fitdistrplus
Loading required package: MASS
Loading required package: survival
$prob
[1] 0.4758602

$scale
[1] 9.328407

$shape
[1] 0.5020363

qmap documentation built on May 1, 2019, 7:31 p.m.