fit_weights: Compute weighting between two models

In IceCast: Apply Statistical Post-Processing to Improve Sea Ice Predictions

Description

Compute weighting between two models based on accuracy in predicting a set of observations. Computation is via the Expectation-Maximization algorithm.

Usage

fit_weights(mod1, mod2, obs, prop_area, w_ini = 0.5, z_ini = 0.5,
  eps = 0.01)

Arguments

mod1	array with estimated sea ice probability from model 1. Dimensions are nuumber of training years x lon x lat.
mod2	array with estimated sea ice probability from model 2. Dimensions are nuumber of training years x lon x lat.
obs	array with observations of sea ice presence (1) and absence (0). Dimensions are nuumber of training years x lon x lat.
prop_area	matrix that gives the proportion of area in each grid box. Should sum to 1. Dimensions are lon x lat.
w_ini	initial value of all w, defaults to 0.5.
z_ini	initial value of all z, defaults to 0.5.
eps	tolerance for EM algorithm to reach convergence, defaults to 0.01.

Value

value between 0 and 1 giving the weight on the first model

Examples

## Not run: 
weight <- fit_weights(mod1 = clim_9_2005_2007, mod2 = ppe_9_2005_2007,
obs = obs_9_2005_2007, prop_area = prop_area)

## End(Not run)

IceCast documentation built on June 24, 2019, 9:03 a.m.