ArticleCI

Obtaining confidence intervals and regions

Assuming you have obtained a converged "fit", here's how you can get the CIs:

    lat <- fit$most.prob.track[, 2]
    std <- sqrt(fit$var.most.prob.track[, 4])
    low <- lat - 2 * std
    hig <- lat + 2 * std

    lon <- fit$most.prob.track[, 1]
    std <- sqrt(fit$var.most.prob.track[, 1])
    low <- lon - 2 * std
    hig <- lon + 2 * std

To export the confidence regions into a shapefile, use the CI2shp in Analyzepsat:

### Assuming you have the input "track" for ukfsst

fmat <- data.frame(track[,c(3,2,1)],
                   fit$var.most.prob.track,fit$most.prob.track,0,0)
names(fmat) <- c("Year", "Month", "Day", "V11", "V12", "V21", 
                       "V22", "Lon_E", "Lat_N", "max_depth", "SST")
CI2shp(fmat, fname = "testshp", level = 0.95, npoints = 100, proj4string = CRS("+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs "))