Runs of No Hurricanes

James B. Elsner & Thomas H. Jagger

In order to quantify the run of no hurricanes a simple model consists of a Markov Chain in which the annual counts are coded as 1 or 0 depending on hurricanes or no hurricanes.

Read in the data.

US = read.table("US.txt", header = TRUE)
US$W = as.integer(US$MUS > 0)
w = US$W

w0 = c(NA, w)
w1 = c(w, NA)
w01 = cbind(w0, w1)[2:length(w), ]

t0 = table(w01[, 1], w01[, 2])
t1 = t0/rowSums(t0)
ww = rep(0, 100000)
tp = t1[, 2]
for(i in 2:100000){
  ww[i] <- rbinom(1, prob = tp[ww[i - 1] + 1], size = 1)
  }

tmp = rle(ww)
tmp2 = tmp$lengths[tmp$values == 0]
csy = sum(tmp2 >= 9)/length(ww) * 100
cy = sum(tmp2[tmp2 >= 9])/length(ww) * 100

The first states that there is a 0.34% that any year is the start of a sequence. The second states that there is a 3.54% chance that any year is in a sequence of at least 9 years with no storms.