ERSSTv3b: Global Frequency Analysis

Here we use the same frequency analysis tools used for HadSST2 to look at ERSSTv3b.

---

Fourier Analysis	annual	monthly
Global	160, 80, 53.3	160, 80, 53.3
Northern Hemisphere	160, 80, 53.3, 32	160, 80, 53.3, 32
Southern Hemisphere	160, 53	160, 53

I tweaked the function that produced the global mean. It now uses a “weighted cell” method rather than a “weighted latitude.” There are reasons I prefer the latter for global or hemispheric means, but this provides a very close match to the NCDC linear trend and will be more suitable for regional work. Using the area weighted rather than latitude weighted method, there is no significant difference between annual and monthly series in regards to the Fourier intensity tests.

getRasterLatMean <- function(r) {

rows = dim(r)[1]
cols = dim(r)[2]

wcell <- area(r,na.rm=F) # cells
wlat <- extract(wcell,seq(1,((rows-1)*cols+1),cols)) # lat bands

m <- mask(wcell,r)
r1 <- r*wcell
r2 <- extract(r1,ncell(r1))
return(sum(extract(r1,1:ncell(r1)),na.rm=T)/sum(extract(m,1:ncell(m)),na.rm=T))
}

---

ERSSTv3b – Global

---

ERSSTv3b – Northern Hemisphere

---

ERSSTv3b – Southern Hemisphere

---

Code is here: trend-ersst-global.R