I am currently @ Stockholm Resilience Center, Stockholm University, Kräftriket 2B, SE-10691 Stockholm, Sweden

World has experienced the Warmest October-January and the Warmest Winter in northern hemisphere

NASA releases Combined Land-Surface Air and Sea-Surface Water Temperature Anomalies (Land-Ocean Temperature Index, LOTI) data based on the combined data from NOAA GHCN v3 (meteorological stations), ERSST v4 (ocean areas), and SCAR (Antarctic stations). They cover data since the formal measurement of earth surface temperature began in 1880, which are updated around the middle of every month.

The data are useful to answer several crucial questions regarding earth system dynamics. I was particularly interested in three questions, pushed by my personal interest in earth dynamics as well as inspired by the colleagues at Stockholm Resilience Centre:

  • What is the temporal variability and trend in the monthly temperature anomaly?
  • Do we observe statistically significant change points in the monthly temperature trend?
  • When are the maximum (positive) anomalies observed?

I ran the following simple time series analysis in R using the monthly LOTI data generated by GHCN-v3 and ERSST v4, and covers temporal period of 1880-01/2016. The anomaly is measured in 0.01 degrees Celsius around the base period 1951-1980 (see here for the data details).

Get and process monthly temperature anomaly data

> library(reshape)
> library(cpm)
> library(ggplot2)

> anom.data <- read.csv("http://data.giss.nasa.gov/gistemp/tabledata_v3/GLB.Ts+dSST.csv", sep=",", header=T)

> head(anom.data)

  Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec J.D D.N  DJF MAM JJA SON
1 1880 -29 -20 -18 -27 -14 -28 -23  -7 -16 -15 -18 -20 -20 *** **** -20 -19 -16
2 1881  -9 -13   1  -3  -4 -28  -5  -2  -8 -19 -26 -15 -11 -11  -14  -2 -12 -18
3 1882  10   9   2 -20 -17 -25 -10   4   0 -22 -20 -24  -9  -9    1 -12 -10 -14
4 1883 -33 -42 -17 -24 -25 -11  -7 -12 -18 -11 -19 -17 -20 -20  -33 -22 -10 -16
5 1884 -17 -11 -33 -35 -31 -37 -33 -25 -22 -21 -29 -28 -27 -26  -15 -33 -32 -24
6 1885 -64 -29 -23 -44 -41 -50 -28 -27 -18 -19 -22  -5 -31 -33  -40 -36 -35 -20

> anom.data <- anom.data[,1:(ncol(anom.data)-6)]

> anom.data <- melt(anom.data, id="Year")

> anom.data$value[anom.data$value=="****"] <- NA

> anom.data$value <- as.numeric(anom.data$value)

> tail(anom.data)

     Year variable value
1639 2011      Dec    52
1640 2012      Dec    51
1641 2013      Dec    66
1642 2014      Dec    78
1643 2015      Dec   111
1644 2016      Dec    NA

Petit-Mann-Whitney test to identify statistically significant change points

> change.point <- list()
> month <- list()
> for(i in anom.data$variable){
  change.point[[i]] <- processStream(anom.data$value[which(
    anom.data$variable==i&!is.na(anom.data$value))],
  cpmType="Mann-Whitney")$detectionTimes+1879
  month[[i]] <- rep(i, length(change.point[[i]]))
  }

> change.point <- data.frame(variable=unlist(month), points=unlist(change.point))

> head(change.point)

     variable points
Jan1      Jan   1940
Jan2      Jan   1945
Jan3      Jan   1984
Jan4      Jan   2006
Feb1      Feb   1941
Feb2      Feb   1987

Identify time points when maximum anomalies are observed

> max.val <- sapply(unique(anom.data$variable),
  function(x) max(anom.data$value[anom.data$variable==x], na.rm=T))

> max.dat <- data.frame(variable=unique(anom.data$variable), max.val)

> head(max.dat)
  variable max.val
1      Jan     113
2      Feb      88
3      Mar      92
4      Apr      87
5      May      85
6      Jun      77

Plot results

> ggplot(data = anom.data, aes(x = Year, y = value)) +
  labs(x = "Year", y = "Anomaly") + geom_point(color="firebrick", size = 0.3) +
  geom_line(color="forestgreen", size = 0.3) +
  geom_hline(aes(yintercept=0), linetype = 2, size=0.5) +
  geom_hline(aes(yintercept=max.val), data=max.dat, linetype = 3, size=0.5, color="darkmagenta") +
  geom_vline(aes(xintercept=points), data=change.point, color="#FF9999") +
  stat_smooth(method = "lm", size = 0.5) +
  facet_wrap(~variable, scales="free", ncol=2)

What do we observe?

We observe frequent changes in the variability (denoted by brick-colored dots and connected green lines) and an increasing trend (denoted by the blue straight line) in the monthly temperature anomalies. This indicates that the earth temperature has been steadily increasing since the measurement started and will likely continue to increase in the following years. The change-point analysis depicts a statistically significant change point (red vertical lines) around 1950 for each month, which has been claimed to be the onset of Anthropocene led by the extensive human perturbation of nature and thus led to anthropogenic global warming. Another change point is observed around 2000 for each month, likely indicating another exponential increase in global warming trend. This can be supported by the maximum anomalies (denoted by the dotted purple horizontal line) concentrated around and mostly after 2000. Importantly, the temperature anomalies observed for October-December 2015 and January 2016 were the highest since the measurement began and thus indicating the warmest for these months and the warmest winter for northern hemisphere in the history.

In a nutshell, global warming is not only evident but also alarming, currently stepping into its very extreme!

Leave a Comment

Your comments and discussion will enrich my blog. You can comment by logging in to your social network or disqus sites (options are shown in the login drop-cap or when you click on the comment box). You can also comment as a guest by clicking on the "Name" box under "sign up with disqus" and ticking "I'd rather post as a guest" (your e-mail will be kept confidential).

comments powered by Disqus