Guest Post by Willis Eschenbach
Well, since I was on a roll with my last post Rainergy, I thought I’d look further at the Copernicus global rainfall dataset. I started by looking at the change in global rainfall over time.
Figure 1. Global monthly rainfall, 1979 – 2022.
Well, that’s interesting. Overall, despite endless hype about increasing floods, there’s no significant trend in rainfall. The main feature is the dropoff in rain from the 2016 peak. Being curious about that drop, I thought I might look at the hemispheres separately to see where it’s happening. Here’s that data:
Figure 2. CEEMD smooths, northern and southern hemisphere monthly rainfall
Zowie, sez I … do you see what I see?
The two hemispheres are basically mirror images! When one is wetter, the other is dryer, and vice versa. And as to why that would be, my only guess is that it’s from the very rainy Intertropical Convergence Zone (ITCZ) wandering above and below the Equator. Other than that, I fear I have no answer except for the quote below.
Figure 3. Quote from Richard Feynman, one of the most outstanding physicists of our time
Seeing the inverse relationship between the northern and the southern hemispheres made me wonder how well the models managed to hindcast the rainfall over the same period, and whether the models found the same mirroring of the NH and SH. For example, in the real world the northern hemisphere (blue line in Figure 2 above) is wetter than the southern (red line) … do the models find this difference?
So I went to the marvelous KNMI website and got the CMIP6 model average data. And when I graphed it up, my eyebrows went up to my hairline and I busted out laughing …
Figure 4. CEEMD smooths of modeled hemispheric rainfall, CMIP6 model average. This model average is created by first averaging all of the model runs of each model, and then averaging the model averages. This is to prevent overweighting the models with lots of runs.
I am totally gobsmacked. I don’t know what I expected, but it sure wasn’t this … in total contradiction to the real-world observations, in modelworld the southern hemisphere is wetter than the north, the northern hemisphere is getting much wetter over time, total annual rainfall is about 75mm (3 inches) or about 8% too large, and there’s no mirroring …
But wait, as they say on TV, there’s more! Here’s the CMIP6 SSP245 model average global rainfall from 1850 to 2100. It is hindcasting using real data up to 2014, and forecasting after that.
Figure 5. Modeled global rainfall, CMIP6 model average, SSP245 scenario. The graphs are taken directly from the KNMI website. Upper panel is full data, lower panel shows residual after removing seasonal variations. This CMIP6 model average is created by first averaging all of the model runs of each model, and then averaging the model averages. This is to prevent overweighting the models with lots of runs.
Seriously? Does that look real to anyone?
And there’s another oddity. Recall from my post Rainergy that evaporating water to create rainfall cools the surface. The modeled rainfall shown above claims that by 2100, the rainfall will have increased from the 20th Century average by ~60 mm. The evaporation necessary to produce this increased rainfall would cool the surface by an additional 4.8 W/m2 … which per IPCC calculations would offset the theoretical increase in forcing resulting from CO2 increasing from 400 ppmv to 980 ppmv.
Right … that’s totally believable …
These are the Tinkertoy™ models that our noble climate cognoscenti are using to predict the climate in the year 2100? We’re abandoning the world’s reliable energy sources based on these ludicrous models??? …
Madness. Tragic madness.
I fear that’s all for today. Although I’m sure that there’s more to be learned from the Copernicus rainfall data, at this moment I’m laughing and crying too hard to do any more mathematical analysis.
My best to all,
w.
Yeah, yeah, you heard it before, but I gotta say it again: When you comment, please quote the exact words you are discussing. I can defend my words. I cannot defend your interpretation of my words.
And if you wish to show that I’m wrong, here are complete instructions on how to show Willis is wrong.
I visually see a slightly increasing trend up to the La Nina events which lead to a big downturn at the end. I’m guessing the short time period and La Nina are the big reason for this result. I suspect longer term we will see some increase.
If hockey and therefore its stick had never been invented, would this have prevented global warming/climate change?
So much depends on hockey sticks in climate ‘science’.
“Overall, despite endless hype about increasing floods, there’s no significant trend in rainfall.”
For floods, the relevant number is the total precipitation over land.
Does the rainfall data exclude snowfall?
Oh, I see the same was just asked by dk_…
Thanks Willis. Interesting find.
It appears that there is a time offset on the potentially correlating peaks and valleys. I was going to look into it, but don’t know how to work with the resulting .nc file type from Copernicus. Particularly, on or about the years of 1979, 1984, 1988, 1996, 2008, 2016.
Thoughts on the potential offsets?
Good afternoon Willis (or for you, still morning). Another in a long list of excellent posts. I have a favor to ask – on your “Where’s the Emergency” post, it appears that comments have been closed, so I am making my request here. I refer back to that post with some frequency and you say there “Updated on a totally irregular basis as new information becomes available. Last update May 9, 2024]. When you say that, could you add what was updated if it’s not too much trouble? If it’s a pain, no worries – it’s one of my go to posts when attempting to refer people to real world data.
Thanks for all the work you do here.
Can do, Barnes, I probably should just start an “Updated” list at the top and add the new ones as they come along.
w.
The observed monthly rainfall data looks as if it correlates with solar activity.
Hi, I’m playing around with the .nc datasets from the Copernicus site but I’m not sure how to interpret their data. E.g. many coordinates have 0 precipitation, I’m wondering if this is true or if the data was simply missing.
How did you deal with that data when you made your plots? Or would you perhaps have a link to the source code for your plots? It would really help me out!
The reality is that for much of the world in one or more months of the year there is zero precipitation. So that’s what I use.
As to the data, I first downloaded the 516 monthly .nc precipitation files, one for each month, Jan 1979 to Dec 2021. Then I run through the files, extract the rainfall data for that month, resample it to 180° latitude by 360° longitude. and store the results as a 180 x 360 x 516 array. Code like this.
(copath="copernicus rainfall/") # path to folder allfiles=paste0(copath,dir(copath)) head(allfiles) (flen=length(allfiles)) bigrain=array(NA,c(180,360,flen)) #make storage array i=1 for (i in 1:flen){ if (i %% 10==1) print(i) whichfile=allfiles[i] nc=open.nc(whichfile) bigrain[,,i]=resampleit(aperm(var.get.nc(nc,"precip"))[72:1,c(73:144,1:72)]) }“resampleit” is a function that resamples a matrix or an array to 180 rows by 360 columns.
resampleit=function(thearray,newx=360,newy=180){ require(raster) thedims=dim(thearray);thedims thedepth=ifelse(length(thedims)==2,1,thedims[3]);thedepth biglight=array(NA,c(180,360,thedepth)) i=1 startrow=(180-newy)/2+1;startrow endrow=startrow-1+newy;endrow for (i in 1:thedepth){ if (i %% 100==0) print(i) x <- raster(nrow=thedims[1],ncol=thedims[2],crs=NA) y <- raster(nrow=newy,ncol=newx,crs=NA) if (thedepth==1) values(x) = thearray else values(x) =thearray[,,i] finaly=raster::resample(x,y) biglight[startrow:endrow,,i]=as.matrix(finaly) } # detach(raster,unload = T) return(biglight) }From there I do whatever kind of analysis I wish—calculate trends, create averages, the possibilities are endless.
w.