Thank you for your kind offer. It is always a lot easier to understand how someone has done an analysis when they share their materials and answer questions about what (and why) optional choices were made in their procedures. I will drop you an e-mail shortly.

I’ll be happy to send you our data and programs (in RATS) if you e-mail me. We used the Jones temperature series. OTOH I don’t think the Nature paper was our best work, just the first one we published. There are links to all the later papers at my website.

Steve: Ever consider using source control?

Steve: I’m trying to make these scripts turnkey, but they are done quickly. Sometimes I edit things in the console and there may be a loose end on my script as posted. LEt me know and I can usually figure it out pretty quickly.

What do you mean by “any room for CO2″?

Model 3. in Box includes trend and CO2 . Model 1. doesn’t have CO2, but is there really anything left for CO2 to explain ?

Some Model 1 realizations, assuming LS-solution is correct:

Another consequence of this tropospheric behaviour is that an eventual increase of water content in the lower levels of the atmosphere (due to CO2 IR absorption, for istance) would enhance the ITCZ activity whereas the region of dry descending air would expand, leading to a less water content in the medium-high troposphere outside the ITCZ region and to a widening and/or thickening of the stratocumulus clouds layer, not to mention the Iris Effect.

> ###CALCULATE TRENDS
> Trend= array(NA,dim=c( 5,18))
> for (k in 1:5) {
+ year=c(time(zonal[[k]]))
+ temp=(year>=1979)
+ for (j in 1:18) {
+ fm= try(lm(zonal[[k]][temp,j]~year[temp]))
+ if (!(class(fm)==”try-error”)) Trend[k,j]=fm$coef[2]
+ }
+ }
Error in lm.fit(x, y, offset = offset, singular.ok = singular.ok, …) :
0 (non-NA) cases
Error in lm.fit(x, y, offset = offset, singular.ok = singular.ok, …) :
0 (non-NA) cases
>
> ##PLOT CONTOUR MAP
> #library(fields)
> hpa=c(1000,740,466,220,75)
> range(Trend,na.rm=T) #-0.08826997 0.05186060
[1] -0.07743676 0.05186060
> breaks0=c(-0.095,seq(-.055,.055,.01),.095);n=length(breaks0)
> filled.contour(x=lat,y=1000-hpa,z=t(Trend),levels=breaks0,col=tim.colors(n-1),main=”RSS/CRU Trends”,
+ plot.axes = { axis(1, seq(-60, 60, by = 20))
+ axis(2, at=1000-hpa,labels=as.character(hpa)) } )
Error in filled.contour(x = lat, y = 1000 – hpa, z = t(Trend), levels = breaks0, :
dimension mismatch

I fixed the “d:” paths and the extra space in v <- open.ncdf(“temp.dat”). Everything that precedes the “CALCULATE TRENDS” section completes without error.

>> rank(I-P,10^-10)

ans =

119

>> rank(I-P_,10^-10)

ans =

119

Sorry about that

Maybe its a “bug” in Matlab.

Rounding problem ? rank.m provides just an estimate of the true rank..

I=eye(129);

P=Xb*pinv(Xb);

P_=Xb*inv(Xb’*Xb)*Xb’;

rank(I-P)

%ans =
%
% 120
rank(I-P_)

%ans =
%
% 119

% but

max(max(abs(P-P_)))

%ans =
%
% 2.0178e-014

(I marked the midlevels as dry descent but actually Hilo is experiencing some moister midlatitude air intrusion at the moment. It’s not a perfect example.)

The inversion, which is common, is distinct. Below the inversion lies moist near-ocean air. Above the inversion is the dry, radiationally-cooling and sinking air originally lifted high into the atmosphere in thunderstorms thousands of miles to the west.

The equivalent potential energy (known as theta e) increases suddenly below the inversion, indicating the high moisture content.

Regions like the air above Hilo, where radiational cooling is important and water content is very low, are where I’d expect to see a distinct effect from increased CO2, a fingerprint unlike any created by solar changes.