cc: Phil Jones <p.jones@uea.ac.uk>
date: Wed, 24 Sep 2008 11:54:27 +0100
from: kate.willett@metoffice.gov.uk
subject: Re: Downward trend in relative humidity over land?
to: Adrian.Simmons@ecmwf.int

Over land I found non-significant and very small decadal trends, with
the Northern Hemisphere trend actually being slightly positive. (G =
-0.03, NH = 0.07, T = -0.10 and SH = -0.34 - SH data is very sparse and
likely of low quality). In contrast, the Marine data showed very
significant negative trends but I'm highly suspicious about the pre-1982
data which has a strong positive bias relative to the rest of the
timeseries.

The climatology period and possibly the way that anomalies are obtained
differ for HadCRUH (1974 to 2003) although I'm guessing this should have
little effect. The scale on my plots is large and so its perhaps
difficult to see any trends and I agree that the additional 3 years in
the ERA data bring make the negative trend clearer. I have had problems
with different trend fitting techniques giving more/less attention to
end points of the series and so giving very different trends. Ideally it
would be nice if ERA and HadCRUH were in good agreement but I think it
may be realistic that they are not given the variability in RH over land
and the likely very different ways that RH is derived for both.  

I like the idea of a rough sanity check to see if the q and RH changes
in ERA are consistent. For HadCRUH I did a very rough version of
percentage change in q for 1K increase in T using the idea that for
constant RH a ~7% increase in q would be expected. Results suggest
global land q increases are consistent with constant RH (~7%), NH land q
increases suggest a slight increase in RH (~8 %) and Tropics land q
increase suggest a decreasing RH (~5.5 %).

Not sure if all the above is helpful or just rambling. I think
presenting the plots at Boulder is a good idea though.

Kate  

Dr Kate Willett   Climate Research Scientist
Met Office Hadley Centre, FitzRoy Road, Exeter, EX1 3PB
tel. +44 1392 884288 fax +44 1392 885681 (mark for my attention)
www.metoffice.gov.uk/hadobs