date: Wed Jul 27 16:38:28 2005 from: Phil Jones subject: Re: Figure for AR4 to: Kevin Trenberth OK. Not sure where it could go then. Lisa's map is much more spotty than a schematic. Phil At 16:23 27/07/2005, you wrote: I put them all together: that schematic is from Groisman 2005. Kevin Phil Jones wrote: Kevin, My view here was that this could replace the schematic we have (3.8.2), unless we now have one instead of that from the Groisman et al. (2005) paper. Phil At 15:02 27/07/2005, Kevin Trenberth wrote: Hi all At present we do not have the figure Lisa sent on precvipitation in our chapter. Should we? If so should we replace one or add it? Kevin Lisa Alexander wrote: All, Oops - sorry I can't spell. Please find new graph attached. I've plotted anomalies as bars like in the temperature plots. Phil, you were right - zero line is 22.53%. Caption:- Observed trends (%) per decade for 1951-2003 for the contribution to total annual precipitation from very wet days i.e. (R95p/PRCPTOT)*100. Trends were only calculated for grid boxes where both PRCPTOT and R95p had at least 40 years of data during this period and had data until at least 1999. Black lines enclose regions where trends are significant at the 5% level. Below the map is the global annual time series anomalies (with respect to 1961-1990) defined as the percentage change from the base period average (22.5%). The red line shows a 21-term binomial fit to the data to show decadal variations. Cheers. Lisa. -----Original Message----- From: Phil Jones [[1]mailto:p.jones@uea.ac.uk] Sent: Tuesday, 26 July 2005 8:26 PM To: Lisa Alexander; Kevin Trenberth; Klein Tank, Albert Subject: RE: Figure for AR4 Lisa, Thanks for all this. The temperature ones look fine now. For the precip plot, the title needs to say 'contribution', so add the 't'. As with the temperature, it would be best if the map and the time series were in the same units, so can you plot the time series as % anomaly rather than % of the total contribution. The plot should be the same. Maybe you can add to the caption that zero is 22.5% of the annual total. It looks about this number for 1961-90. We can do this if you give us the number. Cheers Phil Cheers Phil At 10:29 26/07/2005, Lisa Alexander wrote: Hi All, I have redone the PDF's with the new subset of stations (this makes very little difference). Phil, you wanted the number of stations used in the graphs so I've included that too. Cheers. Lisa. -----Original Message----- From: Lisa Alexander Sent: Tuesday, 26 July 2005 4:41 PM To: 'Kevin Trenberth'; Phil Jones Cc: Klein Tank, Albert; David Parker Subject: RE: Figure for AR4 Hi all, I have converted the temperature maps and timeseries to days and used the same y-axis for each of the timeseries plots. Revised caption should be: Observed trends (days) per decade for 1951 to 2003 for the percentile temperature indices (a) cold nights (TN10p), (b) warm nights (TN90p), (c) cold days (TX10p), (d) warm days (TX90p). Trends were only calculated for grid boxes that had at least 40 years of data during this period and had data until at least 1999. Black lines enclose regions where trends are significant at the 5% level. Below each map is the global annual time series anomalies (with respect to 1961-1990). The red line shows a 21-term binomial fit to the data to show decadal variations. Trends are significant at the 5% level for all the indices shown using a modified Kendall tau test. From Alexander et al. (2005a). I have also created the updated R95pT map and timeseries plot. The map is not currently in the paper but I am going to add it into Fig 9. Note that R95pT is only defined where PRCPTOT and R95p both exist. The R95pT map is as the temperature map but for decadal trends (%) of the contribution to total annual precipitation from very wet days. The timeseries is shown as actual percentage contribution (as in the paper) but could be changed to anomalies for consistency with the temperature plots if required. Finally, I will have to check the PDF's since a slightly different US network has been used but I will send you an updated plot if necessary. Cheers. Lisa. -----Original Message----- From: Kevin Trenberth [[2]mailto:trenbert@ucar.edu] Sent: Tuesday, 26 July 2005 12:34 AM To: Phil Jones Cc: Klein Tank, Albert; Lisa Alexander; David Parker Subject: Re: Figure for AR4 Hi all Thanks for being so responsive, especially Lisa. Agree with Phil's assessment. I would point out that precip is in another category and % is much more appropriate there as it provides a normalization and makes more physical sense. The latter comes from Clausius-Clapeyron which expresses change in saturation vapor pressure as function of itself plus T etc. Thus it is a logarithmic relation de/e = d (ln e) or de/e is a fraction = percent. Kevin Phil Jones wrote: Lisa, Albert, Agree that the paper and the IPCC plots should be in the same units. Also the maps and the time series plots should be in the same units, and agree that days/decade are the most easily understood. The 4 temperature- related indices (warm/cold nights/days) can be in these units. The precip plot (R95pT) currently isn't in, but should with its map replace what is currently there for 3.8.2. By the way Lisa, the plot you sent (once in consistent units) will be with common question 3.1 on changes in extremes. For the text on temperature-related indices we will be using two of the panels (cold/warm nights) - top two plots from Figure 3 of the paper. If these plots have been altered by any of the updated data (it would seem not) then can you send revised plots. So, in summary, I think the consensus for the 4 temp indices will be days/decade. For David, the change in units should make no difference to the REML trends. Cheers Phil At 08:44 25/07/2005, Klein Tank, Albert wrote: Hi Lisa, Thanks for the updates. Yes, I would prefer IPCC (and the paper) to present all temperature indices results (time series plots, maps, tables) in days/decade, which I think avoids confusion about the calculated trends. I have done this in the past for Europe and the Pune paper, but I am aware that others have also used %. Albert. -----Original Message----- From: Lisa Alexander [[3]mailto:L.Alexander@bom.gov.au] Sent: maandag 25 juli 2005 07:56 To: Kevin Trenberth; Klein Tank, Albert Cc: Phil Jones; David Parker Subject: RE: Figure for AR4 Hi All, OK there seems to be some confusion (I must confess I'm a bit confused myself!). Do you want me to put either or both the timeseries and maps as % or days? I changed the timeseries plots into days as requested (hence why the y-axis was different) but I didn't change the gridded maps. I have now redone the maps and timeseries using updated data. The main changes are that additional data have been added over Brazil and Africa (precip only) and a more homogeneous US station network has been used. For some of the indices this has reduced the decorrelation length scales but for others (particularly precip) the correlations have improved. Hence you might note some slight differences in coverage in the new plots. I have also recalculated the timeseries using the new dataset. The attached plot shows the trends on the maps as % (index measure) per decade and the timeseries are shown as anomalies (as in the paper). I'm also attaching the actual global timeseries (% temperature, fraction precip as before) for David to calculate the REML trends. I can see that trends of days might be more understandable but all trends in the paper were calculated using the units pertaining to each particular index. Albert, as a co-author of the paper would you prefer to see the trends of both the maps and timeseries as days per decade? If we can come to a consensus, I would be willing to redo the diagrams if you thought this would make the results clearer (bearing in mind that I haven't received both reviews back yet). Cheers. Lisa. -----Original Message----- From: Kevin Trenberth [[4]mailto:trenbert@ucar.edu] Sent: Saturday, 23 July 2005 6:29 AM To: Albert Klein Tank; Lisa Alexander Cc: Phil Jones; David Parker Subject: Figure for AR4 Hi all, esp Lisa: I was going thru stuff on the figure and I am not sure we have it right. Albert suggested that the change in units of the time series makes it days instead of %. I do notice some small discrepancies with the original if it is % in relative values rescaled . [Is it that 10% is 10% of 365 so it scales to 36.5 days?] The problem is the maps are in %. So shouldn't we do the time series in % too? We can add explanation in terms of days in the text or cpation perhaps? Lisa does this mean the panels for the time series need to be redone with % scaling? Hopefully all the same. Or can we explain what the time series is and why the maps are %? Thanks Kevin -- **************** Kevin E. Trenberth e-mail: [5]trenbert@ucar.edu Climate Analysis Section, NCAR [6]www.cgd.ucar.edu/cas/ P. O. Box 3000, (303) 497 1318 Boulder, CO 80307 (303) 497 1333 (fax) Street address: 1850 Table Mesa Drive, Boulder, CO 80303 Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email [7]p.jones@uea.ac.uk NR4 7TJ UK ----------------------------------------------------------------------- - ---- -- **************** Kevin E. Trenberth e-mail: [8]trenbert@ucar.edu Climate Analysis Section, NCAR [9]www.cgd.ucar.edu/cas/ P. O. Box 3000, (303) 497 1318 Boulder, CO 80307 (303) 497 1333 (fax) Street address: 1850 Table Mesa Drive, Boulder, CO 80303 Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email [10]p.jones@uea.ac.uk NR4 7TJ UK ------------------------------------------------------------------------ ---- -- **************** Kevin E. Trenberth e-mail: [11]trenbert@ucar.edu Climate Analysis Section, NCAR [12]www.cgd.ucar.edu/cas/ P. O. Box 3000, (303) 497 1318 Boulder, CO 80307 (303) 497 1333 (fax) Street address: 1850 Table Mesa Drive, Boulder, CO 80303 Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email [13]p.jones@uea.ac.uk NR4 7TJ UK ---------------------------------------------------------------------------- -- **************** Kevin E. Trenberth e-mail: [14]trenbert@ucar.edu Climate Analysis Section, NCAR [15]www.cgd.ucar.edu/cas/ P. O. Box 3000, (303) 497 1318 Boulder, CO 80307 (303) 497 1333 (fax) Street address: 1850 Table Mesa Drive, Boulder, CO 80303 Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email p.jones@uea.ac.uk NR4 7TJ UK ----------------------------------------------------------------------------