cc: tkarl@ncdc.noaa.gov date: Wed, 7 Apr 1999 09:34:55 -0400 from: kgallo@ncdc.noaa.gov subject: urban analysis mentioned at IPCC mtg. to: christy@atmos.uah.edu, jhansen@giss.nasa.gov, p.jones@uea.ac.uk John, Jim, and Phil: Below are John Christy ’s comments to Jim Hansen about the analysis I mentioned briefly at the IPCC meetings held here at NCDC followed by my "clarification" and additional comments. I’ve included Phil in on this email as he was also interested in the analysis and provided some helpful comments. (6) The analysis in the paper concerning the urban effect is based on population. Kevin Gallo showed evidence at the March IPCC meeting (later that week) that when stations were categorized more objectively (by AVHRR land use) an urbanization effect was found. I'm copying this message to Kevin so he can correct anything I might say. Kevin carefully stratified the stations (urban, suburban and rural) and checked differences (in pairs?). The trend for 1950-1990 for the rural was +0.16 K/decade while that of the urban stations was +0.23 K/decade (or a difference of about 0.3 over the 40 years). The suburban trend was in between, so that is evidence that Kevin's technique is consistent. The best method would be to categorize sites by CHANGE in land-use over the period, but that's a tough problem. I think you are probably correct by stating that "land-use changes" (a better phrase than urbanization) causes "no more than 0.1 C warming" for the global averages. I guess I should clarify John’s comments. In the analysis mentioned at the IPCC meetings, stations included in the GHCN data set were identified as urban, suburban, and rural based on Defense Meteorological Satellite Program-Operational Linescan System (nighttime lights) data in addition to ancillary map and population data. The analysis was partially initiated because the paper by Peterson, Gallo, et al., (1999, J. Geophys. Resr.) on global rural temperatures was being misinterpreted by the press as to the urban station contribution to the overall temperature signal (claims of no impact). Urban stations were not specifically examined in this paper. Thus, the study I initiated was an attempt to directly compare urban and rural signals at the grid cell level. Stations were analyzed on a 5 X 5 degree grid cell basis such that each grid cell included in the analysis must have contained urban and rural stations. The trends were initially analyzed with the first difference method (Peterson et al., 1998, J. Geophys. Resr.) that maximizes the number of stations included in the analysis. We did find significant differences between the urban and rural trends within the grid cells. Subsequent analysis with the traditional anomaly time series (ATS) method of trend analysis revealed no significant difference. Preliminary analysis of the first difference and ATS results, by others here at NCDC, indicates that the ATS may be more appropriate for the scale (5 by 5 degree) of this analysis. We have since begun to identify urban and rural station pairs that will be analyzed, hopefully in time to have a publication that can be cited in the next IPCC report. At this time, we plan to cite the completed and nearly published (J. Climate, May 1999) paper that examined grid cell trends in diurnal temperature range, min, and max temperature trends in the USA over the 1950-1996 interval. This study used the anomaly time series method of analysis. While differences were observed between the satellite-defined rural (-.41 degree C/100 years), suburban (-.81), and urban (-.86), trends in DTR and min temperature, they were not statistically significant. Conclusions included that; the general (urban, suburban, or rural) land use/land cover associated with surface observation stations appears to be one of the factors that can influence the trends observed in temperatures and thus should be considered in the analysis and interpretation of temperature trends. I certainly support John’s suggestion that stations should optimally be categorized by CHANGE in land-use over time. Data sets are under preparation for specific urban areas that will soon permit such analysis on a limited basis. One last comment. Should any of you wish to have stations included in your future analysis classified as urban, suburban, or rural, based on the satellite methodology, please feel free to contact me. Regards, Kevin ---------------------------------------------- Kevin Gallo, Ph.D. NOAA/National Climatic Data Center 151 Patton Ave. Asheville, NC 28801 ph: +1-828-271-4878 fax: +1-828-271-4328 email: Kevin.P.Gallo@noaa.gov National Climatic Data Center