Effects of Site-Change and Urbanisation in the Beijing Temperature Series 1977-2006

1. Effects of Site-Change and Urbanisation in the Beijing Temperature Series 1977-2006 Li Zhen Key Laboratory of Regional Climate-Environment in Temperate East Asia Institute of Atmospheric Physics, Chinese Academy of Sciences 26 – 30 May 2008, Budapest, Hungary 中国科学院东亚区域气候-环境重点实验室

2. Outline • Introduction • Data sets • Methods • Results • Conclusions

3. 1. Introduction • The Beijing station was moved from its suburban location (Daxing district) to an urban location (Beiwa road in Haidian district ) on 1/1/1981 and back on 1/7/1997. • The aims of the present study are: • Analyze and adjustsite change effects in BJ temperature series • Estimate anurbanisation-relatedwarming bias.

4. 2.Datasets Figure 1. Locations of 19 stations in the larger Beijing area used in the present study. The bold are 4 core reference stations (CRS) , the Italic fonts indicate 5 rural stations.

5. Table1. Basic information of Beijing (BJ) and 18 nearby stations with daily temperature data back to 1977 or earlier. Italic are rural stations with less-than-18K population and bold are 4 core reference stations for site-change adjustments at BJ. Correlation coefficients with BJ for 3 sub-periods (c1 for 1977-1980, c2 for 1981-1997 and c3 for 1997-2006) are calculated based on seasonal mean temperature anomalies.

7. 3.Methods The idea here is similar to that of Alexandersson (1986; 1995) . The ξ variable indicating possible site-change-induced bias is expressed as: denotes a daily temperature anomaly at BJ; denotes a (mean) daily anomaly at a (set of) reference station (s); is the difference between the average of daily anomalies of a n-day-span after day Land that of the previousn-day-span at BJ; is the same but at the reference station (s).

8. Adjustment scale: Figure 2. Daily ξ series of (BJ vs mean of MT, MY and TH), from 1/1/1980 to 31/12/1981 (centred at the relocation date for BJ, 1/1/1981, as marked in the series), for n = 1, 30, 91, 366, 732, 1098 days. A site-change effect may be possibly estimated at the seasonal scale.

9. Adjustment variable: denotes the difference between the mean of daily anomalies during the 3 months of a certain season after the relocation day and that of the previous 3 months of the same season at BJ. is the same but at a (set of) reference station (s). denotes the ithyear after/before the relocation (1≤i≤m). is the number of years for calculating the multi-year mean adjustment.

10. 4.Results Figure3. The m-year mean seasonal adjustment variables for BJ temperature records changing with m (MY and MS). • Multi-year mean seasonal adjustments In the present analysis m is set as 3.

11. Core reference stations (CRS) for site-change adjustments Figure 4. Three-year mean seasonal adjustments for BJ temperature records due to relocations in 1981/1997, based on records of different reference stations. Eliminate obvious and less obvious ‘outliers’ stations, MY, PG, TZ and FS were chosen as the CRS .

12. Site-change adjustments for BJ temperature records during 1981-1997 Table 2. Seasonal adjustments based on 4 CRS records and comparisons of three-year mean seasonal temperature anomalies before and after the relocation dates. The two relocations induced larger biases for winter and spring than for summer and autumn. The absolute values of biases for spring(0.22), summer (0.25) and annual(0.12) for 1981 are larger than for 1997 relocation.

13. A liner trend was used for adjusting the daily data at BJ during 1981-1997, for 4 seasons, respectively: -0.95 + 0.22 * n / N -0.69 + 0.25 * n / N -0.73 + 0.01 * n / N -0.86 – 0.02 * n / N

14. The effects of site-changes on trends of BJ series Figure 5. Seasonal mean temperature series at BJ during 1977-2006 (adjusted: solid line; original: dashed line). The annual plot includes for comparison the adjusted series from Li et al (2004) and Yan et al (2001).

15. Table 3. Linear trends in the adjusted and original annual and seasonal temperature series during 1977-2006 (Unit: ℃ /decade) After adjustments, the warming trends in the seasonal and annual series are enhanced. Note that the adjusted series here retain an urban-related trends.

16. Urban-related warming bias at BJ Figure 6. BJ adjusted annual temperature series in 1977-2006, compared with those of 5 rural stations. 0.78℃/decade 0.24 ℃/decade 0.58 ℃/decade The likely urban-related warming at BJ is of 0.54℃/decade , the possible upper limit of urban warming bias, about 69% contribution of the overall warming. This result coincided with Ren et al (2007). PG with a similar elevation to BJ presents 0.58 ℃/decade warming trend, and the firm bottom limit of urban warming bias is 0.20 ℃/decade.

17. Table 4. Linear trends (℃/decade) in annual temperature series for different periods. Bracketed values are the differences of trend between BJ and the station for comparison. The numbers by each site name are elevation (m) and population of the site (K). Compared with FS, the urban warming bias at BJ is likely about 0.3 ℃/decade. This estimate coincides with what Portman (1993) concluded for larger cities in China.

18. 5.Conclusion Based on comparisons between 3-year- mean seasonal temperature anomalies at BJ and the mean of 4 CRS, the site-change-induced biases were between 0.43-0.95. The adjusted annual and seasonal mean series exhibited a slightly enhanced long-term warming trend in 1977-2006. The annual mean UHI effect around the site change in 1981 was 0.12℃ larger than that in 1997. Urban-related warming bias was likely about 0.3℃/decade, accounting for about 40% of the overall warming at Beijing for the last 3-4 decades.

19. Thank you!