全球变化知识资源中心

Time series of historical annual average temperature, total precipitation, and extreme weather indices were constructed and analyzed for 103 (for temperature indices) and 115 (for precipitation indices) U.S. cities with climate records starting earlier than 1900. Mean rate of change and related 95% confidence bounds were calculated for each city using linear regression for the full periods of record. Box-Cox transformations of some time series of climate records were performed to address issues of non-normal distribution. Thirteen cities among the nine U.S. climate regions were selected and further evaluated with adequacy diagnoses and analyses for each month. The results show that many U.S. cities exhibit long-term historical increases in annual average temperature and precipitation, although there are spatial and temporal variations in the observed trends among the cities. Some cities in the Ohio Valley and Southeast regions exhibit decreasing or statistically nonsignificant increasing trends in temperatures. Many of the cities exhibiting statistically significant increases in precipitation are in the Northeast and Upper Midwest regions. The records for the cities are individually unique in both annual and monthly change, and cities within the same climate region sometimes exhibit substantially different changes. Within the full periods of record, discernible decade-long subtrends were observed for some cities; consequently, analysis of selected shorter periods can lead to inconclusive and biased results. These statistical analyses of constructed time series of city-specific long-term historical climate records provide detailed historical climate change information for cities across the United States.