Gormally, Alexandra and Mackenzie, Rob and Tych, Wlodzimierz (2012) Extending Manley’s Lancashire Plain Temperature Record: 1753‐2007. International Journal of Climatology, 32 (12). pp. 1899-1908. ISSN 0899-8418
Full text not available from this repository.Abstract
Before Gordon Manley began work on his famous Central England Temperature (CET) record, he constructed an instrumental temperature record for the Lancashire Plain (LP), covering 1753–1945. We describe the construction of the LP series, and discuss strategies for its extension to the present day. We analyse the relationship between the LP and the Central England Temperature (CET) record in order to assess the value of continuing the LP. Three strategies for extending the LP have been considered: replicating the original methodology Manley used to construct the LP (option 1), using linear regression on archived station data to bring the extended record ‘in line’ with the original LP (option 2), and using the CET to predict the LP (option 3). In extending the LP, the station at Morecambe (Met Office station number 16851) was found to be particularly important. Our analysis highlights the importance of maintaining long-running records at those stations, such as Morecambe, that are representative of a wider region. All three options were successfully extended to the present, providing similar results. The three extended series resulted in annual mean differences of 0.23, 0.26, and 0.29 °C below the CET, respectively. Both options 1 and 2 appear to provide a legitimate extension of the LP that is independent of the CET. Owing to uncertainties in the replication of Manley's methodology, option 2 was used in our final reconstruction of the LP to 2007. The final extension was found to be significantly correlated to (r2 = 0.912), and statistically indistinguishable from, the CET (Student's t-test, p = 1). Anomalies between the two series (CET–LP) resulted in a frequency distribution with mean of 0.26 °C, and relative standard deviation of 60%, showing that, although closely correlated, there is significant inter-annual variability in the bias between the datasets.