Adjusting urban bias in the regional mean surface temperature series of South Korea, 1968–99

Abstract

AbstractThe purpose of this paper is to produce a higher quality regional surface temperature series by removing urban biases in South Korean surface temperatures using statistical procedures. Monthly mean temperatures for 16 stations were obtained for a period of 32 years (1968–99). Each station is defined as an urban station or a rural station. Urban (rural) stations are defined as those that have population densities greater (less) than 1000 persons per squared kilometer in 1995. Ten urban stations and six rural stations are identified. Again, urban stations are subdivided into two groups according to whether their population totals exceed one million to examine magnitude changes of urban biases with the size of urban areas. Estimates of urban bias magnitude are calculated by averaging the difference between each urban station and every rural station. Estimates of mean urban bias magnitude (T̄u–r) are calculated by averaging the yearly urban bias estimates. Estimates of the urban trend (ΔTu–r) are obtained by differencing period means (by doubling the differences obtained between yearly estimates averaged over two 16 year periods, 1968–83 and 1984–99). For annual or seasonal mean temperature Ti, the adjusted temperature T′i is determined. As all estimates of T̄u–r are greater than zero, it suggests that temperatures in urban stations are warmer than those in rural stations. Estimates of the annual mean magnitude of urban bias range from 0.35 °C for smaller urban stations to 0.50 °C for large urban stations. Also, all estimates of ΔTu–r are positive, indicating an increasing trend in the urban bias time series. Seasonal variations are found in T̄u–r and ΔTu–r. After adjusting the urban bias, an increasing trend in surface temperature series is still evident. Copyright © 2003 Royal Meteorological Society