For this situation,using the idealized heat budget model for the bucket,they calculated the heat exchange quantity,and estimated temperature change during the measuring work.Since the heat exchange between the bucket(and water inside)and the surrounding air depends on the various measuring conditions such as air temperature,humidity,solar radiation as well as ship speed and material of bucket(wooden or canvas),they provided correction values to four types of measuring conditions: wooden bucket on slower or faster cruising ships and canvas bucket on slower or faster cruising ships. In addition,since the environment of marine meteorology depends on the sea areas and seasons,these correction values are functions of each calendar month and year,and each 50 x 50 grid points in the world's oceans(see Fig.1).Moreover,as time goes by,measuring conditions changed from wooden bucket to canvas bucket,and from slower ship to faster ship.Therefore,actual correction values are obtained by the mixture of four types of correction values.These correction values are given from 1856 around the beginning of archive of ship reports to 1941 when World War II begun.That is,they regard all measurements since 1942 have been made by the intake method.The readers may refer to Folland and Parker(1995)on details.
For the correction of SST data taken until 1941,we use the correction values of the most adjacent grids to CSST stations.Although CSST data have also been measured by the bucket method as mentioned above,we do not adopt this correction since the method is same even after 1941 at least until mid 1970s.In addition,since CSST data may be influenced by the local geographical situations,we can not expect the agreement between absolute values themselves of CSST and SST data.That is,we will make a comparison between anomaly fields of CSST and SST data.
3. Result of comparison
First,non-seasonal anomalies are calculated by subtracting the mean seasonal variation.Climatologies of mean seasonal variation are calculated using the data of 30 years from 1942 to 1971(reference period)at both CSST stations and circular areas for SST.Here,we use only so-called “match-up” data between CSST and SST:that is,for example,when SST data are blank at some months, then CSST data at those months are not used in calculation of climatologies,and vice verse.This 30-year period is chosen because no correction is needed for the SST data as mentioned before.Then,anomaly fields during the whole period are computed.Here,we prepare two SST data sets:uncorrected one and one corrected using the correction values proposed by Folland and Parker(1995).
In order to examine the validation of correction values proposed by Folland and Parker(1995).scatter plots are prepared between CSST anomalies,and corrected and uncorrected SST anomalies for the two 30-year periods of 1912-41 and 1942-71.Figures 2(a)through(c)show those at Ishigakijima station(Station i)as an example;Here Figs.2(a)and(b)show those using the uncorrected and corrected SST anomalies for the period of 1912-1941,respectively.
