• Allsopp, D., 2004: History of the Digital Climate Archive. Extended Abstracts, Workshop on Climate Data Homogenization, Toronto, ON, Canada, Environment Canada, 1–11.

    • Search Google Scholar
    • Export Citation
  • Baker, G. B., 1975: Effect of observation time on mean temperature estimation. J. Appl. Meteor., 14 , 471476.

  • Belcher, B. N., , and A. T. DeGaetano, 2003: A method for operational detection of daily observation-time changes. J. Appl. Meteor., 42 , 18231836.

    • Search Google Scholar
    • Export Citation
  • Bootsma, A., 1976: A note on minimum temperature and the climatological day at first order stations. Atmosphere, 14 , 5355.

  • Byrd, G. P., 1985: An adjustment for the effect of observation time on mean temperature and degree-day computations. J. Climate Appl. Meteor., 24 , 869874.

    • Search Google Scholar
    • Export Citation
  • Cameron, T., , and H. Wilson, 1996: Minimum temperature bias introduced by a redefinition of the climatological day. Atmospheric Environment Service Internal Rep., Toronto, ON, Canada, 34 pp.

    • Search Google Scholar
    • Export Citation
  • DeGaetano, A. T., 1999: A method to infer observation time based on day-to-day temperature variations. J. Climate, 12 , 34433456.

  • Hutchinson, M. F., , D. W. McKenney, , K. Lawrence, , J. H. Pedlar, , R. F. Hopkinson, , E. Milewska, , and P. Papadopol, 2009: Development and testing of Canada-wide interpolated spatial models of daily minimum–maximum temperature and precipitation for 1961–2003. J. Appl. Meteor. Climatol., 48 , 725741.

    • Search Google Scholar
    • Export Citation
  • Janis, M. J., 2002: Observation-time-dependent biases and departures for daily minimum and maximum air temperature. J. Appl. Meteor., 41 , 588603.

    • Search Google Scholar
    • Export Citation
  • Karl, T. R., , C. N. Williams Jr., , and P. J. Young, 1986: A model to estimate the time of observation bias associated with monthly mean maximum, minimum and mean temperatures for the United States. J. Climate Appl. Meteor., 25 , 145160.

    • Search Google Scholar
    • Export Citation
  • Schaal, L. A., , and R. F. Dale, 1977: Time of observation temperature bias and “climatic change”. J. Appl. Meteor., 16 , 215222.

  • Vincent, L. A., , and D. W. Gullett, 1999: Canadian historical and homogeneous temperature datasets for climate change analyses. Int. J. Climatol., 19 , 13751388.

    • Search Google Scholar
    • Export Citation
  • Vincent, L. A., , X. Zhang, , B. R. Bonsal, , and W. D. Hogg, 2002: Homogenization of daily temperatures over Canada. J. Climate, 15 , 13221334.

    • Search Google Scholar
    • Export Citation
  • Vose, R. S., , C. N. Williams Jr., , T. C. Peterson, , T. R. Karl, , and D. R. Easterling, 2003: An evaluation of the time of observation bias adjustment in the U.S. Historical Climatology Network. Geophys. Res. Lett., 30 , 2046. doi:10.1029/2003GL018111.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Locations of the 121 stations. The stars indicate the stations mentioned in the text. Canada is currently divided into six time zones—1: Newfoundland, 2: Atlantic, 3: eastern, 4: central, 5: mountains, and 6: Pacific standard time.

  • View in gallery

    Hourly temperature from 0000 to 2300 LST at Kapuskasing from 18 to 23 Jul 2007. The highest hourly value is identified by a gray dot for the (a) window ending at 1200 UTC and (c) window ending at 0600 UTC. The lowest hourly value is identified by a gray dot for the (b) window ending at 0000 UTC and (d) window ending at 0600 UTC.

  • View in gallery

    Difference between the annual mean of (a) the highest hourly value from the window ending at 0600 UTC and that from the window ending at 1200 UTC and (b) the lowest hourly value from the window ending at 0600 UTC and that from the window ending at 0000 UTC for 1961–2007. In (b), a filled triangle indicates that the means are significantly different at the 5% level when a t test is used, and the size of the triangle is proportional to the magnitude of the difference.

  • View in gallery

    Difference between the monthly means of the lowest hourly values from the window ending at 0600 UTC and those from the window ending at 0000 UTC at Penticton (black), Regina (gray), and Fredericton (white) for 1961–2007.

  • View in gallery

    Annual mean of the differences between the daily minimum temperatures and the lowest hourly value from the window ending at 0000 UTC for 1953–60.

  • View in gallery

    Annual percentage of adjusted days at five stations. A day is adjusted when the hourly lowest temperature of the window ending at 0600 UTC is smaller than the hourly lowest temperature of the window ending at 0000 UTC.

  • View in gallery

    (a) Seasonal percentage of adjusted days and (b) seasonal mean of the daily adjustment at Penticton (black), Calgary (dark gray), Regina (light gray), Kapuskasing (white), and Fredericton (dotted) for 1961–2007.

  • View in gallery

    Trends in annual mean of the daily minimum temperature for 1950–2007 (a) before adjustment and (b) after adjustment, and (c) the difference between (b) and (a).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 4 4 2
PDF Downloads 1 1 1

Bias in Minimum Temperature Introduced by a Redefinition of the Climatological Day at the Canadian Synoptic Stations

View More View Less
  • 1 Climate Research Division, Science and Technology Branch, Environment Canada, Toronto, Ontario, Canada
  • 2 Custom Climate Services, Regina, Saskatchewan, Canada
  • 3 Climate and Water Department, World Meteorological Organization, Geneva, Switzerland
© Get Permissions
Full access

Abstract

On 1 July 1961, the climatological day was redefined to end at 0600 UTC (coordinated universal time) at all synoptic (airport) stations in Canada. Prior to that, the climatological day ended at 1200 UTC for maximum temperature and 0000 UTC for minimum temperature. This study shows that the redefinition of the climatological day in 1961 has created a cold bias in the annual and seasonal means of daily minimum temperatures across the country while the means of daily maximum temperatures were not affected. Hourly temperatures taken at 121 stations for 1953–2007 are used to determine the magnitude of the bias and its spatial variation. It was found that the bias is more pronounced in the eastern regions; its annual mean varies from −0.2° in the west to −0.8°C in the east. Not all days are affected by this change in observing time, and the annual percentage of affected days ranges from 15% for locations in the west to 38% for locations in the east. An approach based on hourly values is proposed for adjusting the affected daily minimum temperatures over 1961–2007. The adjustment on any individual day varies from 0.5° to 12.5°C. The impact of the adjustment is assessed by examining the trends in the annual mean of the daily minimum temperatures for 1950–2007. Overall, with the adjustment, the trends are becoming either more positive or are reversing from negative to positive, and they have changed by as much as 1°C in numerous locations in the eastern regions.

Corresponding author address: Lucie A. Vincent, 4905 Dufferin Street, Toronto, ON M3H 5T4, Canada. Email: lucie.vincent@ec.gc.ca

Abstract

On 1 July 1961, the climatological day was redefined to end at 0600 UTC (coordinated universal time) at all synoptic (airport) stations in Canada. Prior to that, the climatological day ended at 1200 UTC for maximum temperature and 0000 UTC for minimum temperature. This study shows that the redefinition of the climatological day in 1961 has created a cold bias in the annual and seasonal means of daily minimum temperatures across the country while the means of daily maximum temperatures were not affected. Hourly temperatures taken at 121 stations for 1953–2007 are used to determine the magnitude of the bias and its spatial variation. It was found that the bias is more pronounced in the eastern regions; its annual mean varies from −0.2° in the west to −0.8°C in the east. Not all days are affected by this change in observing time, and the annual percentage of affected days ranges from 15% for locations in the west to 38% for locations in the east. An approach based on hourly values is proposed for adjusting the affected daily minimum temperatures over 1961–2007. The adjustment on any individual day varies from 0.5° to 12.5°C. The impact of the adjustment is assessed by examining the trends in the annual mean of the daily minimum temperatures for 1950–2007. Overall, with the adjustment, the trends are becoming either more positive or are reversing from negative to positive, and they have changed by as much as 1°C in numerous locations in the eastern regions.

Corresponding author address: Lucie A. Vincent, 4905 Dufferin Street, Toronto, ON M3H 5T4, Canada. Email: lucie.vincent@ec.gc.ca

1. Introduction

Daily climatological observations are usually recorded once or twice per day for a 24-h period that does not necessarily correspond to the calendar day. The most common practice at ordinary volunteer climate stations in Canada has been to observe the maximum temperature for the previous day at 0700 local standard time (LST) and minimum temperature for the current day at 1700 LST, according to the information contained in the Station Information System of National Climate Data Archives. At first-order synoptic stations (mainly airports), two observations were also taken every day: the maximum temperature was obtained from the morning observations and assigned to the previous day while the minimum temperature was taken from the afternoon observations and assigned to that day. In the 1930s, 0630 and 1830 LST were officially recognized as the morning and afternoon observing times (Allsopp 2004). During the 1940s, they were slightly changed to 1230 and 0030 coordinated universal time (UTC), and then from 1957 they were changed again to 1200 and 0000 UTC so that the data from all airport stations were comparable in time. Meanwhile, the Atmospheric Environment Service was under pressure to have the climatological day be identified more closely with the calendar day because it was easier for the public to understand (M. K. Thomas 1984, personal communication). On 1 July 1961, the climatological day at all first-order stations was redefined to end at 0600 UTC for both maximum and minimum temperatures because 0600 UTC was the synoptic time closest to the local midnight at all stations across the country. This study shows that the redefinition of the climatological day in 1961 at synoptic stations in Canada has introduced a bias in the annual and seasonal means of the daily minimum temperatures and that adjustment should be made on the daily minimum temperatures to address this problem.

In a preliminary study, the frequency and magnitude of the bias were examined at four Canadian airport stations (Cameron and Wilson 1996). The daily minimum temperatures were obtained from two observing windows, 0000 and 0600 UTC, over the period of 1961–94, using the original synoptic data extracted from microfilms held in the National Climate Data Archives. On microfilms, the maximum and minimum temperatures were given at four observing times (0000, 0600, 1200, and 1800 UTC) so that it was relatively easy to obtain the minimum temperature for a window ending at 0000 UTC and a window ending at 0600 UTC. The results showed that the change in observing window influenced more stations located in the eastern regions of the country, with about 40% of their daily minimum temperatures affected. The annual mean temperature differences between both windows were from 0.6° to 0.8°C in the east while they were much smaller in the west. Even though this procedure based on using synoptic observations on microfilms was appropriate for investigating the change in the observing time, it was impractical and labor intensive, and other alternatives needed to be developed.

Temperature bias due to observation time was also investigated in other studies. Morning and evening observations of 24-h maximum and minimum temperatures were estimated from hourly temperature records and were compared with calendar-day observations for 1961–95 at 209 locations across the contiguous United States (Janis 2002). The results showed near-zero differences on more than 70% of the days; however, large bias occurred when very cold mornings or very warm afternoons influenced the temperature on successive days. It was also found that the bias was more pronounced during the winter. Other studies have indicated that the bias due to differing observing times was accumulated in monthly, seasonal, and annual values of variables such as the heating, cooling, and growing degree-days that are derived from the daily mean temperature, and this fact could provide misleading information for applications in industry and agriculture (Baker 1975; Bootsma 1976; Schaal and Dale 1977; Byrd 1985). It consequently became necessary to develop procedures for adjusting this temperature bias, which sometimes required inferring unknown observation times (Karl et al. 1986; DeGaetano 1999; Belcher and DeGaetano 2003; Vose et al. 2003).

Considerable efforts were devoted to resolve this issue in the United States, but far fewer studies were done in Canada to examine this problem carefully. When a homogeneity procedure was applied to the annual mean of the daily maximum and minimum temperatures at 210 Canadian stations (Vincent and Gullett 1999), a decreasing step of about 0.6°–0.8°C was found in 1961 in the minimum temperatures. This step accentuated a cooling trend observed over eastern Canada for the past 50 yr. In western Canada, a significant step was not detectable. It was concluded that indeed the step resulted from the nationwide change in observing time at the first-order stations. Adjustments were applied on monthly values and, later, on daily values (Vincent et al. 2002). Every daily minimum temperature since 1961 was adjusted by an interpolated value derived from the monthly adjustments. At that time, it was recognized that the minimum temperature bias due to change in observing time needed further investigation, but for the time being it was sufficiently solved for the purpose of preserving consistency between the adjusted daily observations and the homogeneous monthly time series.

The purpose of this study is to examine closely the effects of the redefinition of the climatological day in 1961 on daily maximum and minimum temperatures at synoptic stations in Canada. Hourly temperatures are used to illustrate the problem and to estimate the bias on the annual and seasonal means. A procedure based on hourly temperatures is used to adjust daily minimum temperatures over 1961–2007. The impact of the adjustments is assessed on annual and seasonal trends and on other parameters based on degree-days.

2. Data

Hourly values of temperature were directly retrieved from the National Climate Data Archives of Environment Canada for the period 1953–2007. Hourly observations first began at airports in the late 1940s and early 1950s, but they have been accumulated in digital form only from 1953. Stations were selected based on their data quality and availability. Most stations had fewer than 1% of values missing, and a few had as many as 5%. In the daily record of hourly data, the 24 entries represent the observation taken directly on the hour from 0000 to 2300 LST. Figure 1 shows the location of the 121 synoptic stations used in this study.

Daily maximum and minimum temperatures were also retrieved from the National Climate Data Archives for the same period and for 121 stations. The daily maximum and minimum temperatures represent the highest and lowest values over a 24-h interval of continuous temperature, and they do not necessarily correspond to the highest and lowest values of the hourly observations. Because the observing time is based on coordinated universal time, the temperature is observed at different local times across the country. Figure 1 shows the six current time zones in Canada. The first zone, Newfoundland, is 3.5 h behind coordinated universal time. The second zone, Atlantic, is 4 h behind coordinated universal time. All of the other zones are successively 1 h behind their eastern neighbor. For example, when the observing time is 0600 UTC, the observation is taken at 0230 LST in the Newfoundland zone, 0200 LST in the Atlantic zone, 0100 LST in the eastern zone, 0000 LST (midnight) in the central zone, 2300 LST of the previous day in the mountains zone, and 2200 LST of the previous day in the Pacific zone. The Yukon used to be in a separate time zone with observing time at 2100 LST until it was moved to the Pacific zone in 1975.

3. Examination of the temperature bias

a. Illustration of the bias using hourly temperatures

Prior to July of 1961, the daily maximum temperature was observed at 1200 UTC and the daily minimum temperature was observed at 0000 UTC. On 1 July 1961, the climatological day was redefined to end at 0600 UTC for both maximum and minimum temperatures. Hourly temperatures are used to examine the effect of this change on observing time. The highest (lowest) value of the hourly temperatures is used to estimate the daily maximum (minimum) temperature for a window ending at 1200 UTC (0000 UTC) and a window ending at 0600 UTC. Figure 2 shows the hourly temperatures of Kapuskasing (Ontario) for 18–23 July 2007 with the highest and lowest values identified for the different observing times. The highest values of the window ending at 1200 UTC (0700 LST) and at 0600 UTC (0100 LST) are the same because the maximum temperature usually occurs at the end of the afternoon and both observing times take place during the early morning. However, the lowest values of the window ending at 0000 UTC (1900 LST) and 0600 UTC (0100 LST) are different on 19 July because 0600 UTC is close to when the actual minimum temperature occurs on the following day: the lowest hourly temperature has subsequently changed from 10.3° to 6.9°C with the redefinition of the climatological day.

The highest (lowest) values of the hourly temperatures are identified for the windows ending at 1200 UTC (0000 UTC) and ending at 0600 UTC for 1961–2007 at the 121 stations. The annual and monthly means of the highest and lowest values are averaged over the 47 yr for the different observing times. When the annual means of the highest values are examined, the difference between both windows is near zero at all locations across the country (Fig. 3a). On the other hand, the annual means of the lowest values of the 0600 UTC window are always colder than those of the 0000 UTC window (Fig. 3b). The cooling bias becomes gradually more pronounced in the east, and the differences are ranging from −0.2°C for the stations located in the west to −0.8°C for those in the east. Furthermore, the bias seems to be more important in the east where the difference is statistically significant at the 5% level. There are a few stations on the east coast for which the bias is small and not significant, possibly because of the moderating effect of the Atlantic Ocean, but this would require further investigation to confirm.

The differences between the monthly means of the highest (and lowest) values are also examined. For the highest values, the differences are generally near zero across the country. However, the monthly means of the lowest values taken at 0600 UTC are always colder than those taken at 0000 UTC. The monthly bias is less evident near the Pacific coast and it becomes gradually more pronounced toward the Atlantic coast. For example, the January bias is −0.3°, −0.6°, and −0.8°C at Penticton (British Columbia), Regina (Saskatchewan), and Fredericton (New Brunswick), respectively (Fig. 4). Furthermore, the results show that the bias is not constant during the year, and it seems that the new observing time has generated a greater bias during winter and autumn. For example, the monthly bias is −0.2°, −0.6°, and −1.1°C in September whereas it is only −0.1°, −0.3°, and −0.6°C in July at the same three stations, respectively (Fig. 4).

b. Comparing daily minimum temperatures with lowest hourly temperatures

Daily minimum temperatures are compared with the lowest hourly 0000 UTC window temperatures at the 121 stations over 1953–60 to determine if the lowest hourly 0000 UTC temperature could be a good estimate of the daily minimum. The period 1953–60 is selected because the daily minimums are not yet affected by the redefinition of the climatological day. It is found that the lowest hourly temperature rarely corresponded to the “true” daily minimum temperature and that the daily minimum is often colder than the lowest hourly 0000 UTC value. When the annual means of the difference between the daily minimum temperature and the lowest hourly 0000 UTC temperature are examined, the difference is near −0.5°C across the country (Fig. 5). In fact, the annual mean of the difference varies randomly from −0.3° to −0.7°C with no evidence of spatial variation. The seasonal means of the difference are also near −0.5°C (not shown). For example, at Fredericton, they are −0.6°, −0.4°, −0.5°, and −0.5°C for winter, spring, summer, and autumn, respectively. Therefore it is concluded that the daily minimum temperature is on average 0.5°C colder than the lowest hourly 0000 UTC temperature.

4. Adjusting daily minimum temperature using the lowest hourly temperature

The examination of the bias using the hourly temperatures has clearly indicated that the redefinition of the climatological day in 1961 has affected the daily minimum temperatures and not the maxima. Therefore, a simple procedure is proposed for adjusting the daily minimum temperatures. First, a day is identified on which the lowest hourly 0600 UTC window temperature is colder than the lowest hourly 0000 UTC window temperature. Second, for this particular day, the original daily minimum is replaced by the lowest hourly 0000 UTC window temperature. Third, 0.5°C is subtracted from this estimate to remove the small difference detected between the true daily minimum and the lowest hourly value (see section 3b). For example, the lowest hourly 0600 UTC window temperature at Kapuskasing is 6.9°C on 19 July 2007 whereas its lowest value at the 0000 UTC window is 10.3°C (Figs. 2d,b): therefore 19 July is affected by the change in observing time. The current daily minimum temperature for 19 July is 6.8°C. It is first replaced by the lowest hourly temperature for the window ending at 0000 UTC (10.3°C), and then the small difference between the true daily minimum temperature and the lowest hourly value is removed by subtracting 0.5°C. At the end, the adjusted daily minimum temperature for 19 July is 9.8°C. This procedure ensures that only the days affected by the redefinition of the climatological day are adjusted and not every day of the calendar, as was previously done in Vincent et al. (2002).

The procedure is applied to adjust the daily minimum temperatures at the 121 synoptic stations from 1 July 1961 to 31 December 2007. Because the redefinition of the climatological day did not affect every day of the year, the percentage of affected days is carefully examined. The annual and seasonal percentages of days for which the lowest hourly temperature of the 0600 UTC window is colder than the lowest temperature of the 0000 UTC window are obtained at each station for 1961–2007. The results indicate that the annual percentage of affected days is gradually increasing toward the east, from 15% for stations in the west to 38% for stations in the east. For example, Fig. 6 shows that, on average, 16% of the days are affected by the change in observing time in Penticton whereas 35% of the days are affected in Fredericton. The figure also shows that the percentage of affected days varies from one year to another.

The seasonal percentage of adjusted days also increases gradually from west to east. For example, the percentage of adjusted days during the autumn is 16.2%, 21.3%, 23.4%, 30.9%, and 31.8% for Penticton, Calgary (Alberta), Regina, Kapuskasing, and Fredericton, respectively (Fig. 7a). In addition, the results show that more days are adjusted during the winter and autumn. For example, at Calgary, the percentages of adjusted days are 25.6% and 21.3% respectively in the winter and autumn, whereas they are 12.6% and 9.0% in the spring and summer (Fig. 7a).

Original and adjusted daily minimum temperatures are compared at the 121 stations for 1961–2007 to determine the magnitude of adjustment. Overall, the daily adjustment ranges from 0.5° to 12.5°C with an average of 2.6°C. For example, at Kapuskasing, the original and adjusted daily minimum temperatures are 6.8° and 9.8°C, respectively, on 19 July 2007; therefore the adjustment is 3.0°C. The results show no evidence of spatial pattern in the magnitude of the seasonal and monthly means of the daily adjustments. For example, the winter means are 2.2°, 3.1°, 3.0°, 3.6°, and 3.1°C for Penticton, Calgary, Regina, Kapuskasing, and Fredericton, respectively (Fig. 7b). Therefore, it is concluded that the redefinition of the climatological day in 1961 has created a larger bias at locations in the eastern regions because more days are affected by the new observing time and not because the magnitude of the adjustment on individual days is larger. In a similar way, it is concluded that the bias is larger during the winter and autumn because more days are affected by the new observing time during these seasons.

5. Impact of the adjustments

a. Impact on the trends

Annual and seasonal trends for 1950–2007 were obtained at individual stations before and after adjustments to determine the impact of the adjustments on several climatic parameters based on daily minimum temperature. The annual and seasonal means of the daily minimum temperatures, the 5th and 95th percentiles, and the diurnal temperature range (difference between the daily maximum and minimum temperatures) were examined. The best-fit linear trend was used to estimate the linear change over the 58 yr, and the t test was applied to determine its statistical significance. The trends were computed only if more than 80% of values were present.

Overall, the adjustments alter the trends. Because many low daily minimum temperatures become warmer during the period 1961–2007, the trends are generally more positive or are reversing from negative to positive with the adjustment—in particular, in the eastern regions. For example, the annual mean of the daily minimum temperature initially shows a significant warming in the west and a slight cooling in the east (Fig. 8a). With the adjustments, the warming observed in the west remains while most stations in the east show a small warming instead of a cooling (Fig. 8b). The difference between the trends before and after adjustments is positive, and the trends have changed by as much as 1°C in numerous locations in the eastern regions (Fig. 8c). The impact is alike for the seasons, although the greater change in trends is observed in the winter. Similar trend patterns are also evident in the 5th and 95th percentiles of the daily minimum temperature. For example, the summer 95th percentile indicates more stations with larger and significant increasing trends with the adjustments, and the difference in trends before and after adjustment gradually increases toward the east. For the diurnal temperature range, the trends over the past 58 yr are more negative with the adjustment because several cold nighttime temperatures have been adjusted by a warmer value (not shown).

b. Impact on degree-days

Daily maximum and minimum temperatures are used to derive climatic parameters such as heating, cooling, and growing degree-days. Heating and cooling degree-days are used by the heating and air-conditioning industries in design studies and long-term planning of fuel needs. Growing degree-days are helpful for predicting crop development. The degree-days are based on the daily mean temperature, which is computed from the daily maximum and minimum temperatures. As a consequence, the degree-days are also affected by the redefinition of the climatological day in 1961.

The annual accumulated degree-days were obtained at individual stations using the original and adjusted minimum temperatures, and they were averaged over 1961–2007. The degree-days are based on the daily mean temperature departures from a constant, and the departures are summed to determine the annual accumulated number of degree-days. For this study, the heating, cooling, and growing degree-days are the degrees below 18°C, above 18°C, and above 5°C, respectively. The differences between the annual accumulated degree-day averages were obtained before and after adjustment. The results show that the heating degree-days are considerably reduced with the daily minimum temperature’s adjustment—by 50–150 degree-days per year depending on the location, and this reduction is more pronounced in the eastern regions. The cooling degree-days increase with the adjustment by 15–30 degree days per year in the eastern part of the country. The growing degree-days also show an increase with the adjustment, and several stations indicate an additional 25–75 degree days per year. Overall, the change is more evident in the eastern regions of the country.

6. Discussion

The daily minimum usually occurs early in the morning. The 0000 UTC window was a better time for recording the daily minimum because it corresponded to late afternoon (Pacific)–evening (Atlantic) across the country. The 0600 UTC window is not well suited because it is closer to the time at which the actual minimum occurs—in particular, in eastern Canada. It consequently is more likely to record the same minimum (or a similar minimum) on two successive days, creating in this manner a cold bias in the annual, seasonal, and monthly means of the daily minimum temperatures.

The problem is ongoing, and the adjustments have to be applied to the future records. It would have been easier to adjust the data prior to July 1961 but less viable, as it would make all records at principal stations incompatible with the majority of the ordinary climatological stations (volunteer sites), which have 0700 and 1700 LST observations. In addition, it would have made many daily minimum temperatures “incorrect” before 1961 because the same minimum would have been recorded on two successive days.

There is another concern, due to different observing schedules at ordinary volunteer climate stations and first-order stations, that has not been addressed here. Although the observing time is a temporal homogeneity issue where it may create an artificial step in the station’s historical temperature time series, it is also an important consideration in spatial analyses. Adjacent stations with different observing schedules may have disparate temperature records related to different definitions of the climatological day. The wide range of temperatures at nearby locations could be difficult to resolve by any interpolation scheme and could result in so-called bull’s-eyes of aberrant interpolated values. This problem is being addressed in another work currently under way. Preliminary results show that biases for temperature between ordinary and first-order stations, exclusively due to the different definition of climatological day, range from about −0.2°C in the west and north to −1.0°C in the east and south, and they are more pronounced in winter. All first-order stations across Canada are scheduled to be adjusted for this bias during the preparation of a second version of daily temperature and precipitation grids for Canada (Hutchinson et al. 2009).

7. Summary and conclusions

This study shows that the redefinition of the climatological day on 1 July 1961 has created a cold bias in the annual and seasonal means of the daily minimum temperature at airport stations across the country. Many climate parameters based on the daily minimum temperature are also affected by this change in observing time. The adjustments proposed in this study improve the assessment of long-term trends and the computation of degree-days. The most important outcome is that now the data series after 1961 are more compatible with the series before 1961.

A recommendation is that all first-order stations continue to record the maximum and minimum temperatures every 6 h but that these observations be stored and archived for easy access by all users. The users could then choose the daily minimum temperature according to the requirements of their own analyses and applications. This flexibility would alleviate the problem of changing observing time and also it would partially solve the issues regarding the different observing schedules at first-order and ordinary stations.

Daily homogenized maximum, minimum, and mean temperatures have been prepared for the analyses of climate change in Canada (Vincent et al. 2002). It is believed that these adjusted data are so far the most reliable datasets for trends analyses in Canada. A second generation of homogenized temperature is currently under preparation, and the new datasets will include the new daily adjustments for the change in observing time in 1961 as proposed in this study.

Acknowledgments

The authors thank three anonymous reviewers for their helpful comments that contributed to an improved version of this paper. The authors also thank Tanya Cameron and Hannah Wilson for their great work while spending invaluable time searching for data on microfilm.

REFERENCES

  • Allsopp, D., 2004: History of the Digital Climate Archive. Extended Abstracts, Workshop on Climate Data Homogenization, Toronto, ON, Canada, Environment Canada, 1–11.

    • Search Google Scholar
    • Export Citation
  • Baker, G. B., 1975: Effect of observation time on mean temperature estimation. J. Appl. Meteor., 14 , 471476.

  • Belcher, B. N., , and A. T. DeGaetano, 2003: A method for operational detection of daily observation-time changes. J. Appl. Meteor., 42 , 18231836.

    • Search Google Scholar
    • Export Citation
  • Bootsma, A., 1976: A note on minimum temperature and the climatological day at first order stations. Atmosphere, 14 , 5355.

  • Byrd, G. P., 1985: An adjustment for the effect of observation time on mean temperature and degree-day computations. J. Climate Appl. Meteor., 24 , 869874.

    • Search Google Scholar
    • Export Citation
  • Cameron, T., , and H. Wilson, 1996: Minimum temperature bias introduced by a redefinition of the climatological day. Atmospheric Environment Service Internal Rep., Toronto, ON, Canada, 34 pp.

    • Search Google Scholar
    • Export Citation
  • DeGaetano, A. T., 1999: A method to infer observation time based on day-to-day temperature variations. J. Climate, 12 , 34433456.

  • Hutchinson, M. F., , D. W. McKenney, , K. Lawrence, , J. H. Pedlar, , R. F. Hopkinson, , E. Milewska, , and P. Papadopol, 2009: Development and testing of Canada-wide interpolated spatial models of daily minimum–maximum temperature and precipitation for 1961–2003. J. Appl. Meteor. Climatol., 48 , 725741.

    • Search Google Scholar
    • Export Citation
  • Janis, M. J., 2002: Observation-time-dependent biases and departures for daily minimum and maximum air temperature. J. Appl. Meteor., 41 , 588603.

    • Search Google Scholar
    • Export Citation
  • Karl, T. R., , C. N. Williams Jr., , and P. J. Young, 1986: A model to estimate the time of observation bias associated with monthly mean maximum, minimum and mean temperatures for the United States. J. Climate Appl. Meteor., 25 , 145160.

    • Search Google Scholar
    • Export Citation
  • Schaal, L. A., , and R. F. Dale, 1977: Time of observation temperature bias and “climatic change”. J. Appl. Meteor., 16 , 215222.

  • Vincent, L. A., , and D. W. Gullett, 1999: Canadian historical and homogeneous temperature datasets for climate change analyses. Int. J. Climatol., 19 , 13751388.

    • Search Google Scholar
    • Export Citation
  • Vincent, L. A., , X. Zhang, , B. R. Bonsal, , and W. D. Hogg, 2002: Homogenization of daily temperatures over Canada. J. Climate, 15 , 13221334.

    • Search Google Scholar
    • Export Citation
  • Vose, R. S., , C. N. Williams Jr., , T. C. Peterson, , T. R. Karl, , and D. R. Easterling, 2003: An evaluation of the time of observation bias adjustment in the U.S. Historical Climatology Network. Geophys. Res. Lett., 30 , 2046. doi:10.1029/2003GL018111.

    • Search Google Scholar
    • Export Citation
Fig. 1.
Fig. 1.

Locations of the 121 stations. The stars indicate the stations mentioned in the text. Canada is currently divided into six time zones—1: Newfoundland, 2: Atlantic, 3: eastern, 4: central, 5: mountains, and 6: Pacific standard time.

Citation: Journal of Applied Meteorology and Climatology 48, 10; 10.1175/2009JAMC2191.1

Fig. 2.
Fig. 2.

Hourly temperature from 0000 to 2300 LST at Kapuskasing from 18 to 23 Jul 2007. The highest hourly value is identified by a gray dot for the (a) window ending at 1200 UTC and (c) window ending at 0600 UTC. The lowest hourly value is identified by a gray dot for the (b) window ending at 0000 UTC and (d) window ending at 0600 UTC.

Citation: Journal of Applied Meteorology and Climatology 48, 10; 10.1175/2009JAMC2191.1

Fig. 3.
Fig. 3.

Difference between the annual mean of (a) the highest hourly value from the window ending at 0600 UTC and that from the window ending at 1200 UTC and (b) the lowest hourly value from the window ending at 0600 UTC and that from the window ending at 0000 UTC for 1961–2007. In (b), a filled triangle indicates that the means are significantly different at the 5% level when a t test is used, and the size of the triangle is proportional to the magnitude of the difference.

Citation: Journal of Applied Meteorology and Climatology 48, 10; 10.1175/2009JAMC2191.1

Fig. 4.
Fig. 4.

Difference between the monthly means of the lowest hourly values from the window ending at 0600 UTC and those from the window ending at 0000 UTC at Penticton (black), Regina (gray), and Fredericton (white) for 1961–2007.

Citation: Journal of Applied Meteorology and Climatology 48, 10; 10.1175/2009JAMC2191.1

Fig. 5.
Fig. 5.

Annual mean of the differences between the daily minimum temperatures and the lowest hourly value from the window ending at 0000 UTC for 1953–60.

Citation: Journal of Applied Meteorology and Climatology 48, 10; 10.1175/2009JAMC2191.1

Fig. 6.
Fig. 6.

Annual percentage of adjusted days at five stations. A day is adjusted when the hourly lowest temperature of the window ending at 0600 UTC is smaller than the hourly lowest temperature of the window ending at 0000 UTC.

Citation: Journal of Applied Meteorology and Climatology 48, 10; 10.1175/2009JAMC2191.1

Fig. 7.
Fig. 7.

(a) Seasonal percentage of adjusted days and (b) seasonal mean of the daily adjustment at Penticton (black), Calgary (dark gray), Regina (light gray), Kapuskasing (white), and Fredericton (dotted) for 1961–2007.

Citation: Journal of Applied Meteorology and Climatology 48, 10; 10.1175/2009JAMC2191.1

Fig. 8.
Fig. 8.

Trends in annual mean of the daily minimum temperature for 1950–2007 (a) before adjustment and (b) after adjustment, and (c) the difference between (b) and (a).

Citation: Journal of Applied Meteorology and Climatology 48, 10; 10.1175/2009JAMC2191.1

Save