AGW Observer

Observations of anthropogenic global warming

Papers on time of observation bias

Posted by Ari Jokimäki on August 1, 2012

This is a list of the time of observation bias in temperature measurements. The list is not complete, and will most likely be updated in future in order to make it more thorough and more representative.

Note that list of papers on global surface temperature contains papers relevant to this list also.

UPDATE (August 1, 2012): Kreil (1854) and Kreil (1854) added. Thanks to Victor Venema for pointing them out.

Bias in Minimum Temperature Introduced by a Redefinition of the Climatological Day at the Canadian Synoptic Stations – Vincent et al. (2009) “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.” Vincent, Lucie A., Ewa J. Milewska, Ron Hopkinson, Leslie Malone, 2009: Bias in Minimum Temperature Introduced by a Redefinition of the Climatological Day at the Canadian Synoptic Stations. J. Appl. Meteor. Climatol., 48, 2160–2168. doi: http://dx.doi.org/10.1175/2009JAMC2191.1. [FULL TEXT]

A method to infer time of observation at US Cooperative Observer Network stations using model analyses – Belcher & DeGaetano (2005) “A method to estimate the time of observation employed at US Cooperative Observer Network stations has been developed using rapid update cycle model analyses. This method uses the day-to-day variability in model temperature biases to estimate observation schedules on a time scale of weeks, making it ideal for use in ‘real-time’ applications. Observation time estimates from a two-category system (morning and ‘non-morning’) and three-category system (morning, afternoon and midnight) were both evaluated. The performance of the two-category system was compared with existing techniques that employ this system on monthly time scales. The results were comparable, showing dependence on season and climatological characteristics, but reveal an ability to reach high levels of accuracy (>90% of stations have observation schedules correctly estimated) over similar time periods (10–50 days). To our knowledge, the evaluation of three-category estimation performance for the time scales investigated has not been documented. Accuracy remained high for morning and midnight stations (>90%), and decreased for stations with afternoon observation schedules (85–65%). Additionally, the three-category estimation technique was extended to four categories in order to identify observers who shift temperature records temporally. The accuracy of detecting shifted records within the context of the four-category estimation technique was comparable to the performance of the three-category system, with shifted observations correctly identified more than 75% of the time in most cases.” Brian N. Belcher, Arthur T. DeGaetano, International Journal of Climatology, Volume 25, Issue 9, pages 1237–1251, July 2005, DOI: 10.1002/joc.1183. [FULL TEXT]

An evaluation of the time of observation bias adjustment in the U.S. Historical Climatology Network – Vose et al. (2003) “The U.S. Historical Climatology Network (HCN) database contains statistical adjustments that address historical changes in observation time at each observing station in the network. A paper in 2002 suggested that these adjustments cause HCN temperature trends to be “spuriously” warm relative to other datasets for the United States. To test this hypothesis, this paper evaluates the reliability of these “time of observation bias” adjustments in HCN. The results indicate that HCN station history information is reasonably complete and that the bias adjustment models have low residual errors. In short, the time of observation bias adjustments in HCN appear to be robust.” 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(20), 2046, doi:10.1029/2003GL018111.

A Method for Operational Detection of Daily Observation-Time Changes – Belcher & DeGaetano (2003) “A method to detect observation-time changes within weeks to months of their occurrence was developed using hourly observations from a set of 32 first-order stations across the United States. The procedure operationally requires only daily maximum and minimum temperatures. Development of the detection procedure depends on the utilization of nonclimatic biases that are artificially introduced into simulated daily temperature datasets constructed from hourly data at each of the 32 development stations. These biases are quantified by using interdiurnal temperature differences, which are used as the basis of observation-time-change detection. The procedure is tested on simulated daily temperature datasets constructed from hourly data at nine independent first-order stations and is applied to actual daily temperature observations from a sample of stations that make up the Cooperative Observer Network. The probability of detection is largest over the Midwest and High Plains regions of the United States and is smallest for maritime stations in the southwestern and southeastern regions of the United States. False-alarm occurrence is highest over the southern United States, particularly in the southeastern region and southern California. Overall detection performance is improved in all regions by incorporating more data into the detection tests, but the median amount of time between the occurrence of an observation-time change and its detection also increases as a result.” Belcher, Brian N., Arthur T. DeGaetano, 2003: A Method for Operational Detection of Daily Observation-Time Changes. J. Appl. Meteor., 42, 1823–1836. doi: http://dx.doi.org/10.1175/1520-0450(2003)0422.0.CO;2. [FULL TEXT]

Observation-Time-Dependent Biases and Departures for Daily Minimum and Maximum Air Temperatures – Janis (2002) “Non-calendar-day observations of 24-h minimum and maximum air temperatures can be considerably different from calendar-day or midnight observations. This paper examines the influence of time-of-observation on 24-h temperature observations. Diurnal minimum and maximum temperatures measured at common observation times (0700 and 1700 LST) are compared with minimum and maximum temperatures measured at midnight. The principal methods make use of hourly temperature observations, sampled over 24-h moving windows, to approximate once-daily observations. Surprisingly, non-calendar-day observations are similar to calendar-day observations on a majority of days. When differences do occur, however, they can be large and of either sign. Differences between 1700 LST observations and midnight observations are typically smaller than those arising from 0700 LST observations. Daily differences can be grouped by temperature extrema recorded on the incorrect day (a bookkeeping problem) or temperature extrema recorded on two successive days (bias). Bias scenarios arise when very cold mornings or very warm afternoons influence the temperature measured on successive days. Locations or seasons with the least day-to-day temperature variability often display the least time-of-daily-observation influence on temperature. Determining those days on which large departures and biases are likely to occur is possible by measuring day-to-day temperature persistence. First-order differences of daily time series may be used explicitly in adjustment procedures for morning observations of maximum temperature. Otherwise, first-order differences may be used to determine those days on which large observation-time differences are likely or those days on which observation-time dependencies are trivial.” Janis, Michael J., 2002: Observation-Time-Dependent Biases and Departures for Daily Minimum and Maximum Air Temperatures. J. Appl. Meteor., 41, 588–603. doi: http://dx.doi.org/10.1175/1520-0450(2002)0412.0.CO;2. [FULL TEXT]

A Method to Infer Observation Time Based on Day-to-Day Temperature Variations – DeGaetano (1999) “A method to infer the observation time of a station at annual resolution is developed and tested at stations in the United States. The procedure is based on a tendency for the percentiles of the monthly distribution of positive day-to-day maximum temperature changes (i.e., day n + 1 > day n) to exceed the corresponding absolute percentiles of the distribution of negative day-to-day changes at afternoon stations. Similarly absolute percentiles of negative day-to-day minimum temperature change tend to exceed the corresponding positive interdiurnal changes at morning observation sites. Equal percentiles are generally found at stations that use a midnight observation hour. Based on annual and seasonal summations of these monthly percentile differences, discriminant functions are developed that are capable of differentiating between afternoon, morning, and midnight observation schedules. Across the majority of the United States observation time is correctly classified in over 90% of the station-years tested. Classification success is generally highest for morning and afternoon observations and somewhat lower for midnight observations. Although geographic biases in classification success are not apparent, the procedure’s ability to estimate observation time decreases considerably at stations where the average annual interdiurnal temperature range is less than 1.7°C. In the United States such stations are limited to coastal California, parts of Arizona, and extreme southern portions of Texas and Florida. Application of the procedure to a subset of U.S. climatic normals stations indicates the presence of errors in the corresponding observation time metadata file.” DeGaetano, Arthur T., 1999: A Method to Infer Observation Time Based on Day-to-Day Temperature Variations. J. Climate, 12, 3443–3456. doi: http://dx.doi.org/10.1175/1520-0442(1999)0122.0.CO;2. [FULL TEXT]

Standardization of weekly growingdegreedayaccumulationsbasedondifferencesin temperature observation time and method – DeGaetano & Knapp (1993) “Efforts to develop comparable growingdegreeday (GDD) accumulations across the US-Canadian border have revealed significant anomalies resulting from differencesin observing times among US stations and reporting practices between US and Canadian stations. Simulations using hourly temperature data indicate that the period required to reach a given GDD threshold value during a growing season often varies by 2 weeks or more solely because of observation time differences. Previous work on methods to adjust such biases has concentrated on seasonal totals and long-term averages. In this study, empirical methods are developed to standardize weekly GDD accumulations to a common observing time. Hourly data for a 10 year period from five northeastern United States stations are used in the development of the adjustment procedure. However, only daily maximum and minimum temperatures are needed to implement the adjustment scheme. Different weekly adjustment factors are used for each month from March through November. In addition, the weekly adjustment factors are decreased in magnitude during individual spring and autumn weeks in which fewer than 25 degreedays (13.9°C GDD) accumulate. Correction factors are also reduced during summer weeks which exhibit minimal day-to-day variation in temperature. An additional adjustment is applied to observations taken in the late afternoon and evening to ensure that the maximum temperature observed at these stations has occurred on the same day as that reported by morning observing sites. Validation trials using data from six independent stations indicate that average weekly differences of as much as 10 GDD (5.6°C GDD) are reduced to less than 1 GDD (0.6°C GDD). After applying the standardization procedures, GDD values basedon different observation times and the Canadian observation practice accumulate at similar rates.” Arthur T. DeGaetano, Warren W. Knapp, Agricultural and Forest Meteorology, Volume 66, Issues 1–2, September 1993, Pages 1–19, http://dx.doi.org/10.1016/0168-1923(93)90079-W.

A Model to Estimate the Time of Observation Bias Associated with Monthly Mean Maximum, Minimum and Mean Temperatures for the United States – Karl et al. (1986) “Hourly data for 79 stations in the United States are used to develop an empirical model which can be used to estimate the time of observation bias associated with different observation schedules. The model is developed for both maximum and minimum monthly average temperature as well as monthly mean temperature. The model was tested on 28 independent stations, and the results were very good. Using seven years of hourly data the standard errors of estimate using the model were only moderately higher than the standard errors of estimate of the true time of observation bias. The physical characteristics of the model directly include a measure of mean monthly interdiurnal temperature differences, analemma information, and the effects of the daily temperature range due to solar forcing. A self-contained computer program has been developed which allows a user to estimate the time of observation bias anywhere in the contiguous United States without the costly exercise of accusing 24-hourly observations at first-order stations.” Karl, Thomas R., Claude N. Williams, Pamela J. Young, Wayne M. Wendland, 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, 145–160. doi: http://dx.doi.org/10.1175/1520-0450(1986)0252.0.CO;2. [FULL TEXT]

An Adjustment for the Effects of Observation Time on Mean Temperature and Degree-Day Computations – Byrd (1985) “Biases in mean temperatures due to differing times of daily maximum and minimum temperature observation cause problems in evaluation of temporal and spatial anomalies in temperature and derived degree day values. These biases were examined using six years (1973–78) of digitized hourly temperature data taken at Oneonta, New York. An annual mean temperature difference of 2.5°F is noted between means computed with the 0600 LST and 1500 IST observation times, with individual monthly differences as high as 4.4°F. Maximum seasonal degree day biases were 743 heating degree days (HDD) (10.2%), 169 cooling degree days (CDD) (43.3%), and 299 growing degree days (GDD) (14.3%). A modified version of the Blackburn method for adjusting mean temperature data for observation time bias is presented. The modified method involves adjusting data to a “true” mean obtained by averaging all hourly temperature values for the 24-hour period ending at midnight, rather than adjusting to the midnight standard observational mean obtained by averaging the maximum and minimum values over the same period. The adjustments are applied to mean temperatures from stations with different observation times in the region around Oneonta, resulting in spatial analysis fields which are believed to be more representative than those using the published data. This suggests that application of such an adjustment scheme results in a more homogeneous climatological data set.” Byrd, Gregory P., 1985: An Adjustment for the Effects of Observation Time on Mean Temperature and Degree-Day Computations. J. Climate Appl. Meteor., 24, 869–874. doi: http://dx.doi.org/10.1175/1520-0450(1985)0242.0.CO;2. [FULL TEXT]

A Practical Method of Correcting Monthly Average Temperature Biases Resulting from Differing Times of Observation – Blackburn (1983) “Biases in monthly average temperatures reported by cooperative weather observers arise from once-daily observations of maximum and minimum temperatures at times of day other than local midnight. A scheme for adjusting these reports to eliminate the biases and make them conform to the midnight-to-midnight reports of first-order weather stations is described. This scheme has been applied to climatological reports of the Washington, DC, cooperative network.” Blackburn, Thomas, 1983: A Practical Method of Correcting Monthly Average Temperature Biases Resulting from Differing Times of Observation. J. Climate Appl. Meteor., 22, 328–330. doi: http://dx.doi.org/10.1175/1520-0450(1983)0222.0.CO;2. [FULL TEXT]

Time of Observation Temperature Bias and “Climatic Change” – Schaal & Dale (1977) “Historical changes in time of once daily maximum and minimum temperature observations at cooperative climatological stations from 1905 to 1975 have introduced a systematic bias in mean temperatures. Unless corrected, this bias may be interpreted incorrectly as climatic “cooling” and may also affect the assessment of agricultural production potential and fossil fuel needs. Maximum and minimum temperature data for two years from the National Weather Service station at Indianapolis International Airport were used to evaluate the differences between mean temperatures obtained by terminating the 24 h period at the midnight observation and the mean temperatures obtained by terminating the 24 h period at 0700 and 1900 hours, typical observation times for AM and PM observing stations. The greatest mean temperature bias occurs in March when a 1900 observation day yields a monthly mean temperature 1.3°F above a midnight observation, and a 0700 observational day gives a −1.3°F bias. Since the number of AM observing stations in Indiana have increased from 10% of the total number of temperature stations in 1925 to 55% in 1975, the March mean temperature shows a decrease of 1.2°F in the last 40 years, solely because of the change in substation observational times. Unless the time of observation bias is considered, the mixture of AM and PM observations complicates interpretation of areal temperature anomaly patterns. This bias is accumulated in monthly, seasonal or annual values of the mean temperature-derived variables-heating degree days, cooling degree days and growing degree days—and may provide misleading information for applications in industry and agriculture.” Schaal, Lawrence A., Robert F. Dale, 1977: Time of Observation Temperature Bias and “Climatic Change”. J. Appl. Meteor., 16, 215–222. doi: http://dx.doi.org/10.1175/1520-0450(1977)0162.0.CO;2. [FULL TEXT]

Effect of Observation Time on Mean Temperature Estimation – Baker et al. (1975) “The increased interest and application of heating degree days (HDD) and growing degree days (GDD) prompted this study into the effect of different observation times upon the mean daily temperature. The study was based upon three years of hourly air temperatures measured at St. Paul. These data were used to calculate 1) a true daily mean, 2) a mean of the maximum and minimum between successive midnights as observed at first order stations, and 3) a mean of the maximum and minimum observed at all other hours of the day to simulate cooperative station means. Comparisons of the annual and monthly mean temperatures showed deviations can be of such magnitude as to discourage comparison of station temperatures and temperature-derived quantities such as HDD and GDD unless observation times are the same or corrections are applied.” Baker, Donald G., 1975: Effect of Observation Time on Mean Temperature Estimation. J. Appl. Meteor., 14, 471–476. doi: http://dx.doi.org/10.1175/1520-0450(1975)0142.0.CO;2. [FULL TEXT]

Temperature adjustments for discrepancies due to time of observation – Weaver & Miller (1970) Doesn’t seem to be available online in any form. Weaver, C. R., and M. E. Miller, 1970, Type-written communication directed to Director, National Climatic Center, 7 pp plus attachments, dated 23 November.

Effects of changing observation time on mean temperature – Mitchell (1958) Doesn’t seem to be available online in any form. Mitchell, J. M. Jr., 1958, Bull. Amer. Meteor. Soc., 39, 83-89.

The effect of time of observation on mean temperature – Rumbaugh (1934) No abstract. RUMBAUGH, W. F., 1934: THE EFFECT OF TIME OF OBSERVATION ON MEAN TEMPERATURE1. Mon. Wea. Rev., 62, 375–376. doi: http://dx.doi.org/10.1175/1520-0493(1934)622.0.CO;2. [FULL TEXT]

Report on the temperatures and vapor tensions of the United States reduced to a homogeneous system of 24 hourly observations for the 33-year interval, 1873-1905 – Bigelow (1909) No abstract. Bigelow, F. H., 1909, Bull S.W.B. No. 408, U. S. Weather Bureau, Washington D. C.. [FULL TEXT]

On the difference produced in the mean temperature derived from daily maximum and minimum readings, as depending on the time at which the thermometers are read – Ellis (1890) No abstract. William Ellis F.R.A.S., Quarterly Journal of the Royal Meteorological Society, Volume 16, Issue 76, pages 213–220, October 1890, DOI: 10.1002/qj.4970167605.

Mehrjährige Beobachtungen in Wien vom Jahre 1775 bis 1850 – Kreil (1854) Doesn’t seem to be available online in any form. Kreil K, 1854. Mehrjährige Beobachtungen in Wien vom Jahre 1775 bis 1850. Jahrbücher der k.k. Central-Anstalt für Meteorologie und Erdmagnetismus. I. Band – Jg 1848 und 1849, 35-74.

Mehrjährige Beobachtungen in Mailand vom Jahre 1763 bis 1850 – Kreil (1854) Doesn’t seem to be available online in any form. Kreil K, 1854. Mehrjährige Beobachtungen in Mailand vom Jahre 1763 bis 1850. Jahrbücher der k.k. Central-Anstalt für Meteorologie und Erdmagnetismus. I. Band – Jg 1848 und 1849, 75-114..

5 Responses to “Papers on time of observation bias”

  1. Thank you Ari Jokimäki for this resource; somehow the TOB is suddenly a much talked about topic.🙂

    For those of you that can read German, there are two studies from 1848 on the time of observation bias. Well before the climate conspiracy, eh change.

    Kreil K, 1854a. Mehrjährige Beobachtungen in Wien vom Jahre 1775 bis 1850. Jahrbücher der k.k. Central-Anstalt für Meteorologie und Erdmagnetismus. I. Band – Jg 1848 und 1849, 35-74.

    Kreil K, 1854b. Mehrjährige Beobachtungen in Mailand vom Jahre 1763 bis 1850. Jahrbücher der k.k. Central-Anstalt für Meteorologie und Erdmagnetismus. I. Band – Jg 1848 und 1849, 75-114.

  2. Ari Jokimäki said

    Thank you. I have added the two references to the list above. Unfortunately they don’t seem to be available online. Someone should do world a favour and translate all the interesting scientific articles written in German to English. So, we have first articles on this in 1854, can anyone find even earlier than that?🙂

  3. Just trying to provoke a race to the older publication. These two are probably not especially interesting, except for their age.

  4. Just trying to provoke a race to the oldest publication.

    Because the time of observation bias is now such a hot topic, I have written a short introduction. I hope it is useful for the discussions.

  5. […] see Ari's page on time of observations papers. Printable Version  |  Link to this page | Repost this Article […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

 
%d bloggers like this: