The first 96 records found in the 978/brh/brh97f01 file contained the same dates/times as the latter portion of the 967/brh/brh96702 file (overlap of records in the 1996/97 and 1997/98 files). Although the values in these files did not contain exact matches, a choice had to be made on which records to use. The data manager subjectively decided to use the 1996/97 records. This explains why the first record number (fline#) from the 978/brh/brh97f01 file is '978/brh/brh97f01.96', rather than 978/brh/brh97f01.0. If you have any questions, please contact the data manager. On August 17, 1999, the data manager added the 1998-1999 data to this dataset. In the process, she did some fine-tuning of the data, which included:
- removal of the relative humidity data from November 1995 through April 1996, as requested by Peter Doran. The values during this period were unquestionably low, so it was recommended that the values be marked as bad and not included in the data files.
- making the file layout consistent from year to year. This results in many null values being present for periods when certain values were not collected. However, it simplifies combining data from different years.
- making corrections so that the files containing data for a given year started on January 1 at 0:00 and ended on December 31 at 23:45. Up to this point, records for January 1 between 0:00 and 12:45 were appearing in the previous year's file.
- Fine-tuning of the comments fields so they were more concise and/or explanatory. For example, radiation comments were updated from "SwRadIn flagged with an 'S'" to "Negative value for SwRadIn zeroed out". The 'WDir Comments' field was renamed to 'Wind Comments'.
With the exception of removal of the relative humidity values described above, data values were not changed. On August 27, more fine-tuning of the data occurred. This included:
- addition of NetRad from Nov 24, 1995 @ 16:45 to Jan 23, 1996 @ 11:00.
- addition of SwRadOut and PAR from Nov 13, 1996 @ 15:15 to Jan 25, 1997 @ 10:15.
On July 3, 2001, Thomas Nylen adjusted the data. This included:
- Added 37 degrees (clockwise) to WDir results between 11/20/1995 16:00 and 11/27/1998 11:15
- Subjectively voided WDir and WDirStD between 3/1/1996 00:15: and 04/19/96 23:30
In May 2000, the 1999-2000 data was posted. In the process, Denise Steigerwald changed the layout of the files so that they were sorted by category (air temperatures, humidity, radiation, wind etc.) rather than by year. Files containing monthly and daily averages were generated and minor adjustments were made to the metadata. On May 16, 2000, relative humidity (RH) values were corrected for a systematic error in the measurement created by an instrument manufacturer error. All RH data with air temperatures below freezing were corrected using the vapor pressure over ice (rather than over water which was used initially). The error became quite large for very cold temperatures (the correction could grow to around 30%). The formula used for the correction was: RH_corr=RH_orig/((A*temp*temp)+(B*temp)+C)
where RH_corr = corrected relative humidity RH_orig = original relative humidity temp = air temperature A = 4.165E-5 B = 9.7E-3 C = 1.0 For example, with an air temperature of -2.3 and RH_orig of 50.6, the RH_corr will be 51.7. Records that showed corrected RH values greater than or equal to 100 were changed to 99.99 and flagged. In June 2000, summary files were added to the web containing daily and monthly statistics. These were generated by creating "views" of daily and monthly means, minimums, maximums and counts of the records represented in the Oracle database's meteorology tables. In May 2003, SwRadIn values between 12/30/1994 10:00:00 PM and 4/19/1996 11:30:00 PM were changed because it was determined the wrong multiplier was used. The values were corrected by dividing by 88.4 (mAmps m-2) and multiplying by 113.12 (W m-2). On June 4, 2004, the relative humidity (RH) values after January 26, 2000 @ 1545 were corrected for a systematic error in the measurement created by an instrument manufacturer error. All RH data with air temperatures below freezing were corrected using the vapor pressure over ice (rather than over water which was used initially). The error became quite large for very cold temperatures (the correction could grow to around 30%). The polynomials used for the correction is based on Lowe (1977). The original RH value is multiplied by the saturation vapor pressure over water and divided by the the saturation vapor pressure over ice RH3m_corr=[RH3m]*(6.107799961 + [AirT3m] * (0.4436518521 + [AirT3m] * (0.01428945805 + [AirT3m] * (0.0002650648471 + [AirT3m] *(0.000003031240396 + [AirT3m] * (0.00000002034080948 + 0.00000000006136820929 * [AirT3 m])))))) / (6.109177956 + [AirT3m] * (0.503469897 + [AirT3m] * (0.01886013408 + [AirT3m] * (0.0004176223716 + [AirT3m] * (0.00000582472028 + [AirT3m] * (0.00000004838803174 + 0.0000000001838826904 * [AirT3m])))))))
Records that showed corrected RH values greater than or equal to 100 were changed to 99.99 and flagged.
The ultrasonic ranger sensor that measures surface height change has been broken since November 2010, with no plans to replace as of July 2023.