Standard meteorological data from Barrow, Alaska, is described below. Data was obtained from the National Climate Data Center in November 2004, request was for all hourly observations for the period of record. The Barrow station overview and history may be seen on the web at http://www.ncdc.noaa.gov/oa/climate/stationlocator.html. Also available online are daily surface summaries and Cooperative Summary Of The Month for part of the record. Apparently parts of the record do extend back to at least the 1920s, but apparently not for the full hourly surface observations. This document describes data in files PABR.dat and PABRhourlyINTERPOLATED.dat Data begins in early January 1945 and ends in late October 2004. The following fields were extracted from the original dataset: Date & Time (converted from Alaska Standard Time to GMT (= AST+9 hrs)) Wind Direction (degrees) Wind Speed (m/s, converted from mph) Sky Cover (eighths, converted from decoded fields) Air Temperature (degrees C, converted from degrees F) Dew Point Temperature (degrees C, converted from degrees F) Sea Level Pressure (millibars) Altimeter Setting (millibars, converted from inches of Hg) Station Pressure (millibars) Derived fields are: Uo = East-West component of wind vector (m/s) Vo = North-South component of wind vector (m/s) Notes: ====== Missing data is denoted by "NaN". Note that data quality typically suffers in long-term datasets in the 1970s with the advent of automated weather stations. Observer recording of data before this is generally of good quality. Automated stations of the 1990s typically have better data quality. The progression from manual to automated observations must be kept in mind when assessing subtle long-term trends in the datasets. All fields other than the Sky Cover were subjected to a hand-editing process that removed spikes and obviously contaminated data (ie, stuck sensor readings). In general, the data quality from NCDC for this dataset was very good and only minimal hand-editing was needed or applied. All three pressure fields were extracted because there were some times at which only one or another field was reported. The original raw data files report wind direction in the meteorological convention: wind direction denotes the direction from which the wind blows. The files described herein report wind direction in the oceanographic convention: wind direction has been converted to denote the direction to which the wind blows. Likewise the Uo and Vo components are in oceanographic convention: Positive values of Uo and Vo denote wind blowing to the east and north. The sky cover field has been converted in the following fashion, nominally denoting the fraction of sky covered by clouds, in eighths. This field is highly-nonlinear and has been subjected to no other screening. Note that the method of reporting has changed over time: earlier parts of the record have no OBS or POB observations. OVC = Overcast = 8 BKN = Broken = 6 SCT = Scattered = 2.5 CLR = Clear = 0 OBS = Obscured = -1 POB = Partially Obscured = -2 The PABRhourlyINTERPOLATED file contains data that has been linearly interpolated from the original observation time periods to the 0th minute of each hour for the entire record. A cubic spline was not used becase many observations were near, but not on, the top of the hour, and because of this, caused some unphysical overshoot of various parameters at the 0 minute crossing. The effect of a linear interpolation upon the data fields can be estimated to first order. In the interpolated dataset the standard deviation of all observations decreases by 0.05 m/s for the wind speeds, increases by 0.13 and 0.20 deg C for the air temperatures and dew points, and decreases by 0.02 mb for the sea level pressure. Interpolated data points for each parameter were retained only for those within three hours of an actual observation. (This allows for a complete day of interpolated datapoints from observations taken every six hours). The Sky Cover field was not interpolated because of the discrete nature of the dataset. For this field, I stepped through the dataset, one hour at a time, and selected the closest observation within 3 hours of the time of interest. The wind direction and wind speed fields were recomputed from the interpolated component fields Uo and Vo.