Files with the ext, wts, var, and seas qualifier exist for all sites described in a GLOBALVIEW product. Files with the diu and nta-dta qualifier exist for a subset of extended records derived from high-resolution measurements where the diurnal cycle is a dominant feature of the observations. Files with the dta-vg and nta-vg qualifier exist for a subset of extended records derived from high-resolution measurements from towers insrumented at multiple levels. Files with the tod qualifier accompany a subset of extended records derived from discrete measurements where sample collection times have been made available.

All file types (except for the reference MBL matrix) have 19 lines of descriptive information that include

Following the descriptive information detailed above, the format of each type of file is as follows:

There are no blank fields in any column. Missing values are denoted with a standard default value, -999.999.

Following the descriptive information detailed above, the four (4) columns in the extended ext files are:

1. [UTC] "Weekly" synchronized time steps in Universal Coordinated Time (UTC) as decimal dates, i.e., year plus fraction of the year. Each year has 48 "weekly" steps. "Synchronized" means that the synchronization period and the time steps are the same for all extended record files.
2. [S(t)] Smoothed values extracted from a curve fitted to measurement data that have been selected for conditions where the sampled air is thought to be representative of large well-mixed air parcels. Internal and external gaps in the measurement record are denoted as default values.
3. [REF(t)] The latitude reference time-series, based on marine boundary layer sites, constructed at the sine (latitude) of the measurement site. The latitude reference is defined at all time steps.
4. [diff] The difference climatology describes how the site differs from marine boundary layer (MBL) sites that are nearby in latitude. The difference climatology is defined at all time steps.

The length of the files depends on the number of years in the synchronization period.

USING EXTENDED RECORDS

CAVEATS

Any method used to fill spatial and temporal gaps in observational records is forced to make assumptions creating uncertainty in the resulting data product. Each extended record included in GLOBALVIEW has a corresponding weight file that suggests a relative significance for each value in the extended file. All smooth values (derived directly from the actual measurements) receive a relative weight (ranging from 2 to 10) that depends on sampling density and measurement variability. All filled values (interpolated and extrapolated) receive a fixed weight of 1.

We strongly recommend that users of this data product consider the weight files, which provide an estimate of the relative significance of each value in the extended record.

Following the descriptive information detailed above, the four (4) columns in the weight [wts] files are:

1. [UTC] Synchronization year where the number of years is determined by the synchronization period.
2. [rsd] Residual standard deviation (RSD) of the measurements about the smooth curve, S(t), with annual resolution. Years with fewer than six (6) measurements are assigned default values.
3. [#] The number of residuals per year used in the RSD determination.
4. [weight] Scaled weights determined using the relative weighting scheme described by Masarie and Tans, [1995]. Years where weights cannot be determined are assigned a default minimum weight of one (1).

The length of the files depends on the number of years in the synchronization period. The first row past the descriptive information specifies the residual standard deviation, number of residuals, and derived weight for all years, all observations.

USING WEIGHT FILES

A statistical summary of average atmospheric variability is provided for each measurement record. A residual distribution is determined by fitting a smooth curve, S(t), to the observations, C(t), and computing residuals C(t)-S(t). The residuals for all Januarys, Februarys, etc are aggregated and statistics are determined with monthly resolution. The aggregated monthly statistics include within month and year-to-year variability. Information pertaining to the diurnal cycle is not considered here. Following the descriptive information detailed above, the six (6) columns in the variability [var] files are:

1. [mo] Month (1-12) specification.
2. [stdev] Standard deviation of the residual distribution computed monthly for all years.
3. [50%ile] The 50th percentile or median of the residual distribution.
4. [16%ile] The 16th percentile of the residual distribution.
5. [84%ile] The 84th percentile of the residual distribution.
6. [N] The number of aggregated monthly residual values used to compute the monthly statistics.

A statistical summary of the average seasonal pattern is provided for each measurement record. Monthly means are computed from a detrended smooth fit, S(t)-T(t), to the observations. The monthly means for all Januarys, Februarys, etc. are aggregated and statistics are determined with monthly resolution. The standard deviation of each aggregated monthly mean value is a measure of the year-to-year variability in the monthly mean values. The standard error of the aggregated monthly mean value is an estimate of the uncertainty in the aggregated monthly mean value. Following the descriptive information detailed above, the five (5) columns in the seasonal [seas] files are:

1. [mo] Month (1-12) specification.
2. [mean] Mean of the aggregated detrended monthly means for all years.
3. [stdev] Standard deviation of the aggregated monthly mean distribution.
4. [stderr] Standard error of the aggregated monthly mean distribution.
5. [N] The number of monthly mean values used to compute the aggregated monthly statistics.

A statistical summary of average diurnal patterns by month compiled using data from all years is provided for each measurement record with hour or higher resolution and where the diurnal cycle is a dominant feature in the observations. The residual distribution is determined by subtracting an afternoon-hours daytime average (DTA) mixing or isotope ratio from every hourly-averaged observation for that day [see Data Comparability, Table 1 for details]. Please note that diurnal statistics are now presented in Local Time (LST). Following the descriptive information detailed above, the six (6) columns in the diurnal [diu] files are

1. [mo] Month (1-12) specification.
2. [hr in LST] Hour (0-23) specification.
3. [50%ile] The 50th percentile or median of the residual distribution computed monthly for all years.
4. [16%ile] The 16th percentile of the residual distribution.
5. [84%ile] The 84th percentile of the residual distribution.
6. [N] The number of residual values from all years used to compute the monthly statistics.

A statistical summary of average differences between nighttime and daytime averages at each intake height on a tower instrumented at multiple levels. For each day and at each level, we compute differences between a 4-hour nighttime average (NTA) [0-4 LST] and an afternoon-hours daytime average (DTA) [see Data Comparability, Table 1 for details]. We aggregate the daily differences by month and present a multi-year monthly summary. Following the descriptive information detailed above, the six (6) columns in the diurnal [nta-dta] files are

1. [mo] Month (1-12) specification.
2. [ht] Intake Height in meters above ground level (magl).
3. [50%ile] The 50th percentile or median of the residual distribution computed monthly for all years.
4. [16%ile] The 16th percentile of the residual distribution.
5. [84%ile] The 84th percentile of the residual distribution.
6. [N] The number of residual values from all years used to compute the monthly statistics.

A statistical summary of the daytime and nighttime vertical gradients from measurements at towers instrumented at multiple levels. For each day, we compute an afternoon-hours daytime average (DTA) [see Data Comparability, Table 1] at each level. We difference the day average at each level with the day average at the highest level. If a DTA value is missing for any intake height, the gradient is not determined for that day. We aggregate the daily daytime vertical gradients by month and present a multi-year monthly summary. The nighttime vertical gradient is computed in the same way but using hours 0-4 LST to compute the nighttime average (NTA). Following the descriptive information detailed above, the columns in the DTA [dta-vg] and NTA [nta-vg] vertical gradient files are

1. [mo] Month (1-12) specification.
2. [ht] Intake Height in meters above ground level (magl).
3. [50%ile] The 50th percentile or median of the residual distribution computed monthly for all years.
4. [16%ile] The 16th percentile of the residual distribution.
5. [84%ile] The 84th percentile of the residual distribution.
6. [N] The number of residual values from all years used to compute the monthly statistics.

A summary of sample collection times (in LST) for discrete measurement records where sampling times have been made available. Following the descriptive information detailed above, the three (3) columns in the collection time [tod] files are

1. [hr in LST] Sample collection hour (0-23) specification.
2. [fract] Fraction (of the total number of samples) collected within the hour.
3. [N] Number of samples collected within the hour.

The reference marine boundary layer matrix (ref_mbl_mtx.co2) contains mixing or isotope ratios as a function of time and sine of latitude and is a by-product of the data extension procedure (see Masarie and Tans, [1995]. Be aware that significant information contained in the actual data may be lost in this matrix. In addition, the reference MBL matrix itself may give an unrealistic impression of the comprehensiveness of global atmospheric measurements since it contains "values" at locations and times when no measurements exist.

There is a single header line in the matrix [mtx] file that specifies the format of the reference matrix.

Matrix format: FORMAT="(F12.6, 41(1X,F12.4))"

Following the single header line above, the 42 columns are

1. [UTC] "Weekly" synchronized time steps in Universal Coordinated Time (UTC) as decimal dates, i.e., year plus fraction of the year. Each year has 48 "weekly" steps. "Synchronized" means that the synchronization period and time steps in the matrix are identical to those in the extended record files.
2. [sine of latitude] There are 41 even intervals of 0.05 sine of latitude from 90°S to 90°N, i.e., column 2 represents a reference MBL value at -1.00 (90°S), column 3 at -0.95 (71.8°S), column 4 at -0.90 (64.2°S), and so on.

The number of rows in the matrix depends on the number of years in the synchronization period.