Time Series

Time series are data from a fixed position and are recorded continuously at equally spaced time intervals.  They usually result from moored instrumentation, tide gauges, meteorological stations, etc..  Time series data, available from this site, may have been obtained from SAIC deployed instrumentation, or from other academic, government and industry sources.  In all cases, the data has been inspected by oceanographers and subject to our QA/QC procedures given below.  The approach is to generate time series files that can be used in standard time series analysis and statistics software without further clean-up.  Thus, the files always contain continuous series with no gaps or missing data fill flags, and they may already have been filtered to remove high-frequency noise.  Because of these procedures, data that originate from non-SAIC sources may have slight differences when compared to the same time series obtained from the originator (e.g. NDBC or NWS for meteorological data).

File Naming Conventions

SAIC time series files always contain two components.  They may be the u and v-components of a velocity (current or wind) or be two scalers such as temperature and salinity.  For the purpose of making these data available through NVODS, the files are translated into the netcdf format.  In the netcdf files, velocity vector data (currents or wind) are represented by orthogonal components that are named U_cmpt and V_cmpt, respectively.  Scaler variables have a unique, if cryptic name, such as T_var.  The netcdf attribute "longname" gives the full definition of the variable.  If only one scaler series is a available (e.g. the instrument only measures temperature), the second component just duplicates the first. However, in this cae only one time series array will be given in the netcdf file.  Each instrument in the SAIC database is assigned a unique 8 character ID that is usually references the mooring or platform name (e.g. A1 or 42003) and, for a mooring, its position.  Positions are usually numbered down from the surface.  For example, the ID:

DSC-B2-4

is used for the 4th instrument down on mooring B2 deployed in the Desoto Canyon Eddy Intrusion Study.  All files that use data from this instrument will reference this ID.  Files will usually have a shortened form of the ID as the first 5 or 6 characters of the filename that may contain a deployment number.  Filenames that contain data from meteorological buoys, coastal meteorological stations, and coastal tide gauges use 6 characters plus a 2 character suffix begining with N, W, and E, respectively.  Filenames that contain data from moorings can begin with any other alphabetic character ( consistent for each experiment or program), use 5 characters plus a 3 character suffix.

Filename Suffixes for Meteorological Data

2 Char Suffix	Type of Data
.D	Wind Velocity Components
RD	Rotated Wind Velocity Components
TD	Air Temperature/Atmospheric Pressure
PD	Atmospheric Pressure
SG	SST/Wind Speed Gusts
HD	Significant Wave Height/Mean Wave Period
DP	Dominant Wave Period

Filename Suffixes for Mooring Data

3 Char Suffix	Type of Data
C.D	Current Velocity Components
R.D	Rotated Current Velocity Components
T.D	Temperature/Salinity or Pressure
P.D	Pressure (Depth)
K.D	Conductivity/Salinity
M.D	Turbidity
H.D	Significant Wave Height/Mean or Dominant Period
W.D	Wave Direction/Direction Spread

Notes

• If the final D in the above suffixes is replaced by a 3 or 4, then this denotes filtering by 3-hour low pass (3HLP) or 40-hour low pass (40HLP) Lanczos kernels.  3HLP filtering removes high frequency noise and is similar to taking hourly means of the original data.  40HLP filtering removes tidal and inertial period fluctuations from the data in order to view the underlying long-period variability.  3HLP and 40HLP time series have time steps (delta-t) of 1 and 6 hours, respectively.
• Instruments that generate data at multiple depths such as Acoustic Doppler Current Profilers (ADCP) contain all their data of single type (e.g. velocity) in a single file.  Depth levels are numbered down from the surface irrespective of whether the deployed instrument was directed up or down.  On occasion, the measurements made at greatest distance from the head are incomplete when compared to other depth levels.  These incomplete measurements are assigned a different file in keeping with the philosophy that the data files should have no missing data flags.  The descriptions of these multiple depth files contain the word "profile".
• If the wind velocity components are resolved into speed and direction then the direction is towards where the wind is blowing rather than the meteorological convention of from.  This is consistent with the (u,v) components being components of a horizontal vector.
• Similarly, wave direction is also defined as the direction towards the waves are propagating.  The types of suffix (e.g. HD versus H.D) indicate wave measurements from a surface buoy (e.g. a waverider) or from a subsurface fixed platform (e.g. a bottom mounted ADCP), respectively.
• If the vector (current or wind) is denoted as rotated, then the rotation angle is the clockwise direction of the v-component from true north.  In the majority of the data, this direction is chosen to be the general trend of the isobaths at the mooring site such that the u-component is directed towards deeper water.  If there is no rotation, then the u and v-components are true east and north, respectively.  The sketch illustrates the relationships between the east/north and rotated coodinate axes.

Netcdf Files are used for SAIC's data holdings made available through NVODS.  The Netcdf filenames are derived from the SAIC filenames by converting to lower case, replacing any "." (periods) by underscores, and appending ".nc".  SAIC's Netcdf files follow the community COARDS conventions.

QA/QC Procedures

Time series obtained from instruments and other sources are inspected for obvious faulty values and gaps.  Faulty values are removed and any short gaps (1 or 2 points or less than ~ 1 hour in time) are filled by linear interpolation.  Longer gaps, usually caused by the retrieval and redeployment of moorings, are filled by specialized procedures that preserve the spectral and vertical spatial coherence of the signal in a manner consistent with the dataset.  These type of gaps range from a few hours to less than 2 days.  If gaps are longer than ~ 2 days, the data are split into two or more seperate time series.  The results are always checked by professional oceanographers familiar with the instrumentation and/or data sources.