Level 4 data (velocity profilers): Difference between revisions

From Atomix
m Corrected units for REGRESSION_COEFF_A3
Updated attributes to represent those that are being used in the test datasets
Line 132: Line 132:
| <blockquote>''This group includes the results associated with fitting the structure function to data in level 3. The results for each beam, along with quality indicators and errors are provided. A final estimate for the turbulent kinetic energy dissipation is also provided.''</blockquote>
| <blockquote>''This group includes the results associated with fitting the structure function to data in level 3. The results for each beam, along with quality indicators and errors are provided. A final estimate for the turbulent kinetic energy dissipation is also provided.''</blockquote>
|-
|-
| dll_fitting_method
| regression_model
| statistical technique used for fitting the spectra.
| model regressed against the data to determine <math>\varepsilon</math>
| Examples include:
| Examples include:
linear regression
standard model, i.e. <math>D_{ll} = a_0 + a_1 (\delta r)^{2/3}</math> <br /><br />
modified model, i.e. <math>D_{ll} = a_0 + a_1 (\delta r)^{2/3}+a_3((\delta r)^{2/3})^3 </math>
|-
| regression_method
| method used to regress the model against the data
| Examples include:
Least-squares using evaluated separation distances to r_max.
|-
|-
| rsquared_method
| rsquared_method

Revision as of 14:08, 8 July 2022

Page status: Ready for review
Author(s): Cynthia
Reviewer(s):
Comment:

This will dictate the data required at the final processing level, where we store the estimated dissipation estimates [math]\displaystyle{ \varepsilon }[/math] along with quality metrics.

Only a few attributes for each variable are listed since the page's purpose is to describe the information layout within each NetCDF file. Please refer to the complete list for the additional attributes related to each variable (e.g., units, bounds, cell_methods).

Dimensions

Short name Standard name Dimensions Comments
TIME time TIME Units in Days since reference time specified in variable attribute. Provide bounds attribute to designate the variable containing the limits of each segment (see CF-compliant example).
Z_DIST distance_from_sensor_along_vertical Z_DIST bin centre distance (in meters) from the transducer along the instrument's vertical axis
N_BEAM unique_identifier_for_each_beam N_BEAM Array of 1 to number of beams (3 to 5 typically)

Variables


Short name Standard name Dimensions Comment
EPSI specific_turbulent_kinetic_energy_dissipation

_in_sea_water

TIME, Z_DIST, N_BEAM Dissipation rate of turbulent kinetic energy per unit mass of water [W/kg] estimated from individual beams.
EPSI_FINAL specific_turbulent_kinetic_energy_dissipation

_in_sea_water

TIME, Z_DIST Final (beam-averaged) dissipation rate of turbulent kinetic energy per unit mass of water [W/kg]. best to state cell_methods attribute to indicate what was averaged e.g., "cell_methods= N_BEAM:mean" for averages across beams.
C2 constant_used_in_the_second_order_structure_function Scalar [1 value] This constant appears when estimating the dissipation rate of turbulent kinetic energy from the regression coefficients (see Compute structure functions and dissipation estimates).
Quality-control metrics
EPSI_FLAGS specific_turbulent_kinetic_energy_dissipation_

in_sea_water_status_flag

TIME, Z_DIST, N_BEAM For details see Velocity Profiler data flags
EPSI_CI_HIGH specific_turbulent_kinetic_energy_dissipation_

in_sea_water_high_confidence_limit

TIME, Z_DIST, N_BEAM Computed from the confidence interval of the regression slope as EPSI_CI_HIGH = (SLOPE_CI_HIGH/C2)^(3/2) To be verified.
EPSI_CI_LOW specific_turbulent_kinetic_energy_dissipation_

in_sea_water_low_confidence_limit

TIME, Z_DIST, N_BEAM Computed from the confidence interval of the regression slope as EPSI_CI_HIGH = (SLOPE_CI_HIGH/C2)^(3/2) To be verified.
R_MAX maximum_separation_distance_for_DLL_regression TIME, Z_DIST, N_BEAM maximum R_DEL separation distance [m] used when computing the regression of DLL vs r[math]\displaystyle{ ^{2/3} }[/math]
REGRESSION_COEFF_A0 structure_function_regression_intercept TIME, Z_DIST, N_BEAM Constant term in regression, i.e. [math]\displaystyle{ A_0 }[/math] in [math]\displaystyle{ D_{LL} = A_1 r^{2/3} + A_0 }[/math]. Units are m[math]\displaystyle{ ^2 }[/math]s[math]\displaystyle{ ^{-2} }[/math] and value is proportional to instrument noise. (see Compute structure functions and dissipation estimates)
REGRESSION_COEFF_A1 structure_function_regression_coefficient for_r^2/3 TIME, Z_DIST, N_BEAM Linear term in regression, i.e. [math]\displaystyle{ A_1 }[/math] in [math]\displaystyle{ D_{LL} = A_1 r^{2/3} + A_0 }[/math]. Units are m[math]\displaystyle{ ^{4/3} }[/math]s[math]\displaystyle{ ^{-2} }[/math]
REGRESSION_R2 regression_goodness_of_fit_adjusted_for_number_of_terms TIME, Z_DIST, N_BEAM [math]\displaystyle{ R^2 }[/math] computed from the regression of [math]\displaystyle{ D_{LL} }[/math] vs [math]\displaystyle{ r^{2/3} }[/math]. Specific method should be described in group attributes.
REGRESSION_N structure_function_regression_number_of_observations TIME, Z_DIST, N_BEAM number of data points used in the regression of [math]\displaystyle{ D_{LL} }[/math] vs [math]\displaystyle{ r^{2/3} }[/math]
Optional variables
REGRESSION_COEFF_A3 structure_function_regression_coefficient for_(r^2/3)^3 TIME, Z_DIST, N_BEAM Linear term in regression for modified method, i.e. [math]\displaystyle{ A_1 }[/math] in [math]\displaystyle{ D_{LL} = A_3 r^2 + A_1 r^{2/3} + A_0 }[/math]. Units are s[math]\displaystyle{ ^{-2} }[/math]

Group attributes (metadata)

This section describes attributes that may provide additional information about how the data was processed and manipulated at this stage.


Attribute name Purpose Suggested content
processing_level Boilerplate about the content of the NetCDF group.

This group includes the results associated with fitting the structure function to data in level 3. The results for each beam, along with quality indicators and errors are provided. A final estimate for the turbulent kinetic energy dissipation is also provided.

regression_model model regressed against the data to determine [math]\displaystyle{ \varepsilon }[/math] Examples include:

standard model, i.e. [math]\displaystyle{ D_{ll} = a_0 + a_1 (\delta r)^{2/3} }[/math]

modified model, i.e. [math]\displaystyle{ D_{ll} = a_0 + a_1 (\delta r)^{2/3}+a_3((\delta r)^{2/3})^3 }[/math]

regression_method method used to regress the model against the data Examples include:

Least-squares using evaluated separation distances to r_max.

rsquared_method method used to calculate the goodness of fit Examples include:

????

comment (optional) Any additional information pertinent to other users who test their algorithms against the file.

Return to Level 3 structure function

Go back to the beginning Dataset requirements for ADCP structure function