Further information on preprocesssing#
Block diagram of the preprocessing step#
Focus on the preprocessing step (needs to be slightly modified)
Generalities#
The preprocessing step is made by a specific command line (inputs are listed in the page before) calling several scripts. One script is dedicated for the opening and reformatting of each of the three input data sources, with a xarray.Dataset as output, and after this, the three reformatted datasets are merged and the result is saved to a daily netCDF file after filling the corresponding metadata.
Outline of the preprocessing and different scripts involved#
The script ccres_disdrometer_processing.open_weather_netcdf is used to open and reformat input daily weather netCDF files from CloudNet, and outputs a xarray.Dataset :
– the main weather variables are renamed, and zonal and meridional components of wind are computed :
Variable Name |
Input Name |
Output Name |
|---|---|---|
Wind Speed |
wind_speed |
ws |
Wind Direction |
wind_direction |
wd |
Zonal Wind |
u |
|
Meridional Wind |
v |
|
Air Temperature |
air_temperature |
ta |
Relative Humidity |
relative_humidity |
hur |
Ground Pressure |
air_pressure |
ps |
Precipitation Rate |
rainfall_rate |
ams_pr |
Rain Accumulation |
rainfall_amount |
ams_cp |
– The variables are resampled on a perfect time vector, at a 1 minute sampling. at time T, we keep the first value found in the original dataset within the \([T-30s, T+30s]\) interval
The script ccres_disdrometer_processing.open_radar_netcdf is used to open and reformat input daily DCR netCDF files from Coudnet, and inputs a xarray.Dataset :
– Reflectivity (\(Zdcr\)) and Doppler velocity (\(DVdcr\)) data is extracted from the input file corresponding variables (Zh, v), from the ground to the maximum range specified in the configuration file (by default, from 0 to 2500m).
– The values are resampled on a perfect time vector, at a 1 minute sampling. At a range \(r\), we have :
\(Zdcr(T, r) = median(Zh([T-30s, T+30s], r)\)
\(DVdcr(T, r) = mean(v([T-30s, T+30s], r)\)
The script ccres_disdrometer_processing.open_disdro_netcdf is used to open and reformat input daily disdrometer netCDF files from CloudNet ant to compute forward-modeled reflectivity that will be compared to radar measurement, plus additional scattering-related variables and metrics characterizing the droplet size distribution at each timestep.
– The function read_parsivel_cloudnet_choice (changer le nom de la fonction pour quelque chose de plus clair) deals with the input data opening, renaming of disdrometer variables, resampling on a perfect time vector at 1 minute sampling, and add of useful metadata related to disdrometer data.
The key variable is the droplet size distribution (”psd”), giving the number of particles at each timestep, for each diameter and speed classes. Its dimension is (time, size_classes, speed_classes), with :
\[\begin{equation*} \begin{cases} size\_classes = 32, speed\_classes = 32 & \text{ for Parsivel²} \\ size\_classes = 22, speed\_classes = 20 & \text{ for Thies LPM} \end{cases} \end{equation*}\]The data opening slightly varies depending on the disdrometer model (Thies LPM or Parsivel²). The result is stored in a new Dataset, and a dimension called radar_frequencies is created, to be able to store the scattering-related variables at all the specified frequencies when they will be computed.
Remark: The frequencies at which the code is asked to compute the scattering-related variables are 10, 24, 35 and 94 GHz, which are the frequencies that can be found within the CCRES DCR network. These values are specified in the configuration file for each site, but it is possible to add/change these frequencies in the configuration if the user wants the code to compute forward-modeled reflectivity at other frequencies.
– The function reflectivity_model_multilambda_measmodV_hvfov computes the following values :
characterization of the droplet size distribution at time \(t\):
\(N_i(d, t)= \sum_{v \in speed\_classes}^{} psd(t,d,v), \forall d \in size\_classes\)
average measured droplet fall speed for each diameter class is computed and will be used to normalize several values later in the code : \(\forall d \in size\_classes : V_{measured}(d,t) = \frac{1}{N_i(d,t)} \sum_{v \in speed\_classes}^{} psd(t, d, v) * v\)
it is also possible to use modeled droplet fall speed (as a function of the diameter) for the normalisation of the values computed later in the code. We use the formula given by Gun and Kinzer (1949) : : \(V_{model}(d) = 9.40 \times \left( 1 - \exp\left( -1.57 \times 10^3 \times \left( d \times 10^{-3} \right)^{1.15} \right) \right), \forall d \in size\_classes\)
Other formulas could be implemented in the code as alternatives but it is not done yet. A field in the configuration file is ready for the chose of the formula used for modeling droplet fall speed as a function of diameter, when the implementation will be done.
moments of the droplet size distribution : \(M_i(t) = \frac{1}{V(d, t)FT} \sum_{d \in size\_classes}^{} N_i(d,t) \times d^{i}\), with \(V(d)\) either computed from DSD data or modeled, \(F\) the sampling surface of the disdrometer (available in configuration file) and \(T\) the time resolution of the resampled data i.e. \(T=60s\) in our case. Moments are calculated for both modeled and measured droplet fall speed
\(D_m(t) = \frac{M_4(t)}{M_3(t)}\), the mean volume weighted diameter in \(m\), computed in 2 variants (modeled / measured droplet fall speed)
\(re(t) = 0.5 \times \frac{M_3(t)}{M_2(t)}\), the effective radius in \(m\), computed in 2 variants
\(LWC(t) = 10^{6} \times \frac{\pi}{6} \times M_3(t), \text{in } g . cm^{-3}\) the liquid water content, computed in 2 variants
\(N_0^{*}(t) = \frac{4^{4}}{\pi \times 10^{6}} \times \frac{LWC(t)}{D_m^{4}(t)}\), the total number concentration in \(cm^{-3}\), computed in 2 variants
scattering-related variables at time t, with the help of a T-matrix code : hypothesis of Mie backscattering and non-spherical droplets/scatterers
Rayleigh reflectivity : \(Z_{e_{ray}}(t) = 10 \cdot log \ (\frac{1}{V(d, t)FT} \sum_{d \in size\_classes}^{} N_i(d) \times d^{6})\), computed in two variants
Mie reflectivity : reflectivity is calculated with the assumption of Mie backscattering, for spherical droplets. The values are computed in 8 variants : 4 frequencies \(\times\) 2 variants of computation for droplet fall speed.
The function compute_bscat_mie() in the script ccres_disdrometer_processing.ccres_disdrometer_processing.scattering is dedicated to computing backscattering coefficients under the assumption of Mie scattering. It takes as input the vector of disdrometer size classes, the wavelength at which to compute the values, and the complex index of refraction of water. It outputs a vector \(bscat_{mie}\) containing the backscattering coefficients for the different diameter classes. From there, we derive the equivalent reflectivity :
\(Z_{e_{mie}}(t,f) = 10 \cdot log \ [\frac{1}{V(d, t)FT} \sum_{d \in size\_classes}^{} (N_i(d,t) \times bscat_{mie}(d, f))]\)
T-matrix reflectivity and attenuation : reflectivity and attenuation are calculated with the assumption of Mie backscattering, for non-spherical droplets. The droplets are modeled as ellipses with an axis ratio that depends on the diameter. For the diameter-dependant axis ratio we use a \(4^{th}\) order polynomial fit suggested by (Andsager et al. 1999), based on a formula given by (Beard, Chuang 1987). The values are computed in 8 variants : 4 frequencies \(\times\) 2 variants of computation for droplet fall speed.
The function compute_bscat_tmatrix() in the script ccres_disdrometer_processing.ccres_disdrometer_processing.scattering is dedicated to computing backscattering and attenuation coefficients under the assumption of Mie scattering. It takes as input the vector of disdrometer size classes, a vector containing the axis ratios for the different size classes, the wavelength at which to compute the values, the complex index of refraction of water, and the geometry (horizontal or vertical pointing). It outputs two vectors : \(bscat_{tmatrix}\) contains the backscattering coefficients for the different diameter classes, and \(att_{tmatrix}\) contains the attenuation coefficients. From there, we derive the equivalent reflectivity and attenuation :
\(Z_{e_{tm}}(t,f) = 10 \cdot log \ [\frac{1}{V(d, t)FT} \sum_{d \in size\_classes}^{} (N_i(d,t) \times bscat_{tmatrix}(d, f))] \ \text{in} \frac{mm^6}{m^3}\)
\(att_{tm}(t,f) = 10 \cdot log \ [\frac{1}{V(d, t)FT} \sum_{d \in size\_classes}^{} (N_i(d,t) \times att_{tmatrix}(d, f))] \ \text{in} \frac{dB}{km}\)
These values are computed in 16 variants : 4 frequencies \(\times\) 2 variants of computation for droplet fall speed \(\times\) two geometries (horizontal/vertical pointing)
The reflectivity values that we will use for comparison with radar data are for vertical pointing and modeled droplet fall speed. The corresponding variable is named Zdlog_vfov_modv_tm in the output preprocessing netCDF file
Remark : when we derive Ze_tm we remove the biggest size classes to avoid artefacts from the calculation (5 classes are truncated by default in the code)
We can also derive the doppler velocity :
\(DV_{tm}(t,f) = \frac{\sum_{d \in size\_classes}^{} (N_i(d,t) \times bscat_{tmatrix}(d, f))}{\sum_{d \in size\_classes}^{} (\frac{N_i(d,t) \times bscat_{tmatrix}(d, f)}{V(d,t)})} \)
Summary of the assumptions for the forward-modeling of reflectivity values#
The disdrometer sampling surface F varies from an instrument to an other and is given in configuration file. All Parsivel² instruments have the same value \(F=0.0054m²\). For Thies disdrometers, each instrument has a scale factor called “AU” that has to be applied to a standard surface \(F_0 = 0.0046m²\) to get the sampling area : \(F = AU * F_0\)}. F is used to normalize reflectivity data, and one needs to be careful to specify the good value as a scale factor can introduce an offset in log-reflectivity.
The formula used to model droplet fall speed is described in (Gun and Kinzer, 1949) and is of common use. Other formulas are given in the litterature (Khvorostyanov and Curry 2002, Atlas and Ulbrich 1977, …), that are not implemented in the code yet, but the code is ready to host their implementation if necessary (including a field in config file for the method to use)
For T-matrix computations, droplets are assumed to be non-spherical : elliptic, with a shape descriebd by the ratio between major and minor axes. This ratio is a function of droplet diameter, the relationship used in the code is a 4th order polynomial fil suggested by (Andsager et al. 1999) based on a formula from (Beard, Chuang 1987).
Droplets are assumed to be liquid. In the code we use the complex index of refraction for liquid water n = 2.99645 + 1.54866j Je crois que c’est valid @ f=94GHz and T = 25°C. Should we use n(T,f) or at least minima n(f), how much would it impact the computed reflectivity values ? Some ways to compute the complex index of refraction of water as a function of the frequency are given in the literature ; a way to calculate the index it is given for example in (Ray 1972) with the use of the equations described by (Cole and Cole 1941) and tables of values and fits for the relationship n(f) are also provided in more recent papers. The code is ready to host a refraction index function of the frequency : in the configuration file, one must specify the couples (frequency, index) and the code uses them to compute backscattering coefficients.
The output preprocessing file is the result obtained after the three reformatted datasets (DCR, disdrometer, weather station when available) are concatenated. The command line ccres_disdrometer_processing preprocess, enables to launch this preprocessing with Paths to CloudNet data files, name of the configuration file and save path as inputs.
Detailed content of the preprocessing file#
Coordinates :
Coordinate Name |
Dimensions |
Size |
|---|---|---|
time |
time |
1440 |
size_classes |
size_classes |
32/22 according to disdrometer model |
speed_classes |
speed_classes |
32/20 according to disdrometer model |
radar_frequencies |
radar_frequencies |
4 by default |
range |
range |
depends on DCR range resolution : values between ground and 2500m |
Data variables :
Variable Name |
Long Name |
Unit |
|---|---|---|
ws |
Wind speed |
m s-1 |
wd |
Wind direction |
degree |
ta |
Air temperature |
Celsius |
hur |
Relative humidity |
% |
ps |
Air pressure |
hPa |
ams_pr |
Met station precipitation rate at 1m agl |
mm/hr |
ams_cp |
Rainfall amount |
mm |
u |
Zonal wind |
m/s |
v |
Meridional wind |
m/s |
ams_longitude |
Longitude of site |
degree_east |
ams_latitude |
Latitude of site |
degree_north |
ams_altitude |
Altitude of site |
m |
F |
Disdrometer sampling area |
m^2 |
disdro_pr |
Disdrometer-based precipitation rate |
mm/h |
disdro_cp |
Disdrometer-based cumulated precipitation since 00:00 UTC |
mm |
psd |
Number of droplets per diameter and fall speed bins |
#/bin/mn |
particles_count |
Number of particles in time interval |
1 |
size_classes_width |
Width of the diameter bins |
mm |
speed_classes_width |
Width of the speed bins |
m/s |
disdro_longitude |
longitude of disdrometer set-up |
degree_east |
disdro_latitude |
latitude of disdrometer set-up |
degree_north |
disdro_altitude |
altitude of disdrometer set-up |
m |
disdro_model |
Disdrometer model |
N/A |
time_resolution |
Time resolution of the preprocessed file |
s |
disdro_pr_from_raw |
Disdrometer-based precipitation rate from raw data |
mm/h |
measV |
measured fall speed at each timestep as a function of the diameter |
m/s |
modV |
modeled droplet fall speed as a function of the diameter |
m/s |
psd_sum_n |
Normalized precipitation size distribution |
#/cm^3/mm |
psd_sum |
Normalized precipitation size distribution |
#/cm^3/bin |
Zdlin_hfov_measv_mie |
Disdrometer Mie reflectivity in lin scale for measured fall drop velocity and horizontal field of view |
mm^6.m^-3 |
Zdlog_hfov_measv_mie |
Disdrometer Mie reflectivity in log scale for measured fall drop velocity and horizontal field of view |
dBZ |
Zdlin_hfov_modv_mie |
Disdrometer Mie reflectivity in lin scale for modeled fall drop velocity and horizontal field of view |
mm^6.m^-3 |
Zdlog_hfov_modv_mie |
Disdrometer Mie reflectivity in log scale for modeled fall drop velocity and horizontal field of view |
dBZ |
Zdlin_vfov_measv_mie |
Disdrometer Mie reflectivity in lin scale for measured fall drop velocity and vertical field of view |
mm^6.m^-3 |
Zdlog_vfov_measv_mie |
Disdrometer Mie reflectivity in log scale for measured fall drop velocity and vertical field of view |
dBZ |
Zdlin_vfov_modv_mie |
Disdrometer Mie reflectivity in lin scale for modeled fall drop velocity and vertical field of view |
mm^6.m^-3 |
Zdlog_vfov_modv_mie |
Disdrometer Mie reflectivity in log scale for modeled fall drop velocity and vertical field of view |
dBZ |
Zdlin_hfov_measv_tm |
Disdrometer geometric reflectivity in lin scale for measured fall drop velocity and horizontal field of view |
mm^6.m^-3 |
Zdlog_hfov_measv_tm |
Disdrometer geometric reflectivity in log scale for measured fall drop velocity and horizontal field of view |
dBZ |
Zdlin_hfov_modv_tm |
Disdrometer geometric reflectivity in lin scale for modeled fall drop velocity and horizontal field of view |
mm^6.m^-3 |
Zdlog_hfov_modv_tm |
Disdrometer geometric reflectivity in log scale for modeled fall drop velocity and horizontal field of view |
dBZ |
Zdlin_vfov_measv_tm |
Disdrometer geometric reflectivity in lin scale for measured fall drop velocity and vertical field of view |
mm^6.m^-3 |
Zdlog_vfov_measv_tm |
Disdrometer geometric reflectivity in log scale for measured fall drop velocity and vertical field of view |
dBZ |
Zdlin_vfov_modv_tm |
Disdrometer geometric reflectivity in lin scale for modeled fall drop velocity and vertical field of view |
mm^6.m^-3 |
Zdlog_vfov_modv_tm |
Disdrometer geometric reflectivity in log scale for modeled fall drop velocity and vertical field of view |
dBZ |
Zdlin_measv_ray |
Disdrometer Rayleigh reflectivity in lin scale for measured fall drop velocity and horizontal field of view |
mm^6.m^-3 |
Zdlog_measv_ray |
Disdrometer Rayleigh reflectivity in log scale for measured fall drop velocity and horizontal field of view |
dBZ |
Zdlin_modv_ray |
Disdrometer Rayleigh reflectivity in lin scale for modeled fall drop velocity and vertical field of view |
mm^6.m^-3 |
Zdlog_modv_ray |
Disdrometer Rayleigh reflectivity in log scale for modeled fall drop velocity and vertical field of view |
dBZ |
attd_hfov_measv |
Disdrometer attenuation for measured fall drop velocity and horizontal field of view |
dB/km |
attd_hfov_modv |
Disdrometer attenuation for modeled fall drop velocity and horizontal field of view |
dB/km |
attd_vfov_measv |
Disdrometer attenuation for measured fall drop velocity and vertical field of view |
dB/km |
attd_vfov_modv |
Disdrometer attenuation for modeled fall drop velocity and vertical field of view |
dB/km |
DVd_hfov_measv_mie |
Disdrometer Mie Doppler velocity for measured fall drop velocity and horizontal field of view |
m.s^-1 |
DVd_hfov_modv_mie |
Disdrometer Mie Doppler velocity for modeled fall drop velocity and horizontal field of view |
m.s^-1 |
DVd_vfov_measv_mie |
Disdrometer Mie Doppler velocity for measured fall drop velocity and vertical field of view |
m.s^-1 |
DVd_vfov_modv_mie |
Disdrometer Mie Doppler velocity for modeled fall drop velocity and vertical field of view |
m.s^-1 |
DVd_hfov_measv_tm |
Disdrometer geometric Doppler velocity for measured fall drop velocity and horizontal field of view |
m.s^-1 |
DVd_hfov_modv_tm |
Disdrometer geometric Doppler velocity for modeled fall drop velocity and horizontal field of view |
m.s^-1 |
DVd_vfov_measv_tm |
Disdrometer geometric Doppler velocity for measured fall drop velocity and vertical field of view |
m.s^-1 |
DVd_vfov_modv_tm |
Disdrometer geometric Doppler velocity for modeled fall drop velocity and vertical field of view |
m.s^-1 |
m2_measv |
Second order momentum for measured fall drop velocity |
mm^2 |
m2_modv |
Second order momentum for modeled fall drop velocity |
mm^2 |
m3_measv |
Third order momentum for measured fall drop velocity |
mm^3 |
m3_modv |
Third order momentum for modeled fall drop velocity |
mm^3 |
m4_measv |
Fourth order momentum for measured fall drop velocity |
mm^4 |
m4_modv |
Fourth order momentum for modeled fall drop velocity |
mm^4 |
dm_measv |
Mean diameter of the precipitation for measured fall drop velocity |
mm |
dm_modv |
Mean diameter of the precipitation for modeled fall drop velocity |
mm |
re_measv |
Effective radius of the precipitation for measured fall drop velocity |
mm |
re_modv |
Effective radius of the precipitation for modeled fall drop velocity |
mm |
lwc_measv |
Liquid Water Content for measured fall drop velocity |
g/cm^3 |
lwc_modv |
Liquid Water Content for modeled fall drop velocity |
g/cm^3 |
n0_measv |
Total number concentration for measured fall drop velocity |
#/cm^3 |
n0_modv |
Total number concentration for modeled fall drop velocity |
#/cm^3 |
radar_longitude |
Longitude of site |
degree_east |
radar_latitude |
Latitude of site |
degree_north |
radar_altitude |
Altitude of site |
m |
radar_model |
Radar model |
N/A |
radar_frequency |
Frequency of the DCR, in Hertz |
Hz |
radar_wavelength |
Wavelength of the DCR, in meter |
m |
alt |
Range from instrument |
m |
Zdcr |
Radar reflectivity factor |
dBZ |
DVdcr |
Doppler velocity |
m s-1 |
weather_data_avail |
Availability of weather data at the station |
N/A |
Global attributes in the output file :
Attribute Name |
Value |
|---|---|
year |
[year] |
month |
[01-12] |
day |
[01-31] |
location |
[Location of the NF] |
ams_source |
[Weather station / None] |
ams_pid |
https://hdl.handle.net/[…] |
disdrometer_source |
[Name of disdrometer model : Thies Clima LNM / OTT HydroMet Parsivel² |
disdrometer_pid |
https://hdl.handle.net/[…] |
radar_source |
[Name of radar : MIRA-35 / RPG-FMCW-94 / BASTA] |
radar_pid |
https://hdl.handle.net/[…] |
station_name |
[Name of the station] |
axis_ratioMethod |
BeardChuang_PolynomialFit |
fallspeedFormula |
GunAndKinzer |
title |
CCRES pre-processing file for Doppler cloud radar stability monitoring with disdrometer at Palaiseau site |
summary |
Disdrometer ([disdrometer_source]) data are processed to derive the equivalent reflectivity factor at 4 frequencies (10.0, 24.0, 35.0, 94.0 GHz). Doppler cloud radar ([radar_source]) data (reflectivity and Doppler velocity) are extracted up to some hundreds of meters, and weather station data (temperature, humidity, wind and precipitation rate) are added to the dataset if provided. The resulting pre-processing netCDF file has a 1-minute sampling for all the collocated sensors. |
keywords |
GCMD:EARTH SCIENCE, GCMD:ATMOSPHERE, GCMD:CLOUDS, GCMD:CLOUD DROPLET DISTRIBUTION, GCMD:CLOUD RADIATIVE TRANSFER, GCMD:CLOUD REFLECTANCE, GCMD:SCATTERING, GCMD:PRECIPITATION, GCMD:ATMOSPHERIC PRECIPITATION INDICES, GCMD:DROPLET SIZE, GCMD:HYDROMETEORS, GCMD:LIQUID PRECIPITATION, GCMD:RAIN, GCMD:LIQUID WATER EQUIVALENT, GCMD:PRECIPITATION AMOUNT, GCMD:PRECIPITATION RATE, GCMD:SURFACE PRECIPITATION |
keywords_vocabulary |
GCMD:GCMD Keywords, CF:NetCDF COARDS Climate and Forecast Standard Names |
Conventions |
CF-1.8, ACDD-1.3, GEOMS |
id |
|
naming_authority |
|
history |
created on [date in ISO 8601] by ccres_disdrometer_processing, v X.X.X |
source |
surface observation from [radar_source] DCR, [disdrometer_source] disdrometer and AMS, processed by CloudNet |
processing_level |
2a |
comment |
|
acknowledgement |
|
license |
CC BY 4.0 |
standard_name_vocabulary |
CF Standard Name Table v84 |
date_created |
[date in ISO 8601] |
creator_name |
ACTRIS-CCRES |
creator_email |
ccres_contact@listes.ipsl.fr |
creator_url |
https://ccres.aeris-data.fr |
creator_type |
institution |
creator_institution |
|
institution |
|
project |
|
publisher_name |
|
publisher_email |
|
publisher_url |
|
publisher_type |
|
publisher_institution |
|
contributor_name |
|
contributor_role |
|
geospatial_bounds |
POLYGON ((lat_min, lon_min), (lat_max, lon_min), (lat_max, lon_max), (lat_min, lon_max)) |
geospatial_bounds_crs |
EPSG:4326 |
geospatial_bounds_vertical_crs |
EPSG:5829 |
geospatial_lat_min |
[lat_min] |
geospatial_lat_max |
[lat_max] |
geospatial_lat_units |
degree_north |
geospatial_lat_resolution |
0.001 |
geospatial_lon_min |
[lon_min] |
geospatial_lon_max |
[lon_max] |
geospatial_lon_units |
degree_east |
geospatial_lon_resolution |
0.001 |
geospatial_vertical_min |
[alt_min] |
geospatial_vertical_max |
[alt_max] |
geospatial_vertical_units |
m |
geospatial_vertical_resolution |
0.1 |
geospatial_vertical_positive |
up |
time_coverage_start |
[start epoch in ISO 8601] |
time_coverage_end |
[end epoch in ISO 8601] |
time_coverage_duration |
[Delta in ISO 8601] |
time_coverage_resolution |
[Time resolution in ISO 8601] |
program |
ACTRIS, CloudNet, CCRES |
date_modified |
[date of modification in ISO 8601] |
date_issued |
[same date a priori] |
date_metadata_modified |
|
product_version |
[X.X.X] |
platform |
GCMD:In Situ Land-based Platforms, GCMD:OBSERVATORIES |
platform_vocabulary |
GCMD:GCMD Keywords |
instrument |
GCMD:Earth Remote Sensing Instruments, GCMD:Active Remote Sensing, GCMD:Profilers/Sounders, GCMD:Radar Sounders, GCMD:DOPPLER RADAR, GCMD:FMCWR, GCMD:VERTICAL POINTING RADAR, GCMD:In Situ/Laboratory Instruments, GCMD:Gauges, GCMD:RAIN GAUGES, GCMD:Recorders/Loggers, GCMD:DISDROMETERS, GCMD:Temperature/Humidity Sensors, GCMD:TEMPERATURE SENSORS, GCMD:HUMIDITY SENSORS, GCMD:Current/Wind Meters, GCMD:WIND MONITOR, GCMD:Pressure/Height Meters, GCMD:BAROMETERS |
instrument_vocabulary |
GCMD:GCMD Keywords |
cdm_data_type |
|
metadata_link |
|
references |