NMEG FluxProc Priorities
-
Gapfill and partition the PJC and PJG datafiles
- May require regenerating the "for_gapfilling" files with VPD (in hectapascals?) and a qc column.
-
What to do with separate Met and Li8100 data?
- Archive raw card data for now.
-
Fluxflag scripts?
-
Temperature corrected longwave radiation does not make it to ameriflux files - seems that the uncorrected values are going into the files instead
-
Fluxall files are inconsistent in the dates that they contain. Some start at year/1/1 00:00 (last half hour of previous year), some start a half hour later. Some end a half hour early. This cause problems when combining the data with other dataloggers. Need to address this and create some timestamp conversion tools.
-
An unburned grassland (New_GLand) card may have been processed as GLand on 8/5/2014 (see incorrect datalogger id in the 7z file)
-
Where does the Air temperature in the for_gapfilling file come from - it is not from the qc file (see pj and pj girdle Dec 8, 2011 at 5pm)
- It is actually derived from Tdry (dry bulb temperature in Kelvin) and converted to C.
- but where does Tdry come from - it comes in with the flux data.
- This is sonic air temperature (Ts), and it is corrected to be "dry temperature" by the "dry_air_conversion" file.
-
RBD file assigns the last header matching a list to air_temp_hmp and rH. At sites with multiple sensors this is often wrong - look at air temp at PJC.
-
There is no hmp RH data from PJ girdle for 2011 and 2012 (not in FLUXALL file), but something does get put in the qc and for gapfilling files - what is this?
- It is calculated from other hmp variables found in FLUXALL FILE
-
Rounding errors for PAR in Ameriflux Daily files - see PJ Daily AF files (PAR rounded to thousands place)
-
There are 2 UNM_Ameriflux_filemaker pipelines (regular and "_TWH"). The TWH version uses the output from the MPI online flux partitioning tool (old version). The new one uses the REddyProc R scripts (Accessed with UNM_run_gapfiller.m).
-
PJ Girdle 2014 fluxall file is missing a bunch of data - mostly the 10hz data
-
JSav 2013 Fluxall has lots of NAN columns in the TOB1 data - need to finish making Amflux files
- apparently these exist in the TOB1 data also - maybe they can be filled with wireless data?
-
Soil data headers, including soilT, SWC and shf data are renamed from the incoming raw data by the
UNM_Assign_soil_data_labels.m
script.- This interferes with the header resolution scripts, and it may make more sense to remove this and replace it with the header resolution system
- Soil header changes would be put into the header resolution file (or maybe just for 2013 on?)
- Also, there is a separate file that deals with JSav 2009 - which seems like unnecessary code overlap
- This interferes with the header resolution scripts, and it may make more sense to remove this and replace it with the header resolution system
-
Variable names in old fluxall files (starting in around 2010) are different.
- I have changed 'u_mean' to 'u_mean_unrot' in 2010 fluxall files from some sites.
- I have changed 'windDirection(theta)' to 'wind_direction' in some 2010 fluxall files
-
Something is wrong with PAR at JSav in 2014. It is all getting normalized to zero.
-
PPine 2013 Ameriflux file
- Fix UNM_fill_met_gaps_from_nearby_site.m so that it uses the correct data for PPine.
- This should eliminate gaps in precip and shitty VPD gapfilling.
-
Add criteria and data removal for atmospheric stability parameters and advection at sites
- For stability and advection calculations see chapter in Lee, Massman, and Law book.
- This will be done in RemoveBadData for now.
-
JSav 2013 Ameriflux problems.
- Fluxes Reco, GPP, FC seem very low (magnitude and variability) between days 60-170.
- Is there an incorrect hard cap on R, GPP, or FC for the springtime period in 2013?
- Is there a diurnal StdDev check that isn't functioning?
-
Make ameriflux file spit out Tsoil so we can test whether this makes for better flux filling/partitioning?
-
In the ameriflux files, there is no way to tell what values were locally gapfilled. This can be a problem if values calculated (in RBD, say) from missing met data end up as NaN, but the missing met data is later filled in with local or gapfiller data.
-
New precip problem at PJ Girdled from 1/1/2009 to 6/2/2009 (or a little earlier)
- Data (no of tips) copied from PJC, then converted with the PJG multiplier (so precip is doubled)