** Sun Sep 9 18:04:36 2012 Andy Jacobson ** There are many variables in these files, and you'll need to parse the names carefully to find what you want (hopefully it's obvious, if a bit tedious). There are three sets of "regions" (reg240, tc23, and postagg), four classes of flux (terrestrial bio, ocean, FF, fire), three types of flux (imposed, prior, and optimized), and both fluxes and flux uncertainty are in there as well. All the permutations are listed as individual variables. As this CarbonTracker release is a multi-model inversion, there is total uncertainty, within-model uncertainty, and across-model uncertainty for each optimized flux, and for fossil fuel emisisons. The within-model uncertainty is called "internal" uncertainty, and is the average of each inverse model's posterior uncertainty. The across-model uncertainty is called "external" uncertainty and is computed as the covariance across the multiple models of the posterior flux. Total uncertainty is the sum of the internal and external variances (note variance, not std deviation; uncertainties add in quadrature). Geographic region definitions are in ftp://aftp.cmdl.noaa.gov/products/carbontracker/co2/regions.nc Postaggregate region definitions are in ftp://aftp.cmdl.noaa.gov/products/carbontracker/co2/postagg_definitions.nc. ** Sun Mar 13 15:29:47 2016 Andy Jacobson ** Say you want total land flux over the northern extratropics for each year, and the "official" long-term mean (LTM). The official estimate just excludes 2000 since we suspect there are flux artifacts associated with spinning up the atmosphere from an imperfect CO2 initial field. You would first identify the fluxes you want, probably total land flux is "bio" plus "fire", so you'd take the sum of bio_flux_opt_postagg and fire_flux_imp_postagg. Posterior bio flux is optimized; fire fluxes are imposed. Postagg is because you want a pre-defined postaggregate region, "Northern Land". The pre-defined postaggregate regions are defined in ftp://aftp.cmdl.noaa.gov/products/carbontracker/co2/postagg_definitions.nc. These are aggregations of the 23 TransCom regions (11 land, 11 ocean, 1 unoptimized areas), which themselves are aggreagtes of our 240 regions. Regiona are spatially defined on a 1x1 grid in our ftp://aftp.cmdl.noaa.gov/products/carbontracker/co2/regions.nc file. For convenience, here are the names and numerical indices of those postaggregate regions: 01 Northern Land 02 Northern Oceans 03 Northern 04 Tropical Land 05 Tropical Ocean 06 Tropical 07 Southern Land 08 Ocean in South 09 Southern 10 Americas 11 Atlantic and Arctic 12 Eurasia and Africa 13 Indian Ocean 14 Asia 15 Pacific Ocean 16 All Land 17 All Ocean 18 Global 19 North America 20 Extratropical Eurasia 21 Temperate Northern Ocean 22 Temperate Southern Ocean 23 Tropical and Southern Land 24 Tropical Pacific 25 Temperate Oceans 26 High Latitude Oceans 27 Extratropical Oceans 28 Tropical and Southern Oceans 29 Northern Boreal Land 30 Northern Temperate Land 31 Tropical land minus northern land 32 High North Ocean and Boreal Land 33 Northern Temperate Land and Ocean 34 Southern Temperate Land and Ocean 35 Tropical and Southern Land and Ocean 36 Tropics and south minus north 37 Tropical and southern land minus northern land 38 Africa 39 South America Reading the postagg_definitions.nc file, you would find that postagg region 1, Northern Land, is compoased of TransCom regions 1, 2, 7, 8, and 11. Thumbbnail map is at http://www.esrl.noaa.gov/gmd/ccgg/carbontracker/images/regions/region_Northern_Land.png. Software packages used to access netCDF data like Matlab, R, FORTRAN, C, and Python use different syntax, especially with regards toindices into arrays. I will just show the R syntax here. rfa <- load.ncdf("CT2015.regionfluxes_annual.nc") str(rfa$bio_flux_opt_postagg) num [1:39, 1:15] -6787484 -183806 -6971290 -1499510 -79495 ... - attr(*, "units")= chr "mol region-1 s-1" - attr(*, "_FillValue")= num -1e+34 - attr(*, "long_name")= chr "optimized value of terrestrial biosphere exchange, postaggregated regions" flux_annual <- 12e-15*rfa$seconds_in_year*(rfa$bio_flux_opt_postagg[1,]+rfa$fire_flux_imp_postagg[1,]) "flux_annual" will now contain the land biosphere flux including wildfire emissions, integrated over all Northern Land, in PgC/yr. Note that "load.ncdf" is a shorthand function written at ESRL to access netCDF data; other methods may be used to load the data. Note also the use of the seconds_in_year array, which properly accounts for leap years in the time series. To retrieve the long-term mean, one would use the same syntax, but get the variables from CT2015.regionfluxes_annual.nc. The variable names are identical, but quantities are scalars instead of vectors. Like other variable, seconds_in_year is the average of data for all years excluding 2000.