Satellite-detected time series of primary production in the California Current

Mati Kahru; mkahruATucsd.edu; to email me, please replace AT with @

        Data files at: http://spg.ucsd.edu/Satellite_Data/California_Current/NPP_month/

• This project uses the full resolution (1-km) time series of Chl-a and SST built by merging data from multiple sensors for the California Current area (here).
• Wimsoft (http://wimsoft.com) software was used in the analysis and in generating the images shown below.
• Net Primary Production (NPP) is calculated from various satellite data using a recently adapted NPP model (Kahru, M. et al, Trends in primary production in the California Current detected with satellite data, J. Geophys. Res., 114, doi:10.1029/2008JC004979, 2009; see PDF). The NPP model used here (called VGPM-CAL) is based on hundreds of match-ups with in situ NPP measurements collected over the years by the CalCOFI program and is a simple adaptation of the well-known Behrenfeld-Falkowski VGPM model.
• The following input data are used:
  • Chl-a data is from our full resolution (1-km) time series of Chl-a merged from multiple sensors (if available) (see here)
  • PAR data is from the standard SeaWiFS global Level-3 data (9-km resolution)
  • SST data is merged from MODIS-Terra and MODIS-Aqua daytime 4-km data (if available) or AVHRR-Pathfinder 4-km data (earlier years)
  • The MODIS-A and MODIS-T SST datasets are merged with wam_composite_2sensors. For example, for the 2008 data the following command was used:
    • wam_composite_2sensors F:\MODISA\L3\Monthly\SST_4\A2008*.hdf F:\MODIST\L3\Monthly\SST_4\T2008*.hdf
    • Before merging, the SST datasets were always screened to eliminate suspicious and bad data with wam_screen_sst_ocpg
  • Therefore, the 1-km resolution NPP datasets are produced from 1-km Chl-a, 4-km SST and 9-km PAR data. It is assumed that the lower resolution of PAR and SST is of secondary importance. PAR varies relatively smoothly over space and SST has a rather weak influence on NPP; therefore these omissions are thought to be acceptable. The 4-km global SST data are used instead of the 1-km California Current SST produced locally as it is assume that the quality controlled and flag-assigned SST data are more reliable and the lower spatial resolution is less important.
• NPP datasets are calculated with the following command: wam_npp listfile VGPM-CAL
• The list file specifies the paths and patterns of input files. For example, for 2008 the list has the following text:

  # path for Chl files, matching pattern
  N:\CAL\2008\C2008_chl_month,C2008*comp.hdf
  # path for PAR files, matching pattern
  F:\SEAWIFS\L3\Monthly\PAR_9,S2008*PAR_9
  # path for SST files, matching pattern
  F:\Merged\L3\Monthly\SST_4,M2008*.hdf

•  Currently the monthly NPP datasets (3840 columns x 3405 rows) are generated using monthly composites of Chl-a, SST and PAR. As NPP is a nonlinear process, it would be more accurate to calculate daily NPP data and composite those daily NPP datasets into longer composites. However, that would involve much more computing and it is therefore currently not done.
• The NPP datasets are recorded in Int16 datasets with no scaling (i.e. slope =1, intercept = 0) in HDF4 files in mg C m^-2 day^-1.
• For each HDF data file a corresponding quick-look image in the PNG format is stored. These were created with wam_series. An example PNG image for January, 2002 is shown below.
• As the NPP data files are in Albers Conic equal area projection in the full resolution (1-km) of the Chl-a data, they can be remapped to any area within the domain while maintaining the 1-km resolution. Both the HDF and PNG file are accessible at http://spg.ucsd.edu/Satellite_Data/California_Current/NPP_month/.

  Please note that the files are not available through FTP but through HTTP. For downloading multiple files at once you can use the utility wget (e.g. http://www.gnu.org/software/wget/).