Chirps v1.8 tasks

From CHG-Wiki
Revision as of 14:52, 2 September 2014 by Libby (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search

(see also CHIRPS FAQ)

Notes

When prelim moves to v1.8, back process to the beginning of 2013.

Docs docs docs

reorganized ftp hierarchy to be more shallow and explicit

    global_monthly_tifs
    africa_dekad_bils

etc.

Data Location

     http://tinyurl.com/chg-chirps

points to

     ftp://chg-ftpout.geog.ucsb.edu/pub/org/chg/products/CHIRPS-latest/

CHIRPS v1.8

As of 2014.01.27 decided to do a complete rerun of CHIRPS to be called CHIRPS v1.8

Monthly 1.7 CHIRPS were last run in mid November 2013. These have the problem with flipping the order of the neighbors used to fill in master locations.

  • Ingest new station data,
  Brazil (junior)
  World Bank coffee countries: Guatemala, Honduras, Nicaragua
  Panama
  New GTS? GSOD? 
  • Exclude bad Zambia data
  • Address problem Brent found with daily September data in the Horn.
  • Run some reality checks on the station data in the db.
  • Verify the code fix that was flipping the neighbor list is included in this run

Found something wrong with CHIRP data for June-August 2013. # of pixels with positive rainfall plummets. As much as 60% in S. Africa. Does not seem physically possible.

Finalizing

Once all stations finalized:

  • recalculate neighbor station statistics
  • recalcu;ate master locations
  • pre-compute SPI for all master locations (month/pentads)
  • go through CHIRPS MQB
  Merge
  QC
  Blend

Fixes, etc.

Zambia-Met pptx here,

    ftp://chg-ftpout.geog.ucsb.edu/pub/org/chg/products/CHIRPS-1.7/docs/Zambia-CHIRPS_problem.pptx

Fixed a bunch of station locations in the stations table

    ftp://chg-ftpout.geog.ucsb.edu/pub/org/chg/products/CHIRPS-1.7/docs/CSCD.station.cleanup.pptx

Use the same landmask everywhere, avoid variations in the ocean/land boundary

   /home/world/masks/Global_landmask.2014.50N50S.p05Deg.tif

CHIRP May-September 2013 glitch

   changed CHIRP calc script for a while, will redo entire CHIRP timeseries with current script