Updates to MOD06 Climatology Script and MDO06 processing to isolate the effects of the MOD35 -landcover bias

Adjusted handling of cloud flag to 'hopefully' improve missing data problem in MOD06 climatology

restructured MOD06 processing routine to eliminate functions for more transparent (linear) processing. Also updated to run on either pleiades or litoria

Merge branch 'aw/precip' of code.nceas.ucsb.edu:environmental-layers into aw/precip

Adding intial code to process the NDP-026D station cloud climatology dataset

Updated submit script to process multiple tiles and days in the same batch. Also separated the daily processing from the climatologies

Merging with Pleiades copy

Merging with Pleiades copies

Adding Climatology script to process daily files to climatologies

Merge branch 'aw/precip' of code.nceas.ucsb.edu:environmental-layers into aw/precip

Identified problem in handling of missing values of MOD06 product. In some areas limiting quality control to the highest quality results in ~90% NAs.

initial script to handle modified netcdf files from Pleiades.R script

Merging GAM_LST script from bp/interp to aw/precip

Merging with remote branch"

Merge branch 'aw/precip' of code.nceas.ucsb.edu:environmental-layers into aw/precip

Updated summary script to work with netcdf output files

MOD06 processing running successfully using Pleiades environment set up by Andrew Michaelis at NASA.

Script now runs except ncap2 is missing from the nco module

Currently submission hangs using mqueue when loading rgdal library. Will attempt to add a separate submission script that first loads modules before running Rscript on the slaves

MOD06_L2 now running properly but Rmpi is not running due to (probably) incompatability between Rmpi and the version of sgi-mpi available on Pleiades

Updated script to accurately set paths for HEG tool to facilitate parallel processing of gridding procedure using swtif. Multicore package is used to parallelize which limits processing to a single node. Next step is to explore use of foreach package to allow multiple nodes

Updated MOD06 process to output NetCDF files instead of geotif for faster post-processing

Resolved problem with temporary directory for Rscript call.

Script is successfully running and producing the summary files, though the output looks strange. Maybe a problem with sinusoidal output of HEG? Use Pleiades.R to drive the submission and MOD06_L2_process as the processing script

Successfully running MOD06 processing on Pleiades as an array job (though submissions are limited to < 365 jobs so will have to find another way to submit them)

Started working with script to process MOD05 on NASA's Pleiades

Added script to evaluate climatic stationarity (Task #479)

Updates to run 12 months of mean daily precipitation using a variety of models. Includes using gamm() to incorporate spatial effects in addition to the smooth parameters. Not all clean and cozy yet, uploading as a work-in-progress...

Further updates to interpolation procedures including:

1) bugfix in MOD06_L2_data_compile. Root access is needed to run swtif in its current directory (bad design!).
2) Switched to UTM output from swtif (from sinusoidal) becuase it looked like gdal was struggling with integerized sinusoidal...

Modified GAM.R interpolation script to 1) simplify validation results, 2) add full prediction (rather than just stations) and 3) some other (hopefully) simplifications and improvements.

Updated GHCN_stations script to:

1) draw station location information from postgres database and perform merge/join query within database rather than r
2) add function to buffer region of interest to include stations outside region (to minimize edge effects)

Merging with bp/interp to use some of Benoit's code for precip interpolation

Merge branch 'bp/interp' into aw/precip

More updates to output plots

OR data preparation, initial commit for extraction of covariates from raster stack

OR data preparation task#363, modified code for ghcn to create a shapefile

Added initial version of LST_Landcover exploration (not finished). Also moved prior version of interpolation procedure (bayesian krig) into repository.

Minor updates to dataset size calculation

added processing for 'very cloudy days'

Added PRISM comparison to MOD06_summary and added script to calculate 'back-of-the-envelope' estimates of total daily climate layer sizes

GHNCD station data selection using Postgres database, initial commit, task #363

GAM LST, added specific and general diagnostic measures (e.g.MAE) GAM+Kriging, task #364 and #406

Updated MOD06 compile script to process the tiles in parallel using GRASS. This brings processing time for 10 year archive for oregon from 48 hours to ~2 hours on 24 cores. Also added several exploratory analysis to the data_summary script.

Added cloud flag to outputs

First 10-year processing of daily mean tifs

First full processing of daily mean tifs

GAM LST, adding lines to summarize samling RMSE using plots, task #409

GAM LST, first code to test effect of sampling on GAM, task #409

GAM LST, added model and modified summary plots, run on 365 dates GAM+Kriging, task #364 and #406

GAM LST, modify code to generalize number of models and assessment,GAM+Kriging task #364

GAM LST, major modifications to create GAM+Kriging code, task #364

GAM LST, adding lines to assess RMSE per month over 365 dates, task #406

GAM LST, added model with forest only term -364 dates assessment, task #406

GAM LST, added section to save training and test residuals in shapefiles, task #361

GAM LST, changed triple nesting lat,long,Elev to double, 365 dates run, task #406 and #361

GAM LST, used monthly LST average instead of daily, task #361

GAM LST, adding new model with grass land cover, task #361

GAM using LST as input variable task #361

GAM, modification of nesting of models and clean up, task #361

Kriging, adding LST for co-kriging: task #364

Kriging, slight modification and clean up, task #364

Kriging, modification to save kriged surfaces in geotiff and png files, task #364

Kriging, modifying code to save training and testing samples in shapefiles: task #364

Kriging for OR, this is the first script, task #364

GWR modified code to save plots and images, task #364

GWR assessment by visualizing residuals using kriging

GWR in adding loop for 10 dates

GWR first script for one date in OR, task #364

MOD06 processing runs well in sequence, but not parallel

Reorganization: moved mindmap to it's own directory to isolate temporary files created when exporting to other formats

Add initial version of the 'mind map' for the interpolation project. This file contains an outline of the various stages of the project and links back to issues and code in the blessed repository.
Created during Benoit's visit to Yale May 21-22.

Adding some MOD06 (cloud product) processing routines. Still a work in progress

Initial commit

Added 3 new models and nesting to GAM, task #361

Changes in the diagnostics plots, barplots.

Transform aspect in Eastness and Northness variables in the GAM.

Transformed the aspect variable and added GAM model closer to PRISM

GAM prediction added to the loop with RMSE calculation

Streamlining and slight changes to the loop to create subset dataset for different dates

Additional changes to the loop and format

Start of Task #361, modified code to subset for 10 dates

Minor format changes and clean up to code

Corrected errors in RMSE calculation and added plots to view results

simplified code by using 'a' flag for r.neighbors (improves 3abda63)

reorganized AML terrain scripts to reduce superficial redundancy

made minor changes to PRISM variables

changed RMSE and AIC code in GAM script

updated GAM script for tmax

added initial R script with GAM regression for tmax (one date)

renamed script to multiscalesmooth.R

numerous comment, format, and variable name changes

now actually refining before smoothing (GRASS version)

reversed order of refine-and-smooth step

added GRASS/Python implementation of multiscale smoother

added R script to run multiscalesmooth example on Oregon SRTM

switched to more accurate chisq test

enhanced smoother to accept sd as a raster, not just a constant

tweaked var names to match Gallant pub; added comments

fixed vwg step

continued R translation of multiscale smoother