Site scale simulations

## Setup for site-level simulations

Below, the setup - from processing input data to running simulations - for site-level simulations for all FLUXNET 2015 Tier 1 sites is given as an example.

Summary of setup steps for site-level simulations:

  1. Get fAPAR forcing data (for P-model only)
  2. Get climate forcing data
  3. Get CO2 forcing data
  4. Get site and simulation parameter files
  5. Run SOFUN (see usage)

Model forcing data is available on Imperial HPC’s CX1 (work/bstocker/data) or can be downloaded or read from files using R scripts in repository getin. All described below.

Here, we’re preparing site-specific model forcing data. Each simulation suite (collection of sites that belong to a certain dataset, measurement campaign, or similar), has one meta information file which specifies (at minimum) the following information:

  • mysitename: site ID, character
  • lon: longitude of site, decimal
  • lat: latitude of site, decimal
  • year_start: start year of the simulation (for which data is available)
  • year_end: end year of the simulation (for which data is available)

For certain simulation suites, meta information files are available from the repository getin: getin/metafiles_<simsuite>/siteinfo_<simsuite>_sofun.csv. The meta information file the “centre-piece” that is used by several scripts that loop over sites in a given simulation suite. It may also be generated by the R script in repository getin prepare_paramfils_<simsuite>.R, available in repository getin.

Here, we’re using the FLUXNET 2015 dataset. Therefore, we’re setting:

simsuite = "fluxnet2015"

Forcing data is written into different subdirectories of your_chosen_home/sofun/input_<simsuite>_sofun/. Make sure this directory exists (here for simuite=fluxnet2015):

cd your_chosen_home
mkdir -p sofun/input_fluxnet2015_sofun

1. fAPAR Get forcing data

When executing the calibration on a local machine, download the data from CX1 or process new data.

Downloading from CX1

Site-scale subsets from the MODIS FPAR MCD15A3H (Collection 6) data (4 days, 500 m) for all FLUXNET 2015 Tier 1 sites are available on Imperial HPC’s CX1: work/bstocker/data/fapar_MODIS_FPAR_MCD15A3H_fluxnet2015_gee_subset/fapar_MODIS_FPAR_MCD15A3H_<sitename>_gee_subset.csv

Download the entire directory fapar_MODIS_FPAR_MCD15A3H_fluxnet2015_gee_subset and place it in your_chosen_home/data/. Do not modify the directory and file name structure from your_chosen_home/data/ downwards. Specify the path in the header of file XXX.

Processing new

Downloading site-scale data from Google Earth Engine and interpolating to daily data is done using gee_subset.py from the repository gee_subset by Koen Hufkens and get_sitesubset_gee.R from the repository getin. Gapfilling and interpolation to daily values is done by filtering based on the MODIS quality flags (see gapfill_modis.R from the repository getin) and applying a spline to daily values.

your_chosen_home is the path where you chose to place the repositories getin and gee_subset and the data and sofun directories.

1. Clone gee_subset

In your shell, do:

cd your_chosen_home
git clone https://github.com/stineb/gee_subset

Switch to branch gee_stineb by:

cd your_chosen_home/gee_subset
git checkout gee_stineb

To execute gee_subset.py, you must have a Google Earth Enginge login and authenticate yourself. To set this up, follow steps described in setup_steps.md.

2. Clone getin

In your shell, do:

cd your_chosen_home
git clone https://bitbucket.org/labprentice/getin

You must have a login on bitbucket and belong to the group ‘labprentice’ to get getin.

3. Execute

Change to the directory to where your local clone of the getin repository is located. In the header of the R script get_sitesubset_gee.R (see MANUAL SETTINGS), set then manual settings by hand, for example:

##--------------------------------------------------------------------
## MANUAL SETTINGS
##--------------------------------------------------------------------
myhome               = "/alphadata01/bstocker/"
simsuite             = "fluxnet2015"
bundle               = "fapar"
overwrite_raw        = FALSE
overwrite_nice       = TRUE
do_plot_interpolated = TRUE
start_date           = "2000-01-01"
end_date             = "2017-11-27"
years_out            = 1980:2017   # creating SOFUN input data for these years
##--------------------------------------------------------------------

Then execute it in R:

setwd("your_chosen_home/getin")
source("get_sitesubset_gee.R")

2. Get climate forcing data

Climate model forcing for site-scale simulations is extracted from site-specific meteo data and/or global fields (WATCH-WFDEI and CRU TS implemented), and written to CSV files containing all days of all simulation years, as well as annual ascii text files for each year. All files are stored in a pre-defined directory structure:

  • CSV files: your_chosen_home/sofun/input_<simsuite>_sofun/sitedata/climate/<sitename>/clim_daily_<sitename>.csv
  • Annual ascii text files: your_chosen_home/sofun/input_<simsuite>_sofun/sitedata/climate/<sitename>/<year>/d<var>_<sitename>_<year>.txt

var stands for the variable name (temp, prec, vpd, ppfd, [netrad]).

All climate data for FLUXNET 2015 simulations can be read from data distributed by the FLUXNET 2015 dataset. Download the full dataset from CX1 to your local machine (alternatively, run all the following scripts on CX1). The dataset with daily data is on CX1 under: /work/bstocker/data/FLUXNET-2015_Tier1. Place this directory in your_chosen_home/data/.

Processing new climate input files is done using functionality from the repository getin. Clone this repo as described above. In get_climate2.R, adjust the header manually. For example:

##--------------------------------------------------------------------
## MANUAL SETTINGS
##--------------------------------------------------------------------
myhome           = "/alphadata01/bstocker/"
simsuite         = "fluxnet2015"  
overwrite        = TRUE
overwrite_byst   = TRUE
in_ppfd          = TRUE
in_netrad        = FALSE
startyr_override = 2010
add_cru          = FALSE
add_watch        = FALSE
add_meteo        = TRUE
##--------------------------------------------------------------------

Then execute it in R:

setwd("your_chosen_home/getin")
source("get_climate2.R")

3. Get CO2 forcing data

Execute the script that places a copy of the global CO2 time series in a separate directory for each site. Necessary like this - keeping it flexible for cases where CO2 is manipulated per site/experiment. Adjust header:

##--------------------------------------------------------------------
## MANUAL SETTINGS
##--------------------------------------------------------------------
myhome           = "/alphadata01/bstocker/"
simsuite         = "fluxnet2015"  
##--------------------------------------------------------------------

Then execute in R:

setwd("your_chosen_home/getin")
source("get_co2.R")