Model Initialization Modes

We start this tutorial by running FATES at Barro Colorado Island, Panama, with a single Plant Functional Type (PFT). For this first run we will initialize the model with inventory data, meaning the simulation starts with a size structure that matches observations. However, there are other ways to initialzie FATES, and so for reference we have described them all here.

The state of the terrestrial ecosystem in FATES can be initialized in roughly three different ways.

1. A “restart” simulation: During the initialization phase, the model will read in detailed information of the complete “state” of the FATES model, and its host model (ELM), that was generated during a previous simulation. A restart simulation should generate bit-for-bit results with the previous simulation in absolutely all aspects, including carbon, water, energy, demography and nutrients.

2. A “cold-start”, spin-up type simulation. The energy, water, nutrient and soil carbon states will be initialized with nominal values. The vegetation in this case, is initialized by creating a single cohort for each of the represented plant functional types in the fates parameter file, with an initial number density as specified in the fates parameter file, and a starting size that equals the smallest size of a cohort (the size of a new recruit, defined by h_min). It is assumed that over a suitably long enough period of time, the relevant quantities of interest in the simulation will have responded and reacted to the model boundary conditions such that no (or limited) memory of the arbitrary initial conditions remain.

3. A “cold-start”, with inventory initialization. This is the method used in this tutorial. This mode is similar to Case 2, in that the soil biogeochemistry, hydrology and energy states are given nominal values. However, in this case, the user can direct the initialization to read plant demographic data from a set of files. The canopy structure and composition that is defined in the files is generally limited to the size, type and number of plants that exist, and potentially information on existing litter pools and coarse woody debris.

Guide for Restarting a FATES Simulation (Case 1)

To start up a new case using the restart file from a prior case as your initial conditions, add a line like this to your user_nl_elm file:

finidat=‘full_path_to_restart_file.clm2.r.0000.nc’

To extend a case using a restart file from that case, from the case directory run:

./xmlchange CONTINUE_RUN=TRUE
./case.submit

Guide for a Cold Start Simulation without Inventory (Case 2)

To start a new case completely from scratch do not specify a restart file or point to any inventory initialization files.

Guide for a Cold Start Simulation with Inventory (Case 3)

The key element in an inventory initialization, is specifying what inventory data is available and what format the data is provided in. The pre-tutorial work should have resulted in your data being correctly formatted. But for our first run we will use the data from BCI which is publically available from Dryad. We have already downloaded this data and formatted it for FATES. You will find the BCI data under inventory_data in your container.

All of the inventory initialization code in FATES is contained in this file:main/FatesInventoryInitMod.F90

Namelist Entries

To tell FATES what configuration we want and which optional modules to turn on, we specify a number of name list options when we set up the case. This will be explained in more detail in the next lesson. For inventory initialization two namelist entries are required: a flag that turns on inventory initialization, and the full path to a control file, these entries can be specified in user_nl_elm (described in more detail in the next lesson).

use_fates_inventory_init = .true.
fates_inventory_ctrl_filename = '<full path to the control file>'

Control File Specification

The control file should be text formatted. It should contain 1 header row. It should contain any number of rows greater than 1, that each specify inventory site data that the user would like the model to interperate. This description of the site data should contain: a format specifier for the data at that site, the latitude and longitude in decimal degrees (both 0-360 and -180 to 180 conventions allowed), and then the full path to both the patch file and the cohort file. Below is an example of the control file.

TYPE LATITUDE LONGITUDE PSS_PATH CSS_PATH
1 8.5263157894736 280 /inventory_data/bci/BCI_2015.pss /inventory_data/bci/BCI_2015.css

Inventory Format

‘’Patch’’ file format (*.pss)

Variable	Units	Description
time	years	Year
patch	String	Patch identifier (arbitrary, any unique string that can be used to match cohorts)
trk	0 – non-forest, 1 – secondary, 2 - primary	Vegetation type/history
age	years	Patch age since disturbance
area	proportion	Fractional area represented by patch. For format 1 is area in m2

‘’Cohort’’ file format (*.css)

Variable	Unit	Description
time	year	year
patch	string	Unique identifier (the string matching with the patch its on)
dbh	cm	Stem diameter breast height
pft	integer	Plant Functional Type
n	Stem/m2	Stem density

In the next section we show you how to set up a FATES run and point to these files for an inventory initialization.