Using the framework

This section only gives a brief description of the framework focussing on the extensions made for OpenStreams. A full description of the current version of the framwork can be found at http://www.pcraster.eu.

In order to build a dynamic model the one needs to define a model class and add several methods to the class to describe the model behaviour. The easiest way to get started is to copy and modify the wflow_sceleton.py example model. You can also use the other models for inspiration.

In order to faclitate reusing data between models the data is stored in the following directory tree:

digraph file_system {
//rankdir=LR;
size="8,11";
"Case" -> "inmaps";
"Case" -> "instate";
"Case" -> "intbl";
"Case" -> "intss";
"Case" -> "outstate";
"Case" -> "Run";
"Case" -> "staticmaps";
   "Run" -> " intbl";
   "Run" -> "outmaps";
   "Run" -> " outstate";
   "Run" -> "outsum";
   "Run" -> "runinfo";
}

Although it is possible to deviate from this layout it is highly recommended to adhere to this if you build your own model. Also make sure you use an ini file to specify model settings instead of putting those in the python code.

A basic sceleton of a model is given below:

Definition of the wflow_sceleton model.

This simple model calculates soil temperature using air temperature as a forcing.

Usage: wflow_sceleton -C case -R Runid -c inifile

-C: set the name of the case (directory) to run

-R: set the name runId within the current case

-c name of the config file (in the case directory)

$Author: schelle $ $Id: wflow_sceleton.py 898 2014-01-09 14:47:06Z schelle $ $Rev: 898 $

class wflow_sceleton.WflowModel(cloneMap, Dir, RunDir, configfile)

The user defined model class. This is your work!

__init__(cloneMap, Dir, RunDir, configfile)

Required

The init function must contain what is shown below. Other functionality may be added by you if needed.

stateVariables()

Required

Returns a list of state variables that are essential to the model. This list is essential for the resume and suspend functions to work.

This function is specific for each model and must be present. This is where you specify the state variables of you model. If your model is stateless this function must return and empty array (states = [])

In the simple example here the TSoil variable is a state for the model.

Variables:TSoil – Temperature of the soil [oC]
supplyCurrentTime()

Optional

Supplies the current time in seconds after the start of the run This function is optional. If it is not set the framework assumes the model runs with daily timesteps.

Ouput:

  • time in seconds since the start of the model run
suspend()

Required

Suspends the model to disk. All variables needed to restart the model are saved to disk as pcraster maps. Use resume() to re-read them

This function is required.

initial()

Required

Initial part of the model, executed only once. It reads all static model information (parameters) and sets-up the variables used in modelling.

This function is required. The contents is free. However, in order to easily connect to other models it is advised to adhere to the directory structure used in the other models.

resume()

Required

This function is required. Read initial state maps (they are output of a previous call to suspend()). The implementation showns here is the most basic setup needed.

dynamic()

Required

This is where all the time dependent functions are executed. Time dependent output should also be saved here.

wflow_sceleton.main(argv=None)

Optional

Perform command line execution of the model. This example uses the getopt module to parse the command line options.

The user can set the caseName, the runDir, the timestep and the configfile.

Anotated source code for the above