Skip to content

Latest commit

 

History

History
307 lines (261 loc) · 13.8 KB

README.md

File metadata and controls

307 lines (261 loc) · 13.8 KB
Raw

DSSATTools package

Installation:

You can install the library using Python pip.

pip install DSSATTools

v2.1 Updates

For the latest version the next changes were implemented:

  • The library simultes only one treatment. Therefore, only one option for cultivars, irrigation, fertilizer, field, etc. can be defined.
  • Every set of defined crop or management parameters is a section. Each section is an attribute of the Crop or Management class. Sections won't be created by the user. The user can only modify the value of the parameters of the section, they can't create or add new parameters.
  • The weather is now managed by a single Weather class.
  • A __repr__ method was implemented for the four basic classes (Crop, Management, Weather and SoilProfile), and the Section class.
  • The cultivar is selected when initializing the crop instance. Thus, the crop instance contains parameters only for that cultivar.

Documentation

https://py-dssattools.readthedocs.io/en/latest/index.html

Example Notebooks

You'll find example notebooks in this repo:https://github.com/daquinterop/DSSATTools_notebooks. I'll keep uploading examples as some new feature is introduced.

Module contents

DSSATTools library allows the user to create low-code scripts to run simulations using the DSSAT modeling framework. The library structure allows to executes DSSAT based on four input classes: Crop, SoilProfile, Weather and Management.The simulation environment is managed by the DSSAT Class. There are three stages for the simulation to be performed:

  1. Initialize a DSSAT instance.
  2. setup the simulation environment by using the DSSAT.setup method. When that method is called a new directory is created in the provided location (a tmp directory is default) and all the files that are necessary to run the model are copied in that folder.
  3. run the simulation using the DSSAT.run method. That method needs three parameters to be pased, each one indicating the crop, soil, weather, and management. This step can be performed as many times as one wants.
  4. close the environment using DSSAT.close. This removes the directory and the files created during the environment setup.

The next simple example illustrates how to run a simulation using the five aforementioned classes:

>>> crop = Crop('maize')
>>> weather = Weather(
        df, # Weather data with a datetime index
        {"tn": "TMIN", "rad": "SRAD", "prec": "RAIN", "rh": "RHUM", "TMAX": "TMAX"},
        4.54, -75.1, 1800
    )
>>> soil = SoilProfile(default_class='SIL')
>>> man = Management(planting_date=datetime(12, 3, 2020))
>>> dssat = DSSAT()
>>> dssat.setup()
>>> dssat.run(soil, wth, crop, man)
>>> growth = dssat.output["PlantGro"] 
>>> dssat.close() # Terminate the simulation environment

The parameters for ecach class are described later. It is very important to note that this library will allow the user to run one treatment at a time. If the user is familiar with DSSAT, they must know that DSSAT allows to define multiple treatments in the same experimental file.

All of the parameters and attributes of the four basic clases have the same name you find in the DSSAT files (Take a look at the .CDE files in https://github.com/DSSAT/dssat-csm-os/tree/develop/Data).

At the moment Only the next crops and models are implemented:

Crop Model
Maize* CERES
Millet CERES
Rice CERES
Sugarbeet CERES
Sorghum* CERES
Sweetcorn CERES
Wheat* CERES
Alfalfa FORAGE-Alfalfa
Bermudagrass FORAGE-Bermudagrass
Soybean* CROPGRO
Canola CROPGRO
Sunflower CROPGRO
Tomato* CROPGRO
Cabbage CROPGRO
Potato SUBSTOR
Sugarcane CANEGRO

(*) Only a those crops have been validated. During the validation one DSSAT experiment was run using DSSATTools and the results were compared with those obtained using DSSAT desktop. I'll be validating more crops as long as a I have time to do it.

If you're interested in contributing to this project, don't hesitate in sending me an email (daquinterop@gmail.com).

All the Classes can be imported as:

from DSSATTools import (Crop, SoilProfile, Weather, Management, DSSAT)

or

from DSSATTools import *

DSSATTools.crop module

Crop is the only implemented class in the crop module. DSSAT's crop parameters are grouped in three different files: ecotype (.ECO), cultivar (.CUL) and species (.SPE). Not all crops have the ecotype file though. DSSATTools uses the default .SPE, .ECO, and .CUL files. The ecotype and cultivar parameters are defined as attributes of the Crop instance. Each parameter is accessible and can be modified using the key, value syntax, e.g. crop.cultivar["PARAMETER"] = VALUE.

It is well known that for a species there can be multiple cultivars. Therefore, when initializing a Crop instance, two parameters must be provided: the crop name (species), and the cultivar code. The cultivar codes are defined in the .CUL file. If an unknown cultivar is passed, then the last cultivar in the .CUL file is used and a warning is shown. To get a list of the available cultivars for a crop the user can use the DSSATTools.crop.available_cultivars function passing the crop name as only argument.

If the user wants to modify the cultivar or ecotype parameters they can be through the Crop.cultivar and Crop.ecotype attributes respectively. In these two attributes both the cultivar and ecotype parameters are defined as a Section class (DSSATTools.sections.Section). Section class simply maps the parameter's name to a value; it can be treated as a python dictionary. Each of the different sections of the Management class are defined in the same way.

The next example shows how to define the crop and modify one cultivar and ecotype parameter.

>>> crop = Crop('maize')
>>> crop.cultivar["P1"] = 240
>>> crop.ecotype["P20"] = 13.

DSSATTools.management module

This module hosts the Management class, which includes all the information related to management. There are multiple arguments to initialize a Management instance, however, the only mandatory argument is planting_date. If not provided, simulation start is calculated as the day before the planting date, emergence date is assumed to 5 days after planting, and the initial soil water content is assumed to be 50% of the total available water (PWP + 0.5(FC-PWP)).

Management class has one attribute per management section. Up to date not all of the sections have been implemented and the next sections are available for the user to modify: field, initial conditions, planting details, irrigation, fertilizers, harvest details, simulation controls, automatic management. All the sections are a DSSATTools.section.Sections object. The options that are not defined when initializing the Management instance can be defined by modifying the value of the parameters in each of the sections. An example will be set. If the user is not familiar to the different sections of the DSSAT experimental file then reviewing the DSSAT documentation is suggested.

DSSATTools.section.TabularSubsection class is intended to represent tabular information like irrigation schedules, fertilizer applications, or initial condition through the different soil's layers. The TabularSubsection can be initialized the same way a pandas.DataFrame. It's important to mention that the columns must have the same names as the DSSAT variables the are representing (See example).

In the next example a Management object is created, defining the irrigation method option as non-irrigated; then the location of the field is defined in the field section.

>>> man = Management( # Initialize instance
        planting_date=datetime(2020, 1, 1),
        irrigation="N",
    )
>>> # Modify the location of the field
>>> man.field["...........XCRD"] = 35.32
>>> man.field["...........YCRD"] = -3.21

Even though the irrigation method was defined when the object was created, it can still be modified:

>>> man.simulation_controls["IRRIG"] = "R"
>>> # Create a irrigation schedule as a pandas.DataFrame
>>> schedule = pd.DataFrame([
        (datetime(2000,1,15), 80),
        (datetime(2000,1,30), 60),
        (datetime(2000,2,15), 40),
        (datetime(2000,3,1),  20)
    ], columns = ['date', 'IRVAL'])
>>> schedule['IDATE'] = schedule.date.dt.strftime('%y%j')
>>> schedule['IROP'] = 'IR001' # irrigation operation code
>>> man.irrigation['table'] = TabularSubsection(
        schedule[['IDATE', 'IROP', 'IRVAL']]
    )

DSSATTools.run module

This module hosts the DSSAT class. That class is the simulation environment, so per each DSSAT instance there's a directory where all the necesary files to run the model are allocated. To run the model there are 3 basic steps:

  1. Create a new Dscsm instance.
  2. Initialize the environment by calling the setup() method.
  3. Run the model by calling the run() method. You can close the simulation environment by calling the close() method.

The model outputs are storage in the output attribute. Up to date the next output are available: PlantGro, Weather, SoilWat, SoilOrg.

DSSATTools.soil module

soil module includes the basic soil class SoilProfile. This class contains all the soil information necessary to run the DSSAT model. Each of the layers of the soil profile is a SoilLayer instance. After a SoilProfile instance is created, a new layer can added by calling the SoilProfile.add_layer method passing a SoilLayer object as argument. You can also use the SoilProfile.drop_layer to drop the layer at the specified depth.

SoilLayer class represents each layer in the soil profile. The layer is initialized by passing the layer base depth and a dict with the parameteters as argument. Clay fraction (SLCL) and Silt fraction (SLSI) are the only mandatory parameters when creating a layer, the rest of the parameters are estimated.

There are three basic ways of creating a SoilProfile object:

  1. Specify a .SOL file and Soil id. Of course, the soil id must match one of the profiles in the .SOL file.
>>> soilprofile = SoilProfile(
    file='SOIL.SOL',
    profile='IBBN910030'
)
  1. Passing a string code of one the available default soils.
>>> soilprofile = SoilProfile(
    default_class='SCL', # Silty Clay Loam
)
  1. Pasing a dict with the profile parameters (different from the layer pars). DSSAT.soil.list_profile_parameters function prints a detailed list of the layer parameters. And empty dict can be pased as none of the parameters is mandatory.
>>> soilprofile = SoilProfile(
    pars={
        'SALB': 0.25, # Albedo
        'SLU1': 6, # Stage 1 Evaporation (mm)
        'SLPF': 0.8 # Soil fertility factor
    }
)
>>> layers = [
    soil.SoilLayer(20, {'SLCL': 50, 'SLSI': 45}),
    soil.SoilLayer(50, {'SLCL': 30, 'SLSI': 30}),
    soil.SoilLayer(100, {'SLCL': 30, 'SLSI': 35}),
    soil.SoilLayer(180, {'SLCL': 20, 'SLSI': 30})
]
>>> for layer in layers: soilprofile.add_layer(layer)

That layer must be initialized with the texture information (‘SLCL’ and ‘SLSI’ parameters), or the hydraulic soil parameters (‘SLLL’, ‘SDUL’, ‘SSAT’, ‘SRGF’, ‘SSKS’, ‘SBDM’, ‘SLOC’). If a soil hydraulic parameter is not defined, then it’s estimated from soil texture using Pedo-transfer Functions. The previous parameters are the mandatory ones, but all the available parameters can be includedin the pars dict.

If you want to save your soil profile in .SOL a file, you can use the SoilProfile.write method. The only argument of this method is the filename.

For both classes any of the parameters can be modified after the initialization as each parameter is also an attribute of the instance.

>>> soilprofile = SoilProfile(
    pars={
        'SALB': 0.25, # Albedo
        'SLU1': 6, # Stage 1 Evaporation (mm)
        'SLPF': 0.8 # Soil fertility factor
    }
>>> # Modify the albedo of the created instance
>>> soilprofile.SALB = 0.36

DSSATTools.weather module

This module hosts the Weather class. It also contains the list_station_parameters and list_weather_variables which return a list of the parameters that define the weather station where the data was collected, and the weather variables that can be included in the dataset. A Weather instance is initialized by passing five mandatory parameters: a pandas dataframe including the weather data, a dict mapping each dataframe column to one of the DSSAT weather varaibles, latitude, longitude, and elevation.

The next example illustrates how to define a Weather instance from syntetic data:

>>> DATES = pd.date_range('2000-01-01', '2010-12-31'); N=len(DATES)
>>> df = pd.DataFrame(
        {
        'tn': np.random.gamma(10, 1, N),
        'rad': np.random.gamma(10, 1.5, N),
        'prec': np.round(np.random.gamma(.4, 10, N), 1),
        'rh': 100 * np.random.beta(1.5, 1.15, N),
        },
        index=DATES,
    )
>>> df['TMAX'] = df.tn + np.random.gamma(5., .5, N)
>>> weather = Weather(
        df, 
        {"tn": "TMIN", "rad": "SRAD", "prec": "RAIN", 
        "rh": "RHUM", "TMAX": "TMAX"},
        4.54, -75.1, 1800
    )

The parameters of the weather station are defined as attributes of the Weather class. Those parameters can be listed by calling the list_station_parameters. In the next example the reference height for windspeed measurements is defined for the weather instance created in the previous example:

>>> weather.WNDHT = 2