Model

This class allows to manipulate the datasets in an efficient manner. it contains the necessary functions to load nad pre-process the raw data, calculate GPP, iPAR, Stress Factor, Day Temperature and estimate grain yield.

Model(config, params=None)

Bases: object

load_raw_datasets()

Load raw phenology and AgERA5 datasets

Return

A existing dataset including WeatherFile, PhenoFile, NDVIFile. The raw data can be seen in config object. eg. config['PhenoFile']

preprocess_raw_datasets(data)

Preprocess raw datasets

Parameters:

Name	Type	Description	Default
data	dictionary	The phenology, ndvi and weather data for each location in example dataset.	required

Return

A dataset in a specific format for the tfunct package

getTDay(m=None, w=None, tminFactor=None)

Calculate day time temperature

Parameters:

Name	Type	Description	Default
m	object	A tfunct model with sites and the necessary information	None
w	array	Table of weather data with minimum and maximum temperature records	None
tminFactor	float	Minimum Temperature factor. Default is usually 0.25. It can be reviewed in configuration parameters such as TMIN_PERC_FACTOR.	None

Returns:

Type	Description
array	A number or array of Day Temperature

getIPAR(m=None, ndvi=None)

Total light interception - iPAR

Reference

iPAR = NDVI * 1.25 - 0.19 # between heading and maturity (Campos et al. 2018)

iPAR = NDVI * 1.25 - 0.21 Daughtry et al. (1992)

• Asrar, G., Fuchs, M., Kanemasu, E.T., Hatfield, J.L., 1984. Estimating absorbed photosynthetic radiation and leaf area index from spectral reflectance in wheat. Agron. J. 76, 30-306.

• Campos 2018 Remote sensing-based crop biomass with water or light-driven crop growth models in wheat commercial fields.

Parameters:

Name	Type	Description	Default
m	object	A tfunct model	None
ndvi	array	Array of float values	None

Returns:

Type	Description
array	An array of Total light interception values

getYield(tfun='PRFT', sites=None, is_VPDStress=False, expCSV=True)

Get yield using a defined temperature function and default parameters

Parameters:

Name	Type	Description	Default
tfun	str	Name of the temperature function	'PRFT'
sites	dictionary	List of sites with the basic information to estimate yield.	None
is_VPDStress	bool	Vapor Pressure Deficit stress. Default is False.	False
expCSV	bool	Export grain yield results in an individual file. Default is True.	True

Returns:

Type	Description
array	A dataframe with simulated grain yield values

runCombinations_PRFT(data=None, sites=None, comb=None, is_VPDStress=False, expCSV=True, individualCSV=False, target=None)

Run several combinations using different parameters for PRFT model

Deprecated

Stop using this function.

Parameters:

Name	Type	Description	Default
data	object	Array of base data.	None
sites	object	Array of sites	None
comb	object	Array of combinations	None
is_VPDStress	bool	Vapor Pressure Deficit stress. Default is False	False
expCSV	bool	Save results in CSV format. Default is True	True
individualCSV	bool	Default is False	False
target	str	Name of the target device. Available values are cpu, parallel and cuda. Default is cpu	None

Returns:

Type	Description
object	A dataframe with estimated grain yield of all combinations.

fit(data=None, sites=None, comb=None, ft='PRFT', is_VPDStress=False, expCSV=True, individualCSV=False, n_jobs=4, verbose=False)

Run a model to fit yield

Deprecated on Dec, 2022.

Stop using this function.

Parameters:

Name	Type	Description	Default
data	object	Array of base data.	None
sites	object	Array of sites	None
comb	object	Array of combinations	None
ft	str	Name of the temperature response function. Default is 'PRFT'.	'PRFT'
is_VPDStress	bool	Vapor Pressure Deficit stress. Default is False	False
expCSV	bool	Save results in CSV format. Default is True	True
individualCSV	bool	Default is False	False
n_jobs	int	Number of CPU cores to use in paralell processing. Default is 4.	4
verbose	bool	Display comments during processing. Default is False	False

Returns:

Type	Description
array	An array of sites with intermediate results

setup_dataInput_forCombinations(sites)

Set up data for each sites to estimate grain yield from different combinations

Parameters:

Name	Type	Description	Default
sites	dictionary		required

Return

Array of arrays with weather, iPAR and GPP data

getCombinations(functype='PRFT', cols=None, RUE=None, Tmin=None, Toptmin=None, Topt=None, Toptmax=None, Tmax=None, TminFactor=None, Lvpd=None, Uvpd=None, SFvpd_Lthres=None, SFvpd_Uthres=None, isVPDStress=False)

Generate combinations to establish the optimum temperature response for grain-filling period and estimate grain yield.

Parameters:

Name	Type	Description	Default
functype	str	Type of temperature response function (eg. PRFT, WETF, TPF)	'PRFT'
cols	int	Number of daily records in growing period	None
RUE	array	Array of RUE values	None
Tmin	array	Array of minimum temperature values	None
Toptmin	array	Array of Optimun minimum temperature values	None
Topt	array	Array of Optimun temperature values	None
Toptmax	array	Array of Optimun maximum temperature values	None
Tmax	array	Array of maximum temperature values	None
TminFactor	array	Array of minimum temperature factor	None
Lvpd	array	Array of lower VPD values	None
Uvpd	array	Array of lower VPD values	None
SFvpd_Lthres	array	Array of lower VPD stress factor values	None
SFvpd_Uthres	array	Array of lower VPD stress factor values	None
isVPDStress	bool	True/False value for using VPD stress condition	False

Return

Array of combinations and an array to save results

getGYield_forCombinations(functype, df_GYield, data_input, array_params_to_run, isVPDStress, array_results, saveFile=True, returnDF=True, fmt='parquet')

Estimate grain yield using parameters for each combinations

Parameters:

Name	Type	Description	Default
df_GYield	array	Dataframe with observed grain yield	required
data_input	array	Array of inputs containing weather, iPAR, PRFT and GPP data for each site.	required
array_params_to_run	array	Array of combinations (RUE, TOpt and TminFactor) to simulate	required
isVPDStress	bool	True/False value for using VPD stress condition	required
array_results	array	Array used to save results	required
saveFile	bool	True if save file in results folder	True
returnDF	bool	True if return a pandas dataframe. Use False when the combinations are too large.	True
fmt	str	Format to save file. Comma separate value (csv) or Apache parquet (parquet). Default is parquet	'parquet'

Return

A dataframe or table with results

getCombinations_Metrics(functype, isVPDStress, df_GYield, array_params_to_run, array_results, saveFile=True, fmt='parquet')

Get evaluation metrics for each simulation

Parameters:

Name	Type	Description	Default
functype	str	Type of temperature response function (eg. PRFT, WETF, TPF)	required
isVPDStress	bool	True/False value for using VPD stress condition	required
df_GYield	array	Dataframe with observed grain yield	required
array_params_to_run	array	Array of combinations (RUE, TOpt and TminFactor) to simulate	required
array_results	array	Array used to save results	required
saveFile	bool	True if save file in results folder	True
fmt	str	Format to save file. Comma separate value (csv) or Apache parquet (parquet). Default is parquet	'parquet'

Return

A dataframe with several metrics

Returns:

Name	Type	Description
MAE	float	Mean Absolute Error
MSE	float	Mean Squared Error
RMSE	float	Root Mean Squared Error
RMSRE	float	Root Mean Squared Relative Error
MAPE	float	Mean Absolute Percentage Error
pvalue	float	p-value
R2	float	R Squared metric
EF	float	Nash-Sutcliffe metric
intercept	float	Intercept of the regression model
slope	float	Slope of the regression model
Cb	float	A bias correction factor
CCC	float	Concordance correlation coefficient
Accuracy	float	Accuracy in percentage

gpp

calcGPP(SolRad, TF, iPAR, RUE=3.0, stressFactor=1.0)

Estimate the Gross primary production

Parameters:

Name	Type	Description	Default
SolRad	float	Solar Radiation	required
TF	float	Temperature function result (PRFT, WETF, TPF)	required
iPAR	float	the photosynthetically active radiation (PAR) intercepted by a plant or crop estimated from NDVI	required
RUE	float	Radiation-use efficiency. Default value is 3.0 gMJ^−1	3.0
stressFactor	float	Stress Factor (eg. VPD stress factor)	1.0

Deprecated.

This function only use one value for each paramters. it is not optimized for run models in parallel.

Returns:

Type	Description
float	the gross primary production (GPP)

apply_GPP_v0(SolRad, PRFT, iPAR, RUE=3.0, stressFactor=1.0)

calculateGPP(SolRad, PRFT, iPAR, RUE=3.0, stressFactor=1.0)

Estimate the Gross primary production. The accumulated dry matter from heading to maturity

# GPP = Solar Radiation × TemFun × VPD_StressFactor × iPAR_fromNDVI × RUE 
GPP = SolRad * 0.5 * RUE * PRFT * iPAR * stressFactor

Parameters:

Name	Type	Description	Default
SolRad	float	Solar Radiation	required
PRFT	float	Photosynthesis reduction factor	required
iPAR	float	the photosynthetically active radiation (PAR) intercepted by a plant or crop	required
RUE	float	Radiation-use efficiency. Default value is 3.0 gMJ^-1	3.0
stressFactor	float	Stress Factor	1.0

Returns:

Type	Description
float	a number or array of values with the gross primary production (GPP)

applyGPP_VPDStress(SolRad, PRFT, iPAR, SFvpd, RUE=3.0)

calculateGPP_VPDStress(SolRad, PRFT, iPAR, SFvpd, RUE=3.0)

Estimate the Gross primary production with stressed VPD

GPP = SolRad * 0.5 * RUE * PRFT * iPAR * SFvpd

Parameters:

Name	Type	Description	Default
SolRad	float	Solar Radiation	required
PRFT	float	Photosynthesis reduction factor	required
iPAR	float	the photosynthetically active radiation (PAR) intercepted by a plant or crop	required
RUE	float	Radiation-use efficiency. Default value is 3.0	3.0
SFvpd	float	Vapor Pressure Deficit (VPD) stress factor	required

Returns:

Type	Description
float	a number or array of values with the gross primary production (GPP)

estimate(data, params, is_VPDStress, tf, results)

An optimized function for estimating GPP for all observations using one of the three Temperature Function (TF) such as PRFT, WETF and TPF.

Parameters:

Name	Type	Description	Default
data	object	Array of arrays containing tn, tx, ndvi, solrad, VPDx, ipar, GPP datasets	required
params	object	An array with RUE, tminFactor, Topt, Lvpd, Uvpd, SFvpd_Lthres, SFvpd_Uthres values	required
is_VPDStress	array	Array of one value with T/F if Vapor Pressure Deficit stress affecting grain yield. Default is [False].	required
tf	array	Array of one value representing the type of function. [1]: PRFT; [2]: WETF; [3]: TPF.	required
results	array	Empty array for outputs.	required

Returns:

Name	Type	Description
results	array	An array with estimated yield for each site.

gyield

estimate(data, params, is_VPDStress, tf, results)

An optimized function for estimating grain yield in one step for all observations using one of the three Temperature Function (TF) such as PRFT, WETF and TPF.

Parameters:

Name	Type	Description	Default
data	object	Array of arrays containing tn, tx, ndvi, solrad, VPDx, ipar, GPP datasets	required
params	object	An array with RUE, tminFactor, Topt, Lvpd, Uvpd, SFvpd_Lthres, SFvpd_Uthres values	required
is_VPDStress	array	Array of one value with T/F if Vapor Pressure Deficit stress affecting grain yield. Default is [False].	required
tf	array	Array of one value representing the type of function. [1]: PRFT; [2]: WETF; [3]: TPF.	required
results	array	Empty array for outputs.	required

Returns:

Name	Type	Description
results	array	An array with estimated yield for each site.

ipar

calcIPAR(ndvi)

Total light interception - iPAR

Reference

iPAR = NDVI * 1.25 - 0.19 # between heading and maturity (Campos et al. 2018)

iPAR = NDVI * 1.25 - 0.21 Daughtry et al. (1992)

• Asrar, G., Fuchs, M., Kanemasu, E.T., Hatfield, J.L., 1984. Estimating absorbed photosynthetic radiation and leaf area index from spectral reflectance in wheat. Agron. J. 76, 30-306.

• Campos 2018 Remote sensing-based crop biomass with water or light-driven crop growth models in wheat commercial fields.

Deprecated.

Stop using this function. This function estimate iPAR using only one value for each parameters.

Parameters:

Name	Type	Description	Default
ndvi	array	Array of float values	required

Returns:

Type	Description
array	An array of Total light interception values

estimate_IPAR(ndvi=None)

Total light interception - iPAR.

It is assumed that NDVI at maturity is 0.25.

Reference

iPAR = NDVI * 1.25 - 0.19 # between heading and maturity (Campos et al. 2018)

iPAR = NDVI * 1.25 - 0.21 Daughtry et al. (1992)

• Asrar, G., Fuchs, M., Kanemasu, E.T., Hatfield, J.L., 1984. Estimating absorbed photosynthetic radiation and leaf area index from spectral reflectance in wheat. Agron. J. 76, 30-306.

• Campos 2018 Remote sensing-based crop biomass with water or light-driven crop growth models in wheat commercial fields.

Parameters:

Name	Type	Description	Default
ndvi	array	Array of float values	None

Returns:

Type	Description
array	An array of Total light interception values

prft

calcPRFT(TDay, TOpt=18)

Estimate Photosynthesis reduction factor (PRFT)

PRFT = 1 - 0.0025 * (TDay - TOpt)^2

Deprecated.

Stop using this function. This function estimate PRFT using only one value for each parameters.

Parameters:

Name	Type	Description	Default
TDay	float	Number or array of Day Temperatures	required
TOpt	float	Optimum Temperature. Default value 18	18

Returns:

Type	Description
float	A number or array of PRFT

calculatePRFT(Tday, Topt=18)

Estimate Photosynthesis reduction factor (PRFT) in parallel.

PRFT = 1 - 0.0025 * (TDay - TOpt)^2

Parameters:

Name	Type	Description	Default
Tday	float	Number or array of Day Temperatures	required
Topt	float	Optimum Temperature. Default value 18	18

Returns:

Type	Description
float	A number or array of PRFT

sfvpd

calcSFvpd(VPDx, Lvpd, Uvpd, SFvpd_Lthres, SFvpd_Uthres)

getSFvpd(VPDx, Lvpd, Uvpd, SFvpd_Lthres, SFvpd_Uthres)

apply_SFvpd(VPDMAX, Lvpd=1, Uvpd=4, SFvpd_Lthres=0.2, SFvpd_Uthres=1)

calculateSFvpd(VPDMAX, Lvpd=1, Uvpd=4, SFvpd_Lthres=0.2, SFvpd_Uthres=1)

Calculation of Vapor pressure deficit (VPD) stress factor

Parameters:

Name	Type	Description	Default
VPDMAX	array	Array of daily temperature values	required
Lvpd	float	A number for threshold of lower VPD. Default is 1	1
Uvpd	array	A number for threshold of upper VPD. Default is 4	4
SFvpd_Lthres	array	A number for threshold of stress factor of lower VPD. Default is 0.2	0.2
SFvpd_Uthres	array	A number for threshold of stress factor of upper VPD. Default is 1	1

Returns:

Type	Description
array	A number or array of stressed factors of VPD

tday

calcTDay(Tmin, Tmax, tminFactor=0.25)

Calculate day time temperature. TDay is a function of weighted Tmin and weighted Tmax.

 TDay = 0.75 * Tmax + 0.25 * TMin

Deprecated.

Stop using this function. This function estimate day time temperature (TDay) using only one value for each parameters.

Parameters:

Name	Type	Description	Default
Tmin	float	Number or array of Minimum Temperatures	required
Tmax	float	Number or array of Maximum Temperatures	required
tminFactor	float	Minimum Temperature factor	0.25

Returns:

Type	Description
float	A number of Day Temperatures

getTDay(Tmin, Tmax, tminFactor=0.25)

apply_TDay(Tmin, Tmax, tminFactor)

estimate_TDay(Tmin=None, Tmax=None, tminFactor=0.25)

An optimized function to calculate day time temperature in parallel.

 TDay = 0.75 * Tmax + 0.25 * TMin

Parameters:

Name	Type	Description	Default
Tmin	float	Number or array of Minimum Temperatures	None
Tmax	float	Number or array of Maximum Temperatures	None
tminFactor	float	Minimum Temperature factor	0.25

Returns:

Type	Description
float	A number of Day Temperatures