yonggenzhang / rosetta-soil Goto Github PK

This package provides an implementation of Rosetta, a neural network-based model for predicting unasturated soil hydraulic parameters from basic soil characterization data.

For most Rosetta use cases, we recommend using the web browser interface to rosetta-soil that is available at https://www.handbook60.org/rosetta

There is also an api available at handbook60.org. For example:

import requests

data = {
    "soildata": [
        [30, 30, 40, 1.5, 0.3, 0.1],
        [20, 60, 20],
        [55, 25, 20, 1.1],
    ]
}

def url(rosetta_version: int) -> str:
    return f"http://www.handbook60.org/api/v1/rosetta/{rosetta_version}"

r = requests.post(url(3), json=data)

returns the following:

print(r.json())

{'model_codes': [5, 2, 3], 'rosetta_version': 3, 'stdev':
[[0.013628985468838103, 0.01496917525020338, 0.12948704319399928,
0.0347739236276485, 0.1747797749074611], [0.0067075928037543765,
0.00878582437678383, 0.07413139323912403, 0.013230683219936165,
0.08709445948355408], [0.01277140708670187, 0.013062170576228887,
0.10020312250396954, 0.01763982447621485, 0.14163566888667592]],
'van_genuchten_params': [[0.06872133198419336, 0.38390508534751433,
-2.452968871563431, 0.17827394547955497, 0.9827227259550619],
[0.08994502219206943, 0.4301366480210401, -2.4262357492034043,
0.17568732926631986, 1.192731130984082], [0.09130753033144606,
0.485031958049669, -2.0223880878467875, 0.151071612216524,
1.9060147751706147]]}

See below for information on the expected structure and content of the submitted soildata and returned json payload.

[If your use case involves, e.g., thousands of repeated requests, then please install and use rosetta-soil locally rather than use the api.]

pip install rosetta-soil

>>> from rosetta import rosetta, SoilData

>>> data = [
        [30,30,40,1.5,0.3,0.1],
        [20,60,20],
        [55,25,20,1.1]
    ]

>>> mean, stdev, codes = rosetta(3, SoilData.from_array(data))

>>> print(mean)
[[ 0.06872133  0.38390509 -2.45296887  0.17827395  0.98272273]
[ 0.08994502  0.43013665 -2.42623575  0.17568733  1.19273113]
[ 0.09130753  0.48503196 -2.02238809  0.15107161  1.90601478]]

>>> print(stdev)
[[ 0.01362899 0.01496918 0.12948704 0.03477392 0.17477977]
[ 0.00670759 0.00878582 0.07413139 0.01323068 0.08709446]
[ 0.01277141 0.01306217 0.10020312 0.01763982 0.14163567]]

>>> print(codes)
[5 2 3]

The Rosetta pedotransfer function predicts five parameters for the van Genuchten model of unsaturated soil hydraulic properties

• theta_r : residual volumetric water content
• theta_s : saturated volumetric water content
• log10(alpha) : retention shape parameter [log10(1/cm)]
• log10(n) : retention shape parameter
• log10(ksat) : saturated hydraulic conductivity [log10(cm/d)]

Rosetta provides four models for predicting the five parameters from soil characterization data. The models differ in the required input data

where

• sa, si, cl are percentages of sand, silt and clay
• BD is soil bulk density (g/cm3)
• th33 is the soil volumetric water content at 33 kPa
• th1500 is the soil volumetric water content at 1500 kPa

Three versions of Rosetta are available. The versions effectively represent three alternative calibrations of the four Rosetta models. The references that should be cited when using Rosetta versions 1, 2, and 3 are, respectively:

[1] Schaap, M.G., Leij, F.J., and Van Genuchten, M.T. 2001. ROSETTA: a computer program for estimating soil hydraulic parameters with hierarchical pedotransfer functions. Journal of Hydrology 251(3-4): 163-176. doi: 10.1016/S0022-1694(01)00466-8

[2] Schaap, M.G., A. Nemes, and M.T. van Genuchten. 2004. Comparison of Models for Indirect Estimation of Water Retention and Available Water in Surface Soils. Vadose Zone Journal 3(4): 1455-1463. doi: 10.2136/vzj2004.1455

[3] Zhang, Y. and Schaap, M.G. 2017. Weighted recalibration of the Rosetta pedotransfer model with improved estimates of hydraulic parameter distributions and summary statistics (Rosetta3). Journal of Hydrology 547: 39-53. doi: 10.1016/j.jhydrol.2017.01.004

from rosetta import rosetta, SoilData

The imported function rosetta predicts soil hydraulic parameters from soil characterization data. It has two required arguments:

rosetta_version : int, {1, 2, 3}
soildata : SoilData

The second argument is a SoilData instance. Normally, the instance is created from an array-like collection of soil characterization data using the from_array method.

data = [
    [30,30,40,1.5,0.3,0.1],
    [20,60,20],
    [55,25,20,1.1]
]
soildata = SoilData.from_array(data)

Each element of the array-like data contains soil data in this order:

[%sand, %silt, %clay, buld density, th33, th1500]

Sand, silt, and clay are required; the others are optional. For each entry, rosetta selects the best availabe Rosetta model based on the given data. Note that even if you are predicting for only a single soil record, data still needs to 2D array-like:

data = [[30,30,40]]
soildata = SoilData.from_array(data)

The function rosetta returns a 3-tuple

mean, stdev, codes = rosetta(3, soildata)

mean is a 2D numpy array. The ith row holds predicted soil hydraulic parameters for ith entry in soildata. The array columns are

stdev is 2D numpy array holding the corresponding parameter standard deviations.

codes is a 1D numpy array with the ith entry indicating the Rosetta model and input data used to predict the ith row of mean and stdev.

Predictions can also be made using the Rosetta class

import numpy as np
from rosetta import Rosetta

The class is instantiated for a particular Rosetta version and model. Predictions are then made using a numpy array of soil data.

rose33 = Rosetta(rosetta_version=3, model_code=3)
data = np.array([[30,30,40,1.5],[55,25,20,1.1]], dtype=float)
mean, stdev = rose33.predict(data)

The 2D numpy array data has to be data.shape[1] = model_code + 1. Compared with the function rosetta.rosetta, Rosetta.predict offers fewer checks on arguments and data.

This module wraps files taken from research code developed by Marcel Schaap and Yonggen Zhang at the University of Arizona.

The Rosetta class described above has another method, Rosetta.ann_predict, which returns additional statistical quantities computed by the Schaap and Zhang code and which may be of interest to researchers. The usage is the same as Rosetta.predict,

rose33 = Rosetta(rosetta_version=3, model_code=3)
data = np.array([[30,30,40,1.5],[55,25,20,1.1]], dtype=float)
results = rose33.ann_predict(data, sum_data=True)

However, in this case, the returned results is a dictionay of parameters and statistical results. Note the arrays in results are the transpose of what is returned by other functions and methods in rosetta-soil See the file ANN_Module.py and the code base of Schaap and Zhang for more information.