Yonggen Zhang
Subsurface Hydrology Computational Group



Comparison of van Genuchten and Ks parameters estimated using Rosetta1-H2 and Rosetta3-H2

The following link provides you the BETA version of the Rosetta3 code described in Zhang and Schaap, (2017). The original version of Rosetta code described in Schaap et al. (2001) was released for Windows XP system, and may not usable in current Operating Systems. Therefore, both Rosetta 1 and Rosetta 3 code are implemented in our current version in the form of open-source code with a long-term support. Please find the the following manual for details.

This cdoe is intended for users/developers familiar with the python programming language, and it works in Linux, Windows 7+, and Mac OSX, independent of Operating System. In addition, a web-based interface is under development.

Please download the code and manual here:

Python code


Please note that:

1. The code has been modified to be compatible with python 2.7, python 3.0, and higher versions (updated in May 2020).

2. There are two sets of models: 2, 3, 4, 5 and 102, 103, 104, 105. The models 2, 3, 4, 5 are NEW and better (see Zhang and Schaap, 2017). The latter series are the OLD (Schaap et al., 2001) models. The soil textural class averages models (model 1, i.e., Rosetta-v3 H1 model) is NOT included in the current beta version, but can be downloaded from here for mean and standard diviation values.
Please not that, in the "H1_new_sd.xls", alpha, n, and, Ks columns are standard deviation of log10 transformed values. Since log10 transformed alpha, n, and Ks values belong to the normal distribution, it is meaningful to calculate the mean values and standard deviation based on the log10 transformed quantities. In "H1_new.xls", alpha, n, and, Ks had been back-transformed by raising 10 to the power of mean values.

3. To make an estimation using the code, please make sure that you are using the correct data input format and correct Rosetta model, i.e., 2, 3, 4, 5, 102, 103, 104, or 105. Please see the above manual for details.

A web-based interface of the Rosetta model developed by Todd Skaggs (USDA-ARS) and Ehsan Ghane (Michigan State University) can be accessed, respectively, by https://www.handbook60.org/rosetta/ and https://dsiweb.cse.msu.edu/rosetta/.


The rosetta-soil python package by Todd Skaggs (USDA-ARS) is now on pypi https://pypi.org/project/rosetta-soil/ and can be installed with pip install rosetta-soil. The project page with instructions is at https://github.com/usda-ars-ussl/rosetta-soil..


The R version of the Rosetta3 code was developed by Dr. Todd Skaggs (USDA-ARS) and you may find the details of how to install, make a prediction, and make graphs of the Rosetta model using R code by accessing http://ncss-tech.github.io/AQP/soilDB/ROSETTA-API.html.

Kosugi code

Global map of (a) residual water content, (b) saturated water content, (c) log10(hm), (d) standard deviation of log10(hm), (e) log10(sigma), (f) saturated hydraulic conductivity (log10(Ks)), (g) field capacity (h = 330 cm), and (h) plant available water (field capacity minus permanent wilting point) in 1 km resolution estimated from Kosugi K3 model (using sand, silt, clay percentage, and bulk density as input). Calculations are based on the surface soil of SoilGrids 1 km data set (Hengl et al., 2014)

Global maps of Kosugi parameters, derived quantities including derived field capacity, plant available water and associated uncertainties in 1 km resolution for the surface soils can be downloaded from


Related paper: Yonggen Zhang, Marcel G. Schaap, and Yuanyuan Zha. (2018). A High-Resolution Global Map of Soil Hydraulic Properties Produced by a Hierarchical Parameterization of a Physically-Based Water Retention Model, Water Resources Research, 54. https://doi.org/10.1029/2018WR023539

The code is coming soon.

Multi-model Ensemble Predictions

Global maps of multi-model ensemble predictions of field capacity and wilting points are coming soon.