Merge pull request #21 from TerrainBento/master

specify landlab version less than 2.0

README.md

@@ -4,7 +4,8 @@
 [![Coverage Status](https://coveralls.io/repos/github/TerrainBento/umami/badge.svg?branch=master)](https://coveralls.io/github/TerrainBento/umami?branch=master)
 [![Anaconda-Server Badge](https://anaconda.org/conda-forge/umami/badges/installer/conda.svg)](https://conda.anaconda.org/conda-forge)
 [![Binder](https://mybinder.org/badge_logo.svg)](https://mybinder.org/v2/gh/TerrainBento/umami/master?filepath=notebooks%2FWelcome.ipynb)
-[![status](https://joss.theoj.org/papers/1dea51d02def0ff3ef5108eb2a7af3a4/status.svg)](https://joss.theoj.org/papers/1dea51d02def0ff3ef5108eb2a7af3a4)
+[![DOI](https://joss.theoj.org/papers/10.21105/joss.01776/status.svg)](https://doi.org/10.21105/joss.01776)
+
 
 # What is it?
 
@@ -112,6 +113,6 @@ All contributions are welcome and appreciated. Feel free to:
 Contributors and maintainers to this project are are expected to abide the [Contributor Code of Conduct](https://github.com/TerrainBento/umami/blob/master/CODE_OF_CONDUCT.md).
 
 # Cite umami
-[![status](https://joss.theoj.org/papers/1dea51d02def0ff3ef5108eb2a7af3a4/status.svg)](https://joss.theoj.org/papers/1dea51d02def0ff3ef5108eb2a7af3a4)
+[![DOI](https://joss.theoj.org/papers/10.21105/joss.01776/status.svg)](https://doi.org/10.21105/joss.01776)
 
 Umami is described by a Journal of Open Source Software paper. If you use umami in your research, please cite it. 

environment-dev.yml

@@ -12,7 +12,7 @@ dependencies:
    # for the package
    - scipy
    - numpy
-   - landlab>=1.10.1
+   - landlab>=1.10.1, <2.0
    # for the notebooks
    - jupyter
    - holoviews

joss/paper.md

@@ -46,15 +46,15 @@ There is, however, no clear consensus regarding which set of simulated equivalen
 
 Umami is a package for calculating objective functions or objective function components for Earth surface dynamics modeling. It was designed to work well with models built with the Landlab Toolkit [@hobley2017creative] or with the `terrainbento` multi-model analysis [@barnhart2019terrain]. Umami is complementary to existing topographic analysis tools such as LSDTopoTools [@mudd2014statistical; @club2017geomorphometric], TopoToolbox [@schwanghart2010topo; @schwanghart2014topo] and the Topographic Analysis Kit [@forte2019tak]. Rather than performing topographic analysis, umami is used to distill model output into a form usable by model analysis methods such as sensitivity analysis, calibration, and validation.  
 
-Umami offers two primary classes: a [`Residual`](https://umami.readthedocs.io/en/latest/umami.residual.html#Residual)
-which represents the difference between model and data, and a [`Metric`](https://umami.readthedocs.io/en/latest/umami.metric.html)
-which is a calculated value on either model or data. The set of currently supported calculations are found in the [`umami.calculations`](https://umami.readthedocs.io/en/latest/umami.calculations.html) submodule. Both the `Metric` and `Residual` classes are designed to be fully specified through a YAML-style input-file or python Dictionary interface. Many different calculations can be accomplished through parameter specification. This supports reproducible analysis and systematic variation in metric construction. For example, when used with `terrainbento` one input file can describe the model run, and one input file can describe the model assessment or model-data comparison. This streamlines model analysis applications by making driver files more re-usable and by placing the code that accomplished calculations in the umami package rather than within the driver file. Umami also provides multiple output formats (YAML and Dakota), the latter of which is designed to interface with Sandia National Laboratory's Dakota package [@adams2019dakota].
+Umami offers two primary classes: a [`Residual`](https://umami.readthedocs.io/en/latest/umami.residual.html#Residual),
+which represents the difference between model and data, and a [`Metric`](https://umami.readthedocs.io/en/latest/umami.metric.html),
+which is a calculated value on either model or data. The set of currently supported calculations are found in the [`umami.calculations`](https://umami.readthedocs.io/en/latest/umami.calculations.html) submodule. Both the `Metric` and `Residual` classes are designed to be fully specified through a YAML-style input-file or python Dictionary interface. Many different calculations can be accomplished through parameter specification. This supports reproducible analysis and systematic variation in metric construction. For example, when used with `terrainbento`, one input file can describe the model run, and one input file can describe the model assessment or model-data comparison. This streamlines model analysis applications by making driver files more re-usable and by placing the code that accomplished calculations in the umami package rather than within the driver file. Umami also provides multiple output formats (YAML and Dakota), the latter of which is designed to interface with Sandia National Laboratory's Dakota package [@adams2019dakota].
 
 The novel contribution of the umami package is not primarily found in the specific calculations accomplished (e.g., some of them are as straightforward as the mean of a state variable). Instead it is the flexible and extensible nature of the input file format and the `Metric` and `Residual` classes. Additionally, the package can be extended through the addition of new calculation methods.
 
 Two model-data comparison metrics in umami are novel. First, the `joint_density_misfit` provides a misfit metric of the joint density of two state variables. This comparison measure was inspired by the use of channel $\chi$ index value [@perron2013integral] and topographic elevation to assess river channel long profiles ($\chi$-z plots). While it is not clear how to best quantify the difference between a modeled and observed $\chi$-z plot, it is straightforward to calculate the sum of squared residuals between the joint density of $\chi$ and topographic elevation for a modeled and observed landscape. In umami, this comparison is generalized to the joint density of two state variables.
 
-Second, the `discretized_misfit` calculation seeks to reduce the dimension of a state variable misfit. In some applications it is appropriate to make a direct comparison between a measured and modeled state variable (e.g., difference measured and modeled topography). However, if an application uses a model domain with tens of thousands of grid cells a user is faced with a challenging choice: either treat each residual as an individual observation or reduce all residuals to a single value through a sum of squared residuals (or other norm). The `discretized_misfit` calculation permits the identification of consistent "categories" of grid cells based on other grid fields. A sum of squared residuals is then calculated within each of these categories.
+Second, the `discretized_misfit` calculation seeks to reduce the dimension of a state variable misfit. In some applications it is appropriate to make a direct comparison between a measured and modeled state variable (e.g., difference measured and modeled topography). However, if an application uses a model domain with tens of thousands of grid cells, a user is faced with a challenging choice: either treat each residual as an individual observation or reduce all residuals to a single value through a sum of squared residuals (or other norm). The `discretized_misfit` calculation permits the identification of consistent "categories" of grid cells based on other grid fields. A sum of squared residuals is then calculated within each of these categories.
 
 # Acknowledgements
 

setup.py

@@ -28,5 +28,5 @@
     long_description_content_type="text/markdown",
     zip_safe=False,
     packages=find_packages(),
-    install_requires=["scipy", "numpy", "landlab>=1.10.1"],
+    install_requires=["scipy", "numpy", "landlab>=1.10.1, <2.0"],
 )