The best way to understand OMS and see how it works is probably to give a look to the material of our last winter School on our system GEOframe. Specifically, this is the page of introduction to OMS. But also starting from installations of the material can give insights.
There you can find links to the OMS v3 pages at CSU and original material. Our choice of OMS and Java was long meditated. This was my assessment almost ten years ago. This is instead the assessment made by NIWA six years ago.
Here you can find some thinking about the choice of programming languages. We use, in practice, a mix of Python and Java. Python for treating and visualizing data. Java for writing our models.
When properly used Java is not so slow (twice slower of well written C++) and was favored by having an incredible suite of tools helping proper programming, project building and code maintenance that make the experience to develop in C++ a poor one. Python (specifically some of its libraries) and the Jupyter lab are instead a very nice experience for treating data and, besides, is supported by a vibrant community of developers, even among hydrologists, who really make the experience better and better everyday. Python is also becoming a “lingua franca” among scientists and this makes less difficult and steep to share results. I have a couple of pages to introduce Python to hydrologists that you can access starting from here. For what regards of Java one can start from here.
In any case Java is not mandatory for the use of OMS. Also FORTRAN and C++ codes can be used: NIWA did it with FORTRAN, but we never tried it.
OMS code is distributed with the MIT Open Source license. It is not available on a public repository though, but Olaf David, its chief architect, gave us access to it and does not have problem to give it to others. I recognize that this is a limitation but with more users, it would not be so difficult to get a branch of the source code in a public repository, I guess.
Our own source code is at the GEOframe components Github repository. The software is under GPL3 and the repository is open to anyone willing to collaborate.
The structure of the OMS allows to easily add components with any type of licence, the one more suited to you, and does not oblige you to release the software under the GPLv3 license.
Worth to mention is that OMS has a server side companion, CSIP of which you can find any information here.
Once Olaf David mentioned to me that the overall cost of OMS was closely 10 million of dollars. From my side I guess i invested around 1 million and more on developing GEOframe and its ancestors, Just to give a measure of the investment needed for such an enterprise. But building on top of them is now much less expensive and much more immediate.
Our current task are finishing the 1D-2D-3D solving of Richards equation (extended to treat groundwater and surface waters) and its coupling with the energy budget, evaporation and vegetation dynamics. This would replace our older GEOtop in two years, except for snow modelling for which we do not have started anything so far. We have also ongoing work for hillslope stability and a well established set of tools for doing lumped modelling, including the estimation of travel times and tracers/pollutants concentrations. Some other developments, more on the side of informatics were done by a former students of mine and are reported in his Ph.D. Dissertation. It contains also some work on neural networks.
For the future, an effort should be made to bring OMS to Java 9, which, I think, could make the code more understandable. Classes are a good thing but understanding the connection between them is what matters. Java 9 Modules should help to get the code more clean, and Gradle should do the rest for external libraries. A few years go I started to dig into the code. I could not complete the task for my limited time but I believe one Java programmer can master OMS3 code in six intense months or so. I also envision that, having a good project, OMS could be completely embedded in Jupyter lab making it usable through it making its use more familiar to a multitude of users.
Obviously, we are willing to give any help, if you join us. I hope this information can help you to make your decisions.
There you can find links to the OMS v3 pages at CSU and original material. Our choice of OMS and Java was long meditated. This was my assessment almost ten years ago. This is instead the assessment made by NIWA six years ago.
Here you can find some thinking about the choice of programming languages. We use, in practice, a mix of Python and Java. Python for treating and visualizing data. Java for writing our models.
When properly used Java is not so slow (twice slower of well written C++) and was favored by having an incredible suite of tools helping proper programming, project building and code maintenance that make the experience to develop in C++ a poor one. Python (specifically some of its libraries) and the Jupyter lab are instead a very nice experience for treating data and, besides, is supported by a vibrant community of developers, even among hydrologists, who really make the experience better and better everyday. Python is also becoming a “lingua franca” among scientists and this makes less difficult and steep to share results. I have a couple of pages to introduce Python to hydrologists that you can access starting from here. For what regards of Java one can start from here.
In any case Java is not mandatory for the use of OMS. Also FORTRAN and C++ codes can be used: NIWA did it with FORTRAN, but we never tried it.
OMS code is distributed with the MIT Open Source license. It is not available on a public repository though, but Olaf David, its chief architect, gave us access to it and does not have problem to give it to others. I recognize that this is a limitation but with more users, it would not be so difficult to get a branch of the source code in a public repository, I guess.
Our own source code is at the GEOframe components Github repository. The software is under GPL3 and the repository is open to anyone willing to collaborate.
The structure of the OMS allows to easily add components with any type of licence, the one more suited to you, and does not oblige you to release the software under the GPLv3 license.
Worth to mention is that OMS has a server side companion, CSIP of which you can find any information here.
Once Olaf David mentioned to me that the overall cost of OMS was closely 10 million of dollars. From my side I guess i invested around 1 million and more on developing GEOframe and its ancestors, Just to give a measure of the investment needed for such an enterprise. But building on top of them is now much less expensive and much more immediate.
Our current task are finishing the 1D-2D-3D solving of Richards equation (extended to treat groundwater and surface waters) and its coupling with the energy budget, evaporation and vegetation dynamics. This would replace our older GEOtop in two years, except for snow modelling for which we do not have started anything so far. We have also ongoing work for hillslope stability and a well established set of tools for doing lumped modelling, including the estimation of travel times and tracers/pollutants concentrations. Some other developments, more on the side of informatics were done by a former students of mine and are reported in his Ph.D. Dissertation. It contains also some work on neural networks.
For the future, an effort should be made to bring OMS to Java 9, which, I think, could make the code more understandable. Classes are a good thing but understanding the connection between them is what matters. Java 9 Modules should help to get the code more clean, and Gradle should do the rest for external libraries. A few years go I started to dig into the code. I could not complete the task for my limited time but I believe one Java programmer can master OMS3 code in six intense months or so. I also envision that, having a good project, OMS could be completely embedded in Jupyter lab making it usable through it making its use more familiar to a multitude of users.
Obviously, we are willing to give any help, if you join us. I hope this information can help you to make your decisions.
No comments:
Post a Comment