Driven by Francesco and Olaf, one step in surrogate modelling

 This is one of the paper to which I was more spectator than a real player. However, my name appears among the contributors because I was somewhat crucial for the project to succeed. The idea that bother many is that hydrological models aren too complicate and we need more simple models that get almost the same results. This is the rational for this work that was part of Francesco Serafin Ph.D. Thesis. 

The main ideas was to get into OMS3/CSIP framework an artificial neural network (ANN) system that could surrogate the hydrological models. Surrogate means that it can reproduce most of the dynamic features of the model training the ANN and then making using it. The roadmap is simple but the realization involves several steps that the paper delineates.

So, here it is our paper and its abstract: Serafin, Francesco, Olaf David, Jack R. Carlson, Timothy R. Green, and Riccardo Rigon. 2021. “Bridging Technology Transfer Boundaries: Integrated Cloud Services Deliver Results of Nonlinear Process Models as Surrogate Model Ensembles.” Environmental Modelling and Software[R], no. 105231 (October): 105231.

Environmental models are often essential to implement projects in planning, consulting and regulatory institutions. Research models are often poorly suited to such applications due to their complexity, data requirements, operational boundaries, and factors such as institutional capacities. This contribution enhances a modeling framework to help mitigate research model complexity, streamline data and parameter setup, reduce runtime, and improve model infrastructure efficiency. Using a surrogate modeling approach, we capture the intrinsic knowledge of a conceptual or process-based model into an ensemble of artificial neural networks. The enhanced modeling framework interacts with machine learning libraries to derive surrogate models for each model service. This process is secured using blockchain technology. After describing the methods and implementation, we present an example wherein hydrologic peak discharge provided by the curve number model is emulated with a surrogate model ensemble. The ensemble median values outperformed any individual surrogate model fit to the curve number model. 

Francesco Serafin Ph.D. Defense and Thesis

Francesco Serafin is not anymore a Ph.D. student bu a doctor! His work concerned mainly the structure of the Object Modelling System Version 3 especially in two directions: improving the researchers modelling experience and improving the end users experience. You can learn directly from his video how.

The defense had a more than forty minutes of presentation and the same period of discussion. In order, you can find below, the presentation, the YouTube video of the the defense and the discussion (the latter in Italian).

• The presentation (by clicking on the Figure):

• The Video of the defense

The Discussion (In Italian)

• Last but not least, the Dissertation is here. 

On Complex Network Representation and Computation of Hydrological Quantities

Francesco Serafin participated for us to the AGU fall meeting presenting part of his work. They gave him a poster, actually the poster whose image is below. It is about the representation of Hydrological models as a network of interactions.

You can access a more readable version of the poster by clicking on the Figure. A similar, more recent presentation, given for IEMSS 2018 at Fort Collins can be found below:

An ML based meta modelling infrastructure for environmental models

This is the presentation Francesco gave for his admission to the third year of Ph.D. studies. He summarizes his work done so far and foresees his work during the next year.

Francesco's work is a keystone of the work in our group, since he sustains most of informatics and pur commitment to OMS3. Besides of this two are his major achievements: the building of the Ne3 infrastructure (an infrastructure inside an infrastructure!)  which allows an enormous flexibility to our modelling, and the new road opened towards modeling discharges through machine learning techniques. But there are other connections he opens that are visible through his talk. Please clisk on the figure to access the presentation.

GEOtop was dockerized during last week

Dear GEOtop aficionados. Francesco Serafin with the help of Olaf David fully 'dockerized' GEOtop during last week: https://hub.docker.com/r/omslab/geotop

You can run it now on Linux, Mac, Windows, Amazon, Google, via this Docker image. There is description on how to use it, pretty simple. This is exciting, since it makes building the model on all platform now obsolete and you can get it out easily to more collaborators. In our opinion this is the best way to disseminate a model like this. Please spread the word and have people using it.

If you want to know what Docker is you can click on the Docker logo above.

About graphs, DSL and replicable research in Francesco Serafin's work

This is the summary of what Francesco Serafin (his blog) did in its first year of doctoral studies, defending for his admission to his second year Ph.D. Undoubtedly he did a lot of work and he programs to do even more. Three are the lines of his research:

• implementing a new flexible structure based on graphs for commanding simulations of interacting systems; 
• implementing a domain specific language for doing environmental models (and particularly to solve ordinary and partial differential equations); 
• deploying a system that makes easier to do replicable science. 

Clicking on the Figure above you will find his presentation. Obviously the committee admitted him to the second year of his Ph.D. studies.

The first taste of Ne3 (netree) (at IEMSs in Toulouse)

We had this presentation, this morning at IEMSs, about some of the many lines of work where Francesco Serafin (his blog here) is developing his Ph.D research. The origin of the work is that modelling basins is to model hillslopes (or Hydrologic Response Units, HRUs) connected by a river network. Most of what happens in HRUs is independent from what happens in the others. Some instead is aggregated by river network. So why do not use this natural structure to organise modelling and processes (and send them in parallel, when possible, using the tree structure) ?

Actually this can grow more, since many systems (a plant, ecological interactions, hydrological interactions), can be seen described over a graph of interactions. So the Ne3 can become a more general way to investigate natural systems. However, you can enjoy the presentation, by clicking on the figure above.  Ne3 source code written so far in on Github.

Patterns for the application of modern informatics to the integration of PDEs: the case of the Boussinesq Equation

Today Francesco Serafin graduated finishing his master in civil and environmental  engineering. In brief, the scope of his thesis was to implement a series of classes, eventually ported to OMS, to solve the groundwater Boussinesq equation (under the link, please find also specific reference to previous work), but with a more large scope to envision an object oriented structure which could work for any PDE. Let Francesco's introduction talk:

"Mathematical models play a fundamental role in many scientific and engineering fields in today’s world. They are used for example in geotechnics to evalute the hillslope stability, in weather science to predict weather trends and produce weather reports, in structural design to study the resistance to stress, and in fluid dynamics to compute fluid flows and air flows.

Consequently mathematical models are evolving all the time: more and more new numerical methods are being invented to solve the Partial Differential Equations (PDE)s that describe physical problems with increasing precision, and more and more complex and efficient processor units are being created to reduce the computational time.
Therefore, the code into which the mathematical models are translated has to be “dynamic” in order to be easily updated on the basis of the continuous developments (Formetta et al. (2014) [16]).
On the other hand, completely different physical problems are often de- scribed using similar PDEs. For this reason, the numerical methods which provide solutions to different problems can be the same. This suggest the implementation of an IT infrastructure that hosts a standard structure for solving PDEs and that can serve various disciplines with the minimum of hassles.

This work is focused on the application of what is envisioned above, with the main purpose of the creation of an abstract code for implementing every type of mathematical model described by PDEs.

We work on hydrological topics but we hope to design a structure of general interest. Obviously the final goal of any work of this type is to find a proper numerical solver, and therefore, part of the thesis is devoted to the analysis of the problem under scrutiny, and the description of the solution found."

The thesis can be found here. The presentation given at the thesis discussion is here.  The code can be found here.