Thursday, April 3, 2025

A Ph.D. position on Advancing Physics-Informed Machine Learning for Environmental Modeling and Smart Irrigation Systems

Project Overview

This PhD grant, funded by Fondazione Bruno Kessler, aims to develop an advanced Physics-Informed Machine Learning (PIML) framework for modeling complex hydrodynamic and environmental systems. By integrating physical principles with data-driven methods, the research will focus on optimizing next-generation irrigation strategies. The project will harness the synergy between physical laws (as implemented in the GEOSPACE system) and machine learning to enable predictive, real-time, and scalable modeling tools for sustainable water resource management.

Key Objectives
  • Design hybrid PIML models that combine governing equations with data-driven predictive models (e.g., neural networks);
  • Improve predictive accuracy and generalizability across heterogeneous environmental conditions;
  • Incorporate real-time sensor data inputs to refine model states and parameters;
  • Benchmark PIML approaches against traditional numerical solvers and/or black-box machine learning models.

Methodological Approach

The core innovation of this project lies in the integration of physical constraints (as derived from GEOSPACE) into machine learning models. Building on recent advances in PIML, the goal is to design, develop, and validate models that enforce conservation laws and boundary conditions within neural architectures. This may involve:

  • Embedding partial differential equations (PDEs) directly into the loss functions of machine/deep learning models;
  • Developing adaptive training strategies to trade-oO data fidelity and physical consistency;
  • Utilizing sensor data to dynamically assimilate environmental variability into model predictions;
  • Leveraging high-performance computing to train and deploy models at scale across complex
  • domains.


Expected Outcomes

This integration will enable:
  • Accurate and efficient modeling of water distribution and use in precision irrigation systems;
  • Real-time monitoring and decision-support capabilities for agricultural and environmental applications;
  • Enhanced data efficiency and model robustness through physics-based regularization;
  • Improved understanding of system dynamics under data-scarce and/or non-stationary conditions.

Implementation Timeline

Months 1–6: Literature review on PIML methodologies;
Months 7–18: Design and develop a core PIML architecture, integrating IoT data sources;
Months 19–24: Validate models using lab-scale and field experimental datasets;
Months 25–36: Upscale models to real-world irrigation systems deployed within ongoing local and EU
projects (e.g., IRRITRE, AGRIF .OODTEF), and quantitatively assess their impact on water-saving strategies.

Possible Collaborations

Fabio Antonelli, Fondazione Bruno Kessler; Sara Bonetti and Concetta D'Amato, EPFL

Info: abouthydrology <at>  gmail.com

Sunday, March 30, 2025

A Ph.D. position ! Advanced Soil Biota-Hydraulics Interface for the WHETGEO-GEOSPACE system

Project Overview

This subproject, funded under the ICOSHELL project, aims to develop an integrated modeling
system that explicitly accounts for the dynamic interactions between soil biota activity and soil
hydraulic properties. Building upon the WHETGEO-1D and 2D frameworks, we will implement a
novel coupling between soil fauna population dynamics and plants root growth, evolving soil
hydraulic characteristics. The modelling system implemented will be eventually used for studying
the feedback between soil-vegetation hydrology.

Key Objectives


  • Extend the WHETGEO model architecture to incorporate time-varying soil hydraulic properties influenced by soil biota 
  • Implement the Kosugi soil water retention curve model with parameters that dynamically evolve based on biological activity
  • Develop and integrate a population dynamics module for key soil engineers (earthworms, ants, termites)
  • Create a comprehensive validation framework using laboratory and field experimental
  • data
Figure from Enrico Chiesa Master Thesis


Methodological Approach

The core innovation of this subproject is the implementation of a feedback loop between
 biological activity and soil physics. Following Meurer et al. (2020), we will start to model how earthworm populations modify soil structure, but significantly expand this approach by:

  • Replacing the van Genuchten model with the Kosugi water retention curve formulation, which provides a more direct physical interpretation of pore size distribution
  • Developing a differential equation system where the Kosugi parameters (median pore size and standard deviation) are directly modified by biological activity
  • Implementing these dynamics within the robust NCZ algorithm of WHETGEO, ensuring numerical stability across diverse conditions
  • The population dynamics will be modeled as a set of ordinary differential equations representing different functional groups of soil engineers, their reproduction, mortality, and activity rates as functions of environmental conditions (temperature, moisture, organic matter)

Expected Outcomes


This integration will allow to better capture:

• The temporal evolution of soil infiltration capacity following land-use changes

• The self-reinforcing positive feedback loops of ecosystem restoration, where initial

vegetation changes trigger soil biological activity that further enhances water retention

• The resilience of soil hydrological function under climate change scenarios


Implementation Timeline

Months 1-6: Preliminary studies, doctoral school activities

Months: 6-18 Implement Kosugi model in WHETGEO framework, develop and integrate

population dynamics module Months 18-24: Validate against experimental data Months 32-36:

Upscale to field applications and integration to estimate catchment scale effects. Study effects of

soil management

Possible collaborations

EPFL Lausannne, Prof. Sara Bonetti and Dr. Concetta D'Amato

Info: abouthydrology <at>  gmail.com

Wednesday, March 12, 2025

The Marvelous Physics of Plants: a personal Introduction

 "The Marvelous Physics of Plants" presents an exploration of the physics behind how plants function, particularly focusing on water transport mechanisms. The presentation begins with poetic descriptions of plant processes, then explores Erwin Schrödinger's fundamental question about how physics and chemistry can explain the events within living organisms. The authors examine various physics domains relevant to plants: quantum physics, thermodynamics, hydraulics, micrometeorology, stability, and light.

Among the other things,  the authors examine the physical limits of tree height, discussing how hydraulic restrictions ultimately limit how tall trees can grow. They also demonstrate synthetic tree models that scientists have created to replicate these natural mechanisms.

The slides combine mathematical formulations, anatomical diagrams, and experimental results to illustrate the physical principles governing plant function. A video of the talk is also available.


Friday, February 28, 2025

Three Batchelor Graduation Works

The first  Thesis-poster,  by Agnese Cavazzini, presents a hydrological study of the Secchia River basin using the GEOframe-NewAGE system. The research analyzes water balance and simulates river flow while generating soil moisture maps to identify drought-prone areas. Key elements include watershed division into sub-basins, mass balance equations, and calibration against measured data. Results show flow simulations at two monitoring stations and soil moisture anomaly maps. The successful implementation provides valuable insights into the basin's hydrological dynamics across Modena, Reggio Emilia, and Mantova provinces.


The second Thesis-poster, by Lorenzo Dalsasso, 





the third Thesis-Poster, by Marco Feltrin,







Thursday, February 6, 2025

Biosphere, Atmosphere, Climater Interactions 2025 Class

 This is a place holder



The Hydrological Modeling 2025 class

 Welcome to the 2025 Hydrological Modeling Class!

To better understand the materials provided:

  • Storyboards – Summaries of the lectures, usually in Italian.
  • Whiteboards – Explanations of specific topics, presented on a whiteboard using Notability on an iPad.
  • Slides – Commented in English (available since 2021).
  • Videos – Recorded during lectures to complement the slides, with no editing (as post-production would be too time-consuming).
    • 2025 videos are available on a [Vimeo Showcase] (link here).
  • Additional information & references – Marked in italics, for the curious and the brave who want to explore further.

📅 24 February 2025 – Part I

Syllabus & Introduction to Hydrological Modeling

In this session, I introduced the course and its learning-by-doing philosophy. We cover all theoretical concepts first, followed by the practical applications (with Professor Giuseppe Formetta).

The real start 

To begin is also worth to have a little (philosophical) analysis of what a model is. This is what done in the following parte of the lecture

📅 25 February 2025 – Geomorphometry

This session begins with a discussion of previous lesson topics and the rationale behind introducing geomorphometric concepts. Since catchments are spatially extended, understanding their geometry is essential for studying catchment hydrology.

In the first part, we focus on the geometrical and differential characteristics of topography, including:

  • Elevation
  • Slope
  • Curvature

These parameters are fundamental for extracting the river network and identifying different parts of a catchment.

We then define drainage directions and explore how they are computed using Digital Elevation Models (DEMs)—where topography is discretized on a regular grid. From these drainage directions, we determine the total contributing area at each point of a DEM.

These two key characteristics allow us to:

  1. Identify channel heads and extract the river network.
  2. Define hillslopes and establish an initial framework for Hydrologic Response Units (HRUs).

    📅 3 March 2025 

    Q&A - 

    Interpolations 
    This lecture, assuming that now you have at least the concepts of what a catchment is and theoretically you know how to extract it and subdivide it in parts, deals with the data to feed catchments hydrology models. Because catchments have a spatial distribution, then also the driving data must be distributed. We need therefore methods of interpolation. 

    Installations of the software can be found here, at this link.

    📅 10 March 2025 

     Interpolations part II
    In this class we try to understand how to estimate the errors over the estimates. Besides we introduce a method (the Normal Score) to avoid to obtain negative values when positive interpolated values are required.
    Q&A - 
    Spatial Interpolation (Vimeo2023)

    Hydrological Models. This is a class about hydrological models, so what are they ?

    The title is self-explanatory. A theoretical approach to modelling is necessary because we have to frame properly our action when we jump from the laws of physics to the laws of  hydrology. Making hydrology we do not have to forget physics but for getting usable models we have to do appropriate simplifications and distorsions. The type of model we will use in the course are those in the tradition are called lumped models. Here we also introduce a graphical tool to represent these models.

    📅 17 March 2025 

    Hydrological Models 

    For old material give a look to Hydrological Modelling 2023

    📅 25 March 2025 

     Linear Models for HRUs

    Once we have grasped the main general (and generic) ideas, we try to draw the simplest systems. They turn out to be analytically solvable, and we derive their solutions carefully. From the group of linear systems springs out the Nash model, whose derivation is performed.  Obviously, it remains the problem to understand how much the models can describe "reality". However, this an issue we leave for future investigations.
    A little more on the IUH and looking at the variety of HDSys models

    We introduced previously without very much digging into it the concept of Instantaneous Unit Hydrograph. Here we explain more deeply its properties, Then we observe that there are issues related to the partition of fluxes and we discuss some simple models for obtaining them. Not rocket science here. The concept that we need those tools is more important than the tools themselves. We also observe that linearity is not satisfactory and we give a reference to many non linear models. Finally we discuss an implementation of some of the discussed concepts in the System GEOframe. 

    📅 31 March 2025