Wednesday, January 31, 2024

A little on reviews of Land Surface Models

 Please improve: A nicely comprehensive review of Land-Surface-Models LSM is given in Blyth et al., 2020 being possibly completed by a reading to Fisher, 2020 for a different perspective on processes. The pioneering models were more what we nowadays call Process Based/Mechanistic models which, however contains a lot of parameters and parameterizations that have to ne calibrated, assimilated or characterized. This calibration is essentially a statistical step that in practice transform LSMs in a mix of PDEs, ODEs solver endowed with various techniques derived from statistics or, more recently, from machine learning (ML) (Pal et al., 2021). In some cases, statistical learning pretends to entirely substitute PB model which sometimes happens for some processes but is more rare in LSM as a whole.

With Prof. Prentice we share the quest for rei-inventing LSMs
It is apparent from the reading of the reviews above that the scope of LSM has been greatly expanded over the years above five main modelling domains: surface and canopy exchanges, soil and snow physics, water bodies, biogeochemistry and plant physiology and vegetation dynamics.
Some models have emerged as reference in literature. They include:


Blyth, Eleanor M., Vivek K. Arora, Douglas B. Clark, Simon J. Dadson, Martin G. De Kauwe, David M. Lawrence, Joe R. Melton, et al. 2021. “Advances in Land Surface Modelling.” Current Climate Change Reports 7 (2): 45–71.

Fisher, Rosie A., and Charles D. Koven. 2020. “Perspectives on the Future of Land Surface Models and the Challenges of Representing Complex Terrestrial Systems.” Journal of Advances in Modeling Earth Systems 12 (4).

Pal, Sujan, and Prateek Sharma. 2021. “A Review of Machine Learning Applications in Land Surface Modeling.” Earth 2 (1): 174–90.

Tuesday, January 23, 2024

Mapping and Modeling Flowing Network Dynamics in Temporary Streams (by Gianluca Botter and Nicola Durighetto

 Gianluca Botter and colleagues (among which a notable mention is needed to Nicola Durighetto) recent work in the ERC project Dynet is remarkable (as well as the older one) and a little of it is in this presentation they kindly prepared for the AboutHydrology blog.The first slides (you find them by clicking on the Figure below) are choreographic (slide 1 only a photo, slide 2 a photographic example of network dynamics).

Slide 2 shows that stream intermittency is a pervasive phenomenon in many riverscapes. All river networks, in fact, continuously expand and contract in response to time-varying climatic conditions. Consequently, the same reach can experience flowing water, ponding or no-water at all depending on the survey time, as shown by the examples reported in this slide.
Slides 3 movie (here) represents the simulated network dynamics in the Rio Valfredda, 5 km2 (BELLUNO, Dolomites). The movie has been built combining empirical observations and a hierarchical modeling framework. Network dynamics are very complex, with wet reaches that propagate upstream in response to rainfall events, but also active channels that extend in the downstream direction as the catchment wets up. Multiple disconnections are generated and removed as the network expands and contracts.
Slide 4 summarizes the conceptual model used to identify the timing and the duration of surface flow within a given location along the geomorphic network. According to the model, the presence of surface flow is produced by the imbalance between the local inflow Qin, which is made up by the sum of a surface and a subsurface component, and the maximum discharge capacity of the subsurface of that point, Q*. Consequently in this framework, the surface flow presence condition can be written as Q_in > Q*.
In slide 5 (movie) it is shown that surface flow presence is driven by the imbalance between the local inflow Q_in and the maximum subsurface discharge capacity in the hyporheic region, Q*. One important point of this formulation is that Q_in changes in time as a function of the catchment wetness, but also in space, with larger values of inflows that are associated to downstream sites with a larger contributing area. Instead, the maximum outflow Q* is typically constant in time, and does not necessarily exhibit significant scaling effects as it might depend on local features such as the slope or the subsurface transmissivity. As the catchment gets wet, the local Q_in increases along the network non-uniformly and activate larger and larger portions of the network, as shown in the upper movie of this slide. The same process can be also seen looking at the corresponding “specific” quantities, dividing both sides of the surface flow presence equation by the contributing area A. The advantage is that now the specific inflow (small q_in) can be assumed as nearly uniform along the network, while the max specific outflow, rho*, decreases downstream as the contributing area increases. From this new perspective, when the catchment wets up the local specific inflow increases almost uniformly everywhere in the network, and more and more sites experience surface flow starting from the most downstream nodes for which the max specific max outflow rho* is lower.
From a phenomenological view point (slide 6), this mechanistic formulation originates a hiearchical behaviour: during wetting, nodes are activated sequentially from from the most to the least persistent, thereby originating … a sequence of network configurations in which less and less persistent nodes activate as the network expands
Slides 7 shows that during drying, nodes dry out with an order that is the inverse of the one experienced during the wetting… Thus, more and more persistent nodes are progressively switched off as the network retracts. The hierarchical behaviour is observed also in case of dynamically fragmented networks, as in this example.
The hierachical structuring of channel network dynamics, shown in slide 8, has been the object of several past studies, in which the hierarchy was mathematically defined using graph theories, and the validity of the hierarchical scheme has been proved using empirical data from tens of catchments spread al lover the world. The hierarchical structuring proved to be a powerful tool to extrapolate in space incomplete empirical information on surface flow presence.
Slides 9 shows data about local persistency as derived from field surveys in different catchments belong to a broad range of geomorphoclimatic features, from humid settings (on the left) to a dry mediterranian catchment (shown on the right). As you can see there is a general tendency for the persistency to increase moving downstream along the network (indicated by blue-like colors), in line with the expected decrease of rho* for larger contributing areas. However, the observed patterns of local persistency are much more complex than expected in most cases, owing to spatial heterogeneity of hydromorphological features, such as slope and river bed permeability.
Activelength vs discharge plots in slide 10 are valuable to estimate the changes in the flowing length associated to changes in the catchment wetness. Different catchments show a vary different behaviour though…
Slide 11 (movie here) shows an example simulation derived using a stochastic model for simulating the spatio-temporal dynamics of temporary streams. The model takes advantage of few, widely available climatic and morphologic parameters to generate synthetic timeseries of rainfall, streamflow and active length. Furthermore, the approach allows the reconstruction of the time-varying configuration of the active network. Thanks to its simplicity and limited computational requirements, the model can be easily coupled with ecologic models to simulate specific in-stream processes taking place on temporary streams.
In slide 12 (movie here) the Authors combined synthetic dynamic networks with a metapopulation model. The model stochastically simulates the occupancy of a temporary stream by a target species, which is shown in dark blue in the lower plot. The available habitat for the focus species varies with time and is greatly reduced during droughts. Our results indicate that, when compared to a static network, temporary streams result in a lower average occupancy and a higher extinction probability.
To better understand the importance of network dynamics, slide 13 compares the simulated behavior of a fish species in two different conditions: a dynamic network, shown in the bottom panel, and a static network, shown at the top panel.
Even though the average length of the active network is the same in the two cases, the time variability of the available habitat inherent of the dynamic network results in a lower average occupancy and a higher time variability of the network length occupied by the metapopulation. Consequently, in the dynamic scenario there is also an increase in the probability of complete extinction of the target species within the network.
The results in slide 14 suggest that the presence of disconnections along the network lowers the mean occupancy and increases extinction probability. In fact, the target species always goes extinct in less than 1 year in the scenario characterized by the largest number of disconnections (upper panel), while it can survive in the other cases. Therefore, temporary disconnections produced by stream network dynamics are crucial to the ecological functioning of rivers.
This is an important result also in the frame of climate change, which is globally increasing stream intermittency.

  • N. Durighetto, F. Vingiani, et al. (2020). Intraseasonal Drainage Network Dynamics in a Headwater Catchment of the Italian Alps. Water Resources Research.
  • G. Botter, N. Durighetto (2020). The Stream Length Duration Curve: A Tool for Characterizing the Time Variability of the Flowing Stream Length. Water Resources Research.
  • G. Botter, F. Vingiani, et al. (2021). Hierarchical climate-driven dynamics of the active channel length in temporary streams. Scientific Reports.
  • N. Durighetto, G. Botter (2021). Time‐lapse visualization of spatial and temporal patterns of stream network dynamics. Hydrological Processes.
  • F. Zanetti, N. Durighetto, et al. (2022). Technical note: Analyzing river network dynamics and the active length–discharge relationship using water presence sensors. Hydrology and Earth System Sciences.
  • N. Durighetto, V. Mariotto, et al. (2022). Probabilistic Description of Streamflow and Active Length Regimes in Rivers. Water Resources Research.
  • N. Durighetto, G. Botter (2022). On the Relation Between Active Network Length and Catchment Discharge. Geophysical Research Letters.
  • N. Durighetto, L. Bertassello, G. Botter (2022). Eco-hydrological modelling of channel network dynamics—part 1: stochastic simulation of active stream expansion and retraction. Royal Society Open Science.
  • L. Bertassello, N. Durighetto, G. Botter (2022). Eco-hydrological modelling of channel network dynamics—part 2: application to metapopulation dynamics. Royal Society Open Science.
  • N. Durighetto, S. Noto, et al. (2023). Integrating spatially-and temporally-heterogeneous data on river network dynamics using graph theory. I-Science.

Between Hydrology and Geology: digital twins for preventing the hydrological and geological hazards

 Invited to talk to remember Fabio Rossi I chose to take a little detour describing the perceptual model of small catchments floods dynamics in the interplay between geology, geomorphology,  and hydrology. Not much technical information though, which you can find in the cited papers, but more the vision of what can be done with physically based models. 

The presentation ends with claiming that such approaches that could be seen as overwhelming can instead be pursued in the framework of DARTHs and within a cooperative, participatory action. Enjoy the presentation by vlivking on the figure above. 

Tuesday, January 2, 2024

Elementary Mathematics sheds light on the transpiration budget under water stress

This paper aims to establish a method to accurately describe transpiration by employing appropriate physical equations. Although some simplifications are made, including use of a simplified treatment of turbulence and neglecting of the thermal capacity of transpiring leaves, it is argued that the chosen scheme has general validity in identifying the primary mechanisms governing transpiration. 

To achieve this objective, a traditional treatment involving five equations, including the mass budget, is used. Initially, a simplified approach that does not consider the water budget is introduced to outline the general procedure to explicitly ad- dress canopies. Subsequently, the water budget is incorporated to appropriately account for water stress in transpiration. In this context, a novel linearization of the extended Clausius- Clapeyron equation, incorporating the Kelvin effect, is employed. It is demonstrated that the well-known Penman formula emerges as one of the solutions within a system of equations, providing estimates for temperature (T), vapor content in air (e), and the thermal transport of heat (H). The method, initially conceived for homogeneous canopies, is expanded to encompass sun-shade canopy layers. By employing the water mass balance, the trade-off between atmospheric evaporation demand and the water delivery capacity of the soil and stem is eluci- dated. Notably, it is revealed that the pressure potential within leaves is not solely determined by capillarity, but rather represents the dynamic outcome of the intricate interactions within the soil-plant-atmosphere continuum. These findings highlight differences from more simplistic approaches commonly employed, particularly concerning canopies. Overall, this study presents a methodological framework to accurately describe transpiration, incorporating key equations and addressing the complex dynamics involved in the soil-plant-atmosphere continuum, and suggests various directions of research in the field. The preprint manuscript can be found here.