The RVDM: modelling impacts, evolution and competition processes to determine riparian vegetation dynamics

Abstract

RVDM is an ecohydrological model aimed to study the vegetation dynamics in riparian areas that represents an upgrade respect to previous tools in the way of understanding the riparian dynamics. Important novelties are proposed by this tool, including a high temporal resolution (daily time step), a proposal of a new plant classification approach useful for research and management (successional plant functional types or SPFTs), good representation of the key processes that determine the vegetation dynamics in riparian areas (drought and flood impacts, recruitment, growth, succession and competition), an easy implementation, and feasible inclusion of river morphodynamics in the model implementation (including different daily elevation and soil maps in the inputs). The model implementation in a Mediterranean semi-arid study site resulted satisfactorily (cell by cell calibration accuracy ≥ 65%, cell by cell validation accuracy between 40% and 60%), demonstrating the great potential of this approach for future research and management applications. Although 36 parameters are included in the model conceptualization, the global sensitivity analysis demonstrated that only 8 types of parameters are actually influent.. These parameters are: minimum time since mixed for transition to terrestrial, root depths, transpiration factors, critical shear stress of early stages, minimum biomass required to allow succession, germination minimum capillary water content in the upper soil, effective depth considered for evaporation from bare soil and coverage of pioneers. RVDM model will be a useful tool for gaining a better understanding of the riparian plants behaviour under different ecohydrological conditions.

Exploring the key drivers of riparian woodland successional pathways across three European river reaches

“…Nevertheless, the similarities among the three sites (i.e., colonization stage occurred close to the reference flow whereas mature stage occurred principally in stable spots) proved the usefulness of succession phases on a very wide cross‐regional context where species‐based analyses could be hardly performed (Merritt et al, ), particularly when comparing Alpine and Mediterranean vegetation because species‐specific traits may exert control over its competitive and/or facilitative abilities (Bagstad et al, ; Bejarano, González del Tánago, et al, ). The use of succession phases, which rely on the sequential spatial array of vegetation and its temporal evolution as a response to different drivers, has proven to be transferable between geographical regions and therefore may be better reproducible within numerical process‐based models (e.g., García‐Arias & Francés, ).…”

“…Conversely, the Portuguese and, to a lesser extent, Spanish sites should be addressed with schemes taking into account groundwater levels and soil moisture. In this regard, frameworks accounting for groundwater availability (e.g., Bertoldi et al, ) or novel numerical process‐based models as the riparian vegetation dynamic model (García‐Arias & Francés, ) may be better suited to undertake them. However, model accuracy on narrow sites with complex vegetation arrays, multichannel sections, and the introduction of exotic material may be hardly improved despite the good performance achieved by former studies on the site (García‐Arias & Francés, ; García‐Arias et al, ).…”

“…In this regard, frameworks accounting for groundwater availability (e.g., Bertoldi et al, ) or novel numerical process‐based models as the riparian vegetation dynamic model (García‐Arias & Francés, ) may be better suited to undertake them. However, model accuracy on narrow sites with complex vegetation arrays, multichannel sections, and the introduction of exotic material may be hardly improved despite the good performance achieved by former studies on the site (García‐Arias & Francés, ; García‐Arias et al, ). Thus, more complex approaches accounting for sedimentological and braided channel morphodynamics (e.g., Gran et al, ; van Oorschot et al, ) may not be worth the enormous effort necessary to gather the necessary information.…”

“…Thus, the vegetation dynamics, sediment transport, or bank erosion has typically been modelled, keeping almost constant most of the remaining elements (Solari et al, 2016). Different numerical modelling approaches with fixed river morphologies have arisen in recent years, specifically addressed to model vegetation dynamics under different management or climatic scenarios (e.g., Benjankar, Egger, Jorde, Goodwin, & Glenn, 2011;García-Arias & Francés, 2016), whereas those considering simultaneous and interacting changes of vegetation and channel geometry, based on either cellular automata (e.g., Coulthard, Hicks, & Van De Wiel, 2007;Gran, Tal, & Wartman, 2015) or physics (e.g., Bertoldi et al, 2014), are currently rare (Solari et al, 2016). A few remarkable approaches include the modelling of sediment transport and deposition in vegetated areas or bank accretion (e.g., van Oorschot, Kleinhans, Geerling, & Middelkoop, 2016), which are still in their infancy as they are not yet completely understood (Solari et al, 2016).…”

Calibration of a parsimonious distributed ecohydrological daily model in a data-scarce basin by exclusively using the spatio-temporal variation of NDVI

“…Once the main spatial patterns were identified, we combined our own human perception with the confusion matrices between the main spatial patterns and the spatial maps of model parameterization. Confusion matrices are widely applied for map comparison in distributed modeling comparing actual to predicted values for each specific category defined previously (García-Arias et al, 2016;Bennett et al, 2013;Van Vliet et al, 2013 among many others). Generally, the rows in the matrix represent the values predicted by the model, whereas the columns represent the actual values.…”