Overview

Overview of the MEDFATE model

MEDFATE is a spatially explicit landscape model designed with the purpose to allow investigating the interaction between global-change factors in Mediterranean landscapes. The model allows examining the interaction between climatic changes, fire regimes, land-use changes and vegetation dynamics at short- to medium-term time scales (Fig. 1).

Purpose

The MEDFATE model shares some features with its predecessor, the MEDFIRE model (Brotons et al. 2013). Like MEDFIRE, it has been designed to simulate different fire regimes, allowing the investigation of their effects on landscape composition and configuration. Moreover, it can take into account climate and land-use dynamics modelled externally, to model their effects on fire regimes and vegetation. An important difference between MEDFATE and MEDFIRE is that MEDFATE implements the FATE succession model (Moore & Noble 1990), which deals with vegetation dynamics on a species-individualistic basis — i.e. simulating the life-stages of plant cohorts, their competition for light and their responses to fire perturbations and climatic stressors. In other words, MEDFATE allows the user to study plant community changes as an emergent property of species-level dynamics. Moreover, MEDFATE accounts for direct climate-vegetation relationships (e.g. drought effects), and not only climate change effects on vegetation mediated through fire perturbations, as in MEDFIRE.

State variables

State variables in MEDFATE are spatial variables that describe the landscape context and conditions. They are represented in raster format and cover the full extent of the study area. Spatial resolution is not fixed, but the model has been designed to work under cell sizes from 20x20m up to 100x100m. The temporal scale is fixed and one time step represents one year; simulations are normally run for several decennia.

Land cover type

Land cover type (LCT) is a categorical state variable describing broad land cover types:

  • Wildland – cells that for which complete vegetation structure will be tracked and will be affected by climate, drought and fire disturbances. Vegetation succession (response to fire and drought disturbances, succession, dispersal) is restricted to occur within these cells.

  • Burnable – cells that can burn (i.e. fires can initiate in these cells, or it can spread through them) but without vegetation structure. Typically, this includes agriculture areas, but in some applications it can also include wildland areas that are not of interest (e.g. alpine meadows).

  • Non-burnable – cells that do not burn and do not have vegetation. Typically, non-burnable cover types include urban areas, rock outcrops and water.

Stand structure and composition

MEDFATE describes the structure and composition of vegetation in each raster cell using cohorts of plants that pass through a series of four discrete life stages (Fig. 2): seeds (also called propagules), seedlings, immature plants and mature plants. Several raster layers are used to represent each of these life stages is represented in the model.

    • Seed cohorts: A seed cohort represents the seed bank of a given species, and is distinguished from other seed cohorts by the species identity and the seed type, which indicates whether the species has seed dormancy or not. Three attributes are stored for seed cohorts: species identity (PrSP, integer), seed type (PrType, active or dormant) abundance (PrABU, ‘none’, ‘low’ or ‘high’) and age (PrAGE, integer).

    • Seedlings: The seedling stage is assumed to last less than one time step (a year). Therefore, cohorts at this life stage are not stored.

    • Immature and mature plants: A plant cohort is defined using four attributes: the species identity (SP, integer), abundance (ABU, integer, 0-100), real age (AGE, integer) and effective age (AEF, integer). Real age determines the year of death for the cohort, whereas the effective age is used to determine whether the plant is mature or immature (i.e. whether it can reproduce sexually or asexually), and the stratum where the plant cohort is located. Effective and real ages are equal for most plant cohorts, but effective age can be different from real age if the plants are able to resprout after a perturbation.

Each of the attributes described above for seed or plant cohorts are stored in different raster layers. There is a maximum number of plant or seed cohorts that grid cells can contain.

Two additional raster vectors have values for each vegetation stratum (see Table S1.1) and complete the structural description of vegetation:

  • The total percentage of cover within each stratum (CoverSTR). This set of layers describes the vertical structure of living plants within the cell. There are as many vegetation layers as strata defined by the user.

  • The model stores the presence of dead fine fuels, such as leaves and small branches, in the stand (DeadFineSTR). Dead fine fuels arise after plant cohorts die (i.e., due to light or water limitations, or because they are too old) and remain in the corresponding vegetation stratum for a few years. These fuels may modify the vertical continuity of fuel in the cell, and hence fire behaviour (see FuelType below). There are as many raster layers of dead fine fuels as strata defined by the user.

State variables influencing fire behaviour and effects

Two state base variables are related to fire behaviour and its effects:

  • FuelType is an ordinal variable describing the level of vulnerability of the cell to crown fire events according to the vertical structure of fuel (Levels: NoCrownFires, LowVulnerability, ModerateVulnerability, HighVulnerabilty). Fuel type is calculated taking into account both live plant cohorts and dead fine fuels.

  • FireType is a categorical variable describing whether a cell burnt in the current year and, if fire occurred, whether it was a surface or a crown fire (Levels: Undisturbed, SurfaceFire, CrownFire). The fire type is determined taking into account both the structure of fuel (FuelType) and other factors conditioning fire behaviour (wind strength, topography…).

Static environmental state variables

A number of layers describe the environmental conditions prevailing in each cell and are static. They modulate the behaviour of fire perturbations:

  • MainWindDirection is used to determine the direction of fire spread in wind-driven fires.

  • Elevation and Aspect layers are used to determine spread rate in fires driven by relief.

  • ProbIgnition is a static layer used to determine the location of fire ignitions.

  • FireRegimeZone is used to determine the proportion of wind- or topography-driven fires.

Dynamic environmental state variables

MEDFATE is designed to account for temporal variation in climate by allowing externally calculated climatic variables to change during the simulated period. The following state variables can be updated every step (one year):

  • Temperature of the coldest month is the temperature of the coldest month and it is calculated by comparing the twelve temperature monthly averages of the current year. Its value is used to limit seedling establishment.

  • Cumulative soil water deficit indicates the level of water deficit experienced in the cell (actually, an average of the deficit experienced in the current year and the preceding one). Used to define the intensity of drought perturbations and to determine whether the fire season occurs in climatically adverse conditions.

MEDFATE is composed of six different sub-modules (Fig. 3). Each time step (a year), the model starts by executing of the Succession sub-model, which determines vegetation maturation and successional changes (including growth, mortality and recruitment of plant cohorts). In particular, this sub-model implements bioclimatic limitations by restricting the establishment of plants to areas where the temperature of the coldest month is not too low. Then, the Reproduction & dispersal sub-model determines the amount of seeds produced by each mature plant cohort and the seeds arriving to each cell, including both locally produced seeds and seeds dispersed from other cells. After that, the model executes the Drought effects sub-model, which is responsible for determining whether plant cohorts will die (or resprout after ‘loosing’ some biomass) as a result of cumulative soil water deficit and taking into account the level of competition experienced. After drought-driven mortality effects have been determined, the Fuel type sub-model may be called if the user wants fire spread and fire severity to be dependent on vegetation structure. After possibly determining the fuel type, the Fire regime sub-model is executed. This sub-model begins by setting the potential total area to be burned using experimental distributions calibrated for the study area. Then, it simulates as many wildfires as necessary until the potential annual area to be burnt is reached. For each fire, the sub-model stochastically chooses a potential size and an ignition location. The location chosen for ignition is used to determine the spread type (relief- or wind-driven). If fire fighting is not considered, the fire is allowed to spread until the potential fire size is attained. If fuel type available the fire sub-model determines the fire type in burned cells to surface or crown fire depending on both fire spread rate and fuel type. Otherwise, all fires burn as crown fires. Finally, the Fire effects sub-model determines the response of plant cohorts to fire (i.e., mortality, survival and resprouting) as well as the response of seed cohorts (i.e., mortality and dormancy break) according to the fire type.

Process overview and scheduling