Desertification and land degradation: origins, processes and solutions

Table of Contents

1. Authors and acknowledgements
2. Introduction
   1. Aims
   2. Definitions and key concepts
   3. Synthesis of previous and ongoing projects
   4. Key issues and outline
3. Evolution of desertification in the Mediterranean
   1. Climatic fluctuation
   2. Human influence
4. Primary drivers of desertification
   1. Biophysical and socio-economic causes
   2. Pathways of desertification
   3. Driving forces in the Mediterranean
   4. Synthesis of causes for DESIRE hotspots
   5. Conclusions
5. Processes and consequences of desertification in the Mediterranean
   1. Socio-economic and political factors
   2. Biophysical processes
   3. Conclusions
6. Indicators: monitoring desertification
   1. History of monitoring
   2. Indicators
   3. Monitoring and mapping techniques
   4. Concluding remarks
7. Modelling desertification
   1. Definition of modelling
   2. Model types
   3. Biophysical modelling
   4. Socio-economic and participatory modelling
   5. Gaps and progression in modelling
8. Solutions
   1. Biophysical solutions
   2. Political and socio-economic solutions
9. References

Biophysical modelling

Categorization of models can be done according to several criteria, possibly including process description, scale, complexity or scientific theme or subject addressed. Here, models are classified based on themes that are relevant to desertification:

• Climate (i.e. modelling climatic variability and climate change (GCMs))
• Land surface - atmosphere exchange: Soil Vegetation Atmosphere Transfer models (SVATs)
• Land surface models
• Vegetation models
• Erosion and hydrological models

Climate

Modelling climate variability and ultimately climate change is done using climate models. These use quantitative methods to simulate the interactions of the atmosphere, oceans, land surface and ice. Climate models can range between simple zero-dimensional models of the radiative equilibrium of the earth to complex coupled atmosphere-ocean global climate models. In between are energy-balance models, in which horizontal energy transport in the atmosphere is considered, and EMICs (Earth system Models of Intermediate Complexity) bridging the gap between conceptual models and GCMs. One of the most common uses of climate models is to explore the impact of perturbations caused by human activity (Pitman, 2003).

EMICs: Earth system Models of Intermediate Complexity

To bridge the gap between conceptual, inductive, simple on the one hand and comprehensive, quasi-deductive models on the other, Earth system Models of Intermediate Complexity (EMICs) have been proposed (Claussen et al., 2002, see Fig. 6.1). These describe the natural earth system excluding the interaction of nature and humans. EMICs include most processes described in comprehensive models, but in a more reduced (parameterized) form. They explicitly simulate the interactions among several components of the natural earth system, mostly including biogeochemical cycles (Claussen et al., 2002). On the other hand, they are simple enough to allow for long-term climate simulations over several thousands of years. A list of currently existing EMICs can be found through the website of the Potsdam Institute for climate impact research (Claussen, 2005). The latest update is May 2005 and updating is done every two years, when new EMICs are included in the table . For every model, the principal investigators are given, its scope, the model components, its limitations and performance, the applications and references.

Fig. 6.1: Graphical definition of EMICs (from: Claussen et al., 2002)

GCMs: Global Climate Models or General Circulation Models

Global Climate Models, or General Circulation Models aim to describe climate behaviour by integrating a variety of fluid-dynamical, chemical or even biological equations that are either derived directly from physical laws or constructed by more empirical means. Both atmospheric GCMs (AGCMs) and oceanic GCMs (OCGMs) exist, which can be coupled to form an atmosphere-ocean coupled general circulation model (CGCM), integrating the knowledge on atmospheric and oceanic circulation (Grassl, 2000). A recent trend is to extend GCMs to become earth system models that include submodels e.g. for atmospheric chemistry or carbon cycling.

Extensive information on climate change, including model evaluation, can be found in the IPCC TAR report (IPCC, 2001).

Two well-known CGCMs are HadCM3 (Hadley centre Coupled Model, version 3; described by Gordon et al. (2000) and Pope et al. (2000)) and CGCM3 of the Canadian Centre for Climate Modelling and Analysis (CCCma) and Flato et al., 2000). A list of 21 models, about all CGCMs existing at the time, that participated in the first phase of the CMIP project (Coupled Model Intercomparison Project) is given in Meehl et al. (2000).

In his review, Mulligan (2004), states that it is increasingly certain that greenhouse induced global climate change will have significant effects on regional climates of the Mediterranean. A general increase in temperature is fairly certain, but the impact on regional rainfall and evapo-transpiration in the Mediterranean is much less certain and local scale impacts are very unclear (Mulligan, 2004). Further advances and results of projects using GCMs for predicting regional climate change in the Mediterranean (notably the MEDALUS project) can be found in Mulligan (2004).

Land surface - atmosphere exchange

The nature of a land surface affects the land surface-atmosphere energy, water and momentum exchange. This characterizes the regional planetary boundary layer which controls the regional climate (Mulligan, 2004). To study these interactions between soil, vegetation and atmosphere, so-called Soil-Vegetation-Atmosphere Transfer (SVAT) models are developed (Dolman et al., 2001). Their purpose is to provide coupling between the near-surface atmosphere and the hydro-ecological processes that take place in the zone that extends typically from a few metres below the ground, through the vegetation into the lower atmospheric boundary layer (Shuttleworth, 2005). SVATS are the main mechanism by which complex land surface-atmosphere processes are integrated in GCMs (Mulligan, 2004). The upper boundary conditions are incoming solar and long-wave radiation, precipitation, atmospheric variables such as temperature, humidity and wind speed and if relevant, concentration of atmospheric constituents. In most SVAT models, the lower boundary conditions are weakly specified: often gravity drainage of soil water to a remote, unspecified groundwater table is assumed (Shuttleworth, 2005).

In the EFEDA II project a significant modelling effort was made, concentrated on the development of regional SVATS (Mulligan, 2004). As with GCMs, many SVATS exist. In Moran et al. (2004) and references therein, several SVATS are named. Comparison over wheat fields of several SVATS of varying complexity is done by Olioso et al. (2002).

Land surface models (LSMs)

The land surface is a key component in climate models, controlling the partitioning of energy between sensible and latent heat and of water between evaporation, infiltration and run-off (Pitman, 2003). Changes in land use are directly linked to many environmental problems at both global and regional scale, and are intrinsically related to the evolution of the regional and global climate (Salmun and Molod, 2006). Land surface schemes or models account for the parameterization of the surface and subsurface mass and energy transfers (Salmun and Molod, 2006). The character of the land surface is spatially variable (e.g. variability in vegetation cover, terrain type, soil texture and wetness etc), complicating calculations of land-atmosphere exchange. Mostly, the scale of heterogeneity is (much) smaller than the grid scale used in GCMs (about 200km). Techniques to account for this include 'dominant', çomposite', 'mosaic' and recently 'extended mosaic' (briefly explained in Salmun and Molod, 2006). Using these land surface models, many studies have been conducted to simulate the impact of land cover changes on regional or even global climate. A summary concerning (tropical) deforestation and desertification is given in Salmun and Molod (2006).

Within a climate model (e.g. a GCM), the element that simulates the initial effect of land cover changes is the land surface model. The evidence is very strong that regional-scale land surface perturbations cause continental-scale changes in climate (Pitman, 2003). In his comprehensive review, Pitman (2003) argues why the land surface should be important in climate models, including a description and examination of the historical development of LSMs.

Vegetation models

Vegetation cover provides a dynamic feedback between the atmosphere and the soil and land surface. Impacts of vegetation change may have strong effects on hydrology, geomorphology (e.g. protection against erosion) and climate and at the same time affect humans and livestock as it provides a means of food (Mulligan, 2004). Vegetation response to environmental change therefore, is an important issue and the modelling of vegetation changes is discussed here in two parts: vegetation models as part of a GCM and as smaller scale independent models.

Dynamic Global Vegetation Models (DGVMs)

Following the relationship between global patterns of vegetation cover and climate, several models of global vegetation patterns have been developed, e.g. BIOME (Prentice et al., 1992), BIOME-3 (Haxeltine and Prentice, 1996), MAPSS (Neilson, 1995) and DOLY (Woodward et al., 1995). Changes in climate affect the distribution of global vegetation communities, while vice versa changes in vegetation structure may significantly influence the climate (see examples in Foley et al., 2000 and references therein) at several timescales. While most climate models describe the rapid biophysical processes, longer-term ecological phenomena are not yet considered (Foley et al., 2000). In most land surface models, vegetation and soil properties are prescribed as boundary conditions which are not allowed to change with the climate, neglecting long-term changes in vegetation cover and resultant feedbacks (Foley et al., 2000). With the advance of Dynamic Global Vegetation Models (DGVMs), the coupling of vegetation models in which long-term changes in vegetation dynamics with GCMs has become possible, using various coupling techniques (see examples in Foley et al., 2000).

A well-known DGVM is the Lund-Potsdam-Jena Dynamic Global Vegetation Model (LPJ) which combines process-based, large-scale representations of terrestrial vegetation dynamics and land-atmosphere carbon and water exchanges in a modular framework (Sitch et al., 2003).

Smaller scale independent vegetation models

The aforementioned bio-geographic models are used to predict broad-scale patterns in vegetation for regions, continents and the globe. Another type of models used to predict vegetation dynamics, acting on a smaller scale (<1 to 100 m²), are the species-based successional models or gap models (e.g. Pausas, 1999; Sitch, 2003 and references therein; Peters, 2002). These simulate the recruitment, growth and mortality of individual plants and complex interactions such as landscape-scale processes and feedbacks between vegetation and soil processes can be represented by these models (Peters and Herrick, 2001). However, they are limited computationally in the spatial extent that can be simulated, due to the small plot size and detailed processes included. Attempts to extend their spatial scale include linking gap models with landscape-scale models (Peters and Herrick, 2001).

The Mediterranean

Two models focusing on the Mediterranean area are the vegetation components of the MEDALUS model and the ModMED model. They will be briefly discussed here.

The MEDALUS model is described in Kirkby et al., (1996) and the vegetation part is reviewed in Mulligan (2004). The vegetation component of the MEDALUS model plays an important role in the hydrological budget and in predicting erosion (Mulligan, 2004). Given the diversity of the Mediterranean vegetation, a model with a number of functional types with clear distinction between herbaceous primary grassy vegetation and woody types, thus a grass and a shrub model, are developed. The model has a large number of parameters, requiring an intensive field effort (Mulligan, 2004).

ModMED, acronym for Modelling Mediterranean Ecosystem Dynamics, aims at predicting the development of vegetation patterns in the landscape in response to changes in land use. The model simulates the processes of ecosystem dynamics integrating knowledge on the plant, community and landscape scale. While the primary objective of the model is to make predictions of vegetation change at the landscape scale, the fundamental principle behind it is that successful predictions result from modelling the system at a lower level; at the community and individual levels (Mulligan, 2004).

Erosion and hydrological models

There are many erosion and hydrological models, which makes it impossible to name them all here. A general overview of categorisation of erosion models is discussed and some well-known models are mentioned. Furthermore, models are reviewed in extensive reviews, such as Aksoy and Kavvas (2005) and Merritt et al., 2003, to which is referred for detailed comparison between erosion models.

Morgan and Quinton (2001) describe the history of erosion modelling, whereby the need to evaluate soil conservation practices is seen as the impetus for developing erosion models. They divide the models into empirical and process-based models. Aksoy and Kavvas (2005), in their review of hillslope and watershed scale models, discuss conceptual models apart from empirical and process-based ones, which is also done by Merritt et al. (2003) in their extensive review of erosion and sediment transport models. An overview of the 17 models reviewed by them is given in their paper, including type of model, scale, input requirements and reference.

Empirical models

Empirical models are based on determining statistically significant relationships between an intended model output and model inputs. The Universal Soil Loss Equation (USLE) is the most widely-used empirical model, with its greatest advantage being its simplicity. The disadvantage of all empirical models is that they are only valid for the database and conditions for which they were derived (Morgan and Quinton, 2001; Aksoy and Kavvas, 2005). Other examples of empirical models include the Soil Loss Estimator for Southern Africa (SLEMSA), the Morgan-Morgan-Finney (MMF) model, adapted by De Jong (1994) to the Soil Erosion Model for Mediterranean Areas (SEMMED). See also the list of models and reviews of individual models in Aksoy and Kavvas (2005).

Process-based models

Physics-based models use mathematical relations to describe the processes of erosion and simulate the movement of water and sediment over the land surface (Morgan and Quinton, 2001). As many of the equations still have an empirical base, these models are considered to be process-based rather than physics-based. They typically contain separate runoff and erosion components and employ some form of kinematic wave procedure for routing water and sediment (Morgan and Quinton, 2001).

A very large number of process-based models have been developed. Division between them can be made based on various criteria. A list of properties of 12 well-known physically-based erosion models is given in Aksoy and Kavvas (2005). Morgan and Quinton (2001) divide them in two broad groups: continuous simulation models and event models. The first require large amounts of data and are used to assess the long-term effects of land management of climatic change on run-off and erosion. Examples include CREAMS (Knisel, 1980); WEPP (Nearing et al., 1989), SEM/SHE (Storm et al., 1987) and PESERA (Kirkby et al, 2004). Event models, simulating the response of catchments to single storms, require less data but they do require assumptions about the starting conditions for each event. Examples include ANSWERS (Beasley et al., 1980); KINEROS2 (Woolhiser et al., 1990); GUEST (Misra and Rose, 1990); EROSION 2D/3D (Schmidt, 1991); LISEM (De Roo et al., 1996a,b) and EUROSEM (Morgan et al., 1998). Differences in the approach to simulate the erosion processes are described in their review (Morgan and Quinton, 2001).