Faraday Discuss

Introduction Conclusions References Tables Figures ◭ ◮ ◭ ◮ Back Close Full Screen / Esc Papers published in Hydrology and Earth System Sciences Discussions are under open-access review for the journal Hydrology and Earth System Sciences Abstract Introduction Conclusions References Tables Figures ◭ ◮ ◭ ◮ Back Close Full Screen / Esc Printer-friendly Version Interactive Discussion EGU Abstract Plant ecosystems in arid and semiarid zones show high complexity from the point of view of water resources, since they depend on water availability to carry out their vital processes. In these climates, water stress is the main factor controlling vegetation development. 5 The available water in the system results from a water balance where the soil, vegetation and the atmosphere are the key issues; but it is the vegetation which modulates (to a great extent) the total balance of water and the mechanisms of the feedback between soil and atmosphere, being the knowledge about soil moisture quite relevant for assessing available water and, as a consequence, for growth and plants maintenance 10 and the final water balance in the system. A conceptual dynamic vegetation-soil model (CDVSM) for arid and semiarid zones was developed. This model based in a tank type conceptualization represents in a suitable way, for Mediterranean climate, the vegetation responses to soil moisture fluctuations. Two tanks interconnected were considered using the water balance equation 15 and the appropriate dynamic equation for all considered fluxes. The first one corresponds to the interception process done by the vegetation. The second one models the upper soil moisture determination. In this tank parameters are based on soil and vegetation properties. The transpiration of the vegetation is a function of the soil moisture , the vegetation type and the biomass. Once all water state variables are evaluated 20 at each time step, the modifications in the biomass are made as a function of transpi-ration rate and water stress. Simulations for monoculture of Quercus Coccifera L. were carried out. Results shows that CDVSM is able to represent the vegetation dynamic, reflecting how the monocul-ture is stabilized around 0.7 of relative biomass, with adaptation to the soil moisture 25 fluctuations in the long term. The model shows the vegetation adaptation to the variability of the climatic conditions, demonstrating how either in the presence or shortage of water, the vegetation regulates its biomass as well as its rate of transpiration trying 3470 Abstract Introduction Conclusions References Tables …

Nowadays, there are many ways to answer the vegetation modeling question. Most of the hydrological models are able to represent hydrological processes, at watershed scale, but all of them consider the vegetation like a static parameter. Models of the physiological processes of vegetation like light interception (Kiniry et al., 2005;Dewar et al., 1998;Kiniry et al., 1999); water interception (Eltahir and Bras, 1993 2003;Nouvellon et al., 2000a) were developed at plant or vegetation plot scale, but with high parameter requirements. Terrestrial ecosystems models (TEM) simulate mainly photosynthesis processes, autotrophic and heterotrophic respiration, allocation, turnover, mortality, fire, land-use change, competition, etc. The TEM involves the plant physiological processes at re-5 gional to global scales to be coupled to Global Circulation Models to measure climate changes (Arora and Boer, 2005;White et al., 2000).
Models designed to simulate agriculture management (SWAP Kroes and van Dam 2003; SWAT Neitsch et al., 2002) are limited to simulate crops growing, irrigation practices, pesticides use, nutrients requirements, nevertheless it needs a large amount of 10 data.
More recently proposals such as Soil-Vegetation-Atmosphere Transfer scheme (SVAT) (Arora, 2002;Dawes et al., 1997;Federer, 1979), or Land-Surface Models (LSM) coupled to TEM are designed to simulate energy and carbon fluxes, requiring that the processes of photosynthesis, respiration from vegetation and soil carbon com- 15 ponents and allocation of net carbon uptake to several vegetation components to be explicit, requiring too many data (Montaldo et al., 2005;Arora, 2003;Nouvellon et al., 2000b;Cao and Woodward, 1998;Noilhan and Mahfouf, 1996;Famiglietti and Wood, 1994;Wigmosta et al., 1994;Dawes and Hatton, 1993;Mackay and Band, 1997;Spittlehouse and Black, 1981). SVAT or LSM-TEM are done to solve the static character 20 of the vegetation in hydrological processes simulation, to study vegetation response to disturbances like fire, or simply to manage land and water resources.
Soil moisture dynamic and bare-soil evaporation are highly important processes in semiarid and arid ecosystems dynamic. These processes are modelled together due to closest interaction between both, neglecting or simplifying the other hydrological Introduction  (Parlange and Katul, 1992). The main objectives of this paper include: -Developing a conceptual vegetation-model for arid and semiarid ecosystems that represent the vegetation response to the soil moisture fluctuations.
-Formulating biomass model based on its water demand and water-soil availability.

5
-Using water stress concept indicating survival plant conditions as function of the plant water demand versus soil offer.

Model description
A conceptual dynamic vegetation-soil model for arid and semi-arid zones was developed to model soil-vegetation-atmosphere relations. The vegetation-soil model was 10 based in a tank type conceptualization, considering two tanks interconnected, using the water balance equation at each tank and the appropriate dynamic equation for all considered fluxes. The vegetation model is mainly a monoculture, functional vegetation type or dominant species. This means that we only considered parameters for the species that 15 represent the group of vegetation to simulate the vegetation-soil processes.
In Fig. 1, the rainfall X 0 enters to the system; a quantity of water D 1 is derived to first tank; this quantity of water intercepted (H 1 ) is function of biomass, precipitation and previous water intercepted, and is available to direct evaporation (Y 1 ). The quantity of rainfall that can not enter to first tank is considered the throughfall (X 1 ), and is an 20 amount of water able to enter the second tank. The quantity of derived water to second tank depends on soil and vegetation properties, and is available to transpiration (T ) and bare-soil evaporation (BSE). Finally, the amount of water that not participate in initial abstractions and capillary water storage in upper soil represents the water excedeence (X 2 ) and it is considered available to infiltration and direct runoff. Introduction

EGU
The CDVSM only models the initial abstractions and upper soil storage, because there occurs the main processes related with the vegetation; but for its structure CD-VSM is easily coupled with hydrological models, given that, this evaluates the first processes in the hydrologic cycle and returns the water available to infiltration and direct runoff.

Interception and direct evaporation
The first tank represents the water retained by leaves and only is able to outflow by direct evaporation. The maximum capacity of the first tank is I mx [mm], depending on shape, quantity and intensity of rainfall, leaf biomass and vegetation type. According to the model scheme, the rainfall X 0 [mm] is stored in the first tank until the maximum 10 capacity is reached. Then the throughfall X 1 is defined by, where R [-] is the relative biomass. The water intercepted can outflow mainly by evaporation or by leaf and stem absorption in low percentages so, this last way is rejected. The water evaporated from interception Y 1 [mm day −1 ] is evaluated by, where H 1 [mm] is the water intercepted and PET [mm day −1 ] is potential evapotranspiration rate.

Upper soil storage and evapotranspiration
The second tank represents the water retained by capillary-soil forces in upper part of 20 the soil or rooting zone. This storage has a maximum capacity (H u ) function of field capacity and effective root depth (z e ). The throughfall X 1 is stored in the second tank up to H u then, the water exceedence X 2 is determined by,

EGU
where H 2 [mm] is the available water in the second tank. The water inside the second tank can outflow by bare-soil evaporation (BSE) or by transpiration (T ). The actual evapotranspiration Y 2 represents the sum of losses of water by T [mm day −1 ] and by BSE [mm day −1 ]. The model gives priority to T instead to BSE, that is, in the sequence of water extraction from the second tank, T has the 5 first chance to do it following the expression, where θ is the upper soil moisture content and is related with available water H 2 computed by The variable T is limited on one hand, by atmospheric conditions represented by PET and the residual potential evapotranspiration (PET -Y 1 ); on the other hand, is restricted by soil moisture conditions represented by extraction curve f(θ) and H 2 . When θ is between optimum soil moisture (θ * ) and field capacity (θ f c ) contents, T depends on type of plant (biomass and soil moisture threshold of normal physiological processes) 15 and climatic conditions (temperature, relative humidity, etc.). As long as θ decreases, T is reduced by stomatal closure to prevent water losses, and H 2 determines T , which continues until the θ reaches the wilting point (θ w ), where suction to extract water from soil produce damage in the plant tissues. This relationship was studied widely at the level of both individual plant and plantation scale (Federer, 1979;Spittlehouse and 20 Black, 1981;Daly et al., 2004) and has been demonstrated that can be approached to a linear piecewise function, when θ * determines if the plant is unstressed or stressed.
The BSE process is limited to the area not covered by vegetation (1-R) and is considered to take place into the soil surface layer with the same soil texture than rooting zone Introduction EGU and soil surface depth z ss [mm], much smaller than z e . The BSE process is limited by the actual residual potential evapotranspiration (PET -Y 1 -T ), bare-soil surface and θ, assuming the same distribution along the z e , 2.3 Dynamic vegetation modelling 5 As mentioned before, in spite of the fact that vegetation growth requirements are depending on light, water and nutrients, in semiarid and arid environments the vegetation is highly conditioned to the availability of the water to carry out vital processes. To represent this dependency, the model considers the relation among vegetation growing, T and vegetation water stress (VWS) to estimate the relative biomass production (Daly et al., 2004;Kramer and Boyer, 1995;Kramer, 1969;Rosenzweig, 1968;Dachnowski, 1914)

EGU
[-] is the wilting soil moisture, related to the water potential value ψ w [MPa], below it, the vegetation can not extract more water and can suffer damage on its tissues; and exponent q [-] is a measure of the nonlinearity of the effects of soil-moisture deficit on plant conditions (see Fig. 2).

Model application 5
Our Conceptual Dynamic Vegetation-Soil Model (CDVSM) was applied to a Mediterranean semiarid slope covered by kermes oak species. The kermes oak, Quercus coccifera L., is an evergreen sclerophyllous shrub which covers extensive areas of Mediterranean garrigue in Mediterranean watershed (Le-Houérou, 1981). Quercus coccifera L., inhabits in regions when the edaphic conditions lead to a great aridity, and is considered pyrostable since shows a high regeneration after fire, due to the continuity of belowground biomass after fire plays an important role in determining the optimum tactics to be adopted during succeeding cycles (Delitti et al., 2005;Cañellas and San-Miguel, 2000;Abril and Gracia, 1989). The kermes oak is found commonly in Spain in continental vegetation structure, in meso-Mediterranean belt and varied om- 15 brotype climates (dry-humid, semiarid-dry, dry-subhumid and dry) reaching up to 2 m of height; or in coastal formation, in thermo-Mediterranean belt and several ombrotype climates (semiarid-humid, dry, dry-subhumid and semiarid) the kermes covers 75% of terrain impeding development of herb substrate, reaching no more to 1 m of height. Sanchis et al. (2003), point out the kermes oak species is able to live in soils with any 20 kind of chemical characteristics, but it is very frequently found in soils with low depth, over Chromic Luvisols ("terras rossas"). Model time discretization is daily, with the aim of modelling as well as the dynamic vegetation, vegetation water stress response and fluctuations of the soil moisture content along the year. Spatial scale is at 10 m order, because done possible to model the 25 physiological process that occurs at level both of individual and set of plant, taking into account the hydrological processes scale. year, total precipitation is near to 137.31 mm in winter, decreasing to 116.10 mm in spring until 56.49 mm in summer and increasing 204.65 mm in autumn (see Fig. 3).
For the recorded period, the mean temperature was 17 • C. The whole recorded period is characterized by a summer season with temperature around 30 • C and 18 mm precipitation; the winter season is characterized by reach -1 • C of minimum tempera-20 ture and 45 mm of mean precipitation. Spring and autumn seasons have 16 • C of mean temperature but with mean precipitation of 38 mm and 70 mm respectively, being both the optimal seasons for vegetation growing. The mean annual potential evapotranspiration (PET ) computed with Hargreaves equation (Allen et al., 1998) is approximately 1250 mm, which is much larger than the precipitation. Along the mean year PET is 25 around 55 mm day −1 in winter, increasing to 122 mm day −1 in spring and then up to 150 mm day −1 in summer and decreasing 90 mm day −1 in autumn, as it is shown in Fig. 3

Parameters estimation
The parameters of CDVSM can be separated into soil and vegetation parameters. Soil parameters, as was explained before, are related with the estimation of available water by transpiration and evaporation processes at different soil moisture contents (θ). According to the soil-water retention curve of Campbell (1974) soil moisture content can 5 be computing by where aeration pressure ψ ae [MPa], porosity distribution index b dimensionless and porosity φ [-] are required to determine the typical curves for main kind of soils; ψ ae and b were experimentally determined by Clapp and Hornberger (1978). Table 1 includes parameters required for Eq. (10) for loam sand soil texture, which will be used in basic scenario by analysis purposes. To compute soil moisture content at field capacity (θ f c ) and wilting point (θ w ), Larcher (2003)   Vegetation parameters are related to interception, vegetation water stress response and relative biomass calculation. Parameters as A n,mx , B pot , T mx , k and I mx depend on the species selected to be modelled; whereas, θ * depends on both, soil and vegetation properties. Respect to leaf shedding parameter k, Castro-Diez and Montserrat-Martí (1998) point out that Quercus coccifera L. leaves falls in spring and continues through 20 summer season and occasionally in autumn, so, the leaf shedding used correspond to 0.0018 in winter and autumn 0.002 in spring and 0.0019 in summer. For the maximum interception (I mx ) parameter estimation, Federer (2002) proposes a simplified version of Gash model

EGU
where C LAI and C SAI are the interception capacity by unit of leaf area index and stem area index respectively; and, LAI and SAI are maximum leaf area index and maximum stem area index respectively. Fundamental physical characteristics and resulting model parameters are defined in Table 1.

5
Although the time resolution model is daily, the CDVSM calibration was made minimizing the root mean square error (RMSE) of annual relative biomass simulated with CDVSM compared with biomass field observations obtained from Cañellas and San-Miguel (2000). The parameters include in Table 1 are considered the basic scenario, and were used 10 in calibration processes of c parameter, which is considered the key factor in Eq. (8), since it determines the relationship between vegetation growing and transpiration rate, according to relative biomass conceptualization. Daily biomass simulated with CDVSM and Cañellas and San Miguel (2000) field observations are showed in Fig. 4. After the first year of simulation the leaf biomass 15 is 2.6 t ha −1 , continues arising until reaches a mean of 4.6 t ha −1 over 4-10 years of simulation, and then it is stabilized around 4.1 t ha −1 along the simulation period. In Fig. 4 the dots, represent Cañellas and San Miguel (2000) field observations, these correspond to 20 communities in Valencia, with similar soil and climatic conditions that our basic scenario, but there are plots previously devastated by fire; the year represent 20 the age of community after fire. Younger communities show maximal development until 6-8 year, reaching 4.9 t ha −1 on average; the oldest communities show a biomass stabilization around 4.0 t ha −1 . Our CDVSM has been developed considering mature communities and neglecting successional processes. So, in spite of these technical differences, the CDVSM model and Cañellas and San Miguel (2000) Fig. 5 shows the general behaviour (means) of the main state variables of CD-VSM (transpiration T , bare-soil evaporation BSE, available water H 2 , vegetation water stress VWS and relative biomass R) due to changes in climate and soil conditions. 4.1 Soil types 15 The Fig. 5a shows the variation respect to H u as a function of soil texture. Thus, H u for loam sand is 51.99 mm (100%), silty loam soil texture is 126.95 mm (245 % of H u for loam sand); sandy loam soil texture is 76.93 mm (148 %); and clay loam soil texture is 92.98 mm (178%). The most sensible state variables to changes in soil type were H 2 and BSE. Changes in H 2 from 17 (10% of H u ) to 180 % (190% of H u ) showed positive 20 correlation with H u , which was in agreement to the expected, given that any increment of H 2 is favoured by chemical and physical characteristics of the silt and clay and, as soon as z e increases, H 2 also increases. Positive correlation between BSE and H u was found, changing from 11% (10% of H u ) to 185% (250% of H u ), clearly reflecting soil characteristics. I.e., silty loam soil evaporates more water quantity (173.72 mm; 185% of BSE) than sand loam (125.01 mm; 133% of BSE) and clay loam (146.18 mm; 156% of BSE) since fine grained soils hold more water than coarse soils and, consequently, Introduction EGU evaporation losses are greater than in coarse soils (Wythers et al., 1999). In spite of that H 2 increases with changes in soil texture, T does not it at the same rate, exhibiting fast increment from 48% (10% of H u ) to 100% (100% of H u ); but, after this only increments until 103% (250% of H u ); this behaviour may be because the main T restriction is the evaporative demand of the atmosphere and the total leaf biomass.

5
Also as it was expected, VWS decreases from 120% (10% of H u ) to 98% (250% of H u ). Contrary behaviour shows R, while H u increases from 10% to 250 %, R increases slowly from 75% to 102%.
In general, the model exhibits a consistent behaviour to the expected one. Changes in soil texture reports great changes in BSE and H 2 , but not thus in R, VWS and T .

Effective root depth
In Fig. 5b effective root depth (z e ) was the variable parameter, and the texture soil was fixed to loam sand (the basic scenario). The most sensible state variable was H 2 which is positive correlated with z e , changing from 3% (10% of z e ) to 230% (190% of z e ). Slow changes in BSE were exhibited, changing from 71.5% (10% of z e ) to 101% 15 (190% of z e ); it would be explained given that BSE depends on one hand by available water, so, to at lower values of H 2 low values for BSE has been recorded; on the other hand, at highest values of H 2 , the evaporative demand of the atmosphere controls BSE, reaching no more than 101%, being neglected the relation z ss /z e. These results are in agreement with Ritchie (1972) bare-soil conceptualization, which suggests that BSE 20 processes occur in two stages: stage 1, is determined by the evaporative demand of the atmosphere; and stage 2, evaporation rates are limited by the lack of water in the upper soil layer and soil hydraulic (Snyder et al., 2000).
Variables like T and R had similar behaviour, for lower values of z e , they decrease below to 30% and 60% respectively, but at higher values of z e , they not increase more 25 than 120%. These values reflect that vegetation-soil system is more sensible to shortage of water (shallow soils) than abundance of it.

Precipitation
In Fig. 5c changes of the model to variations on precipitation were analyzed. Variables T and H 2 showed positive correlation with precipitation, reaching 155% and 140% respectively. Changes in BSE between 22% (10% of P ) to 95% (70% of P ) were found, but after 70% of P until 190%, BSE reached a steady state around 99%. This be-5 haviour may be because z ss and z e continue fixed to 50 and 500 mm respectively and, according to model conceptualization, the soil can not evaporate more water that can storage. Although R showed increasing values, it does it slowly, starting from 72% (10% of P ) and only increasing to 110% (190% of P ). I.e., the sensitivity of the vegetation biomass is higher for reductions in precipitation than for increases.

Potential evapotranspiration
Finally, Fig. 5d shows model sensibility to changes in PET. The variables VWS, T and BSE had the same behaviour. At 10% of PET, T and BSE presented percentages near to 30%, and VWS near to 10%, increasing slowly to 115% (190% of PET ). These values would mean that at higher evaporative demand of the atmosphere, T and VWS 15 will be regulated by vegetation.
Negative correlation between H 2 and PET was found, at lower values of PET (10%) higher values of H 2 has been recorded (382% of H 2 ); but, as soon as PET increases more than 100%, slow changes in H 2 are recorded (64% of H 2 for 190% of PET ). About R, changes between 90% (10% PET ) to 115% (190% PET ) were recorded. Changes 20 in PET affect gradually the relative biomass changes, exhibiting adaptation strategies to this kind of climate change.

Model dynamics
The Fig. 6  EGU BSE and available water H 2 ) through the year. By graphical convenience T and BSE has been scaled by mean daily evapotranspiration (DET ) equal to 3.45 mm day −1 and H 2 has been scaled by H u equal to 51.99 mm (maximum available water for loam sand soil texture). The set of parameters used were the same for previous analysis, but the initial condition for relative biomass R was equal to 0.73.

5
In winter (Fig. 6a) the model recorded the maximum VWS variability, with a standard deviation of 0.35. Most of the time, VWS was below than 0.85, with a mean value of 0.54. Mean value of T was less than 17% of DET, in spite of mean value of H 2 was around 40% of H u ; this result may be because the vegetation is dormant due the lower temperatures in winter season and the transpiration processes is minimized. Lower variations on R were recorded, remaining around 0.72 with a standard deviation of 0.04. In respect of BSE, a low mean value was recorded (9% of DET ).
In spring (Fig. 6b), temperature and precipitation increase and vegetation is activated. Mean values of T and R, 24% of DET and 0.73 respectively, were recorded; these increments in T and R with respect to winter values produce that VWS increases 15 and H 2 decreases, registering 0.74 and 23% of H u respectively. In respect of BSE, a low mean value of 7% of DET was recorded, due to low value of H 2 after water extraction for T . Spring dynamic could be explained, because despite P increment, the vegetation reactivates its growth (R) and consume more water (T ), so, H 2 in soil decreases, and the system records a moderated VWS. 20 In summer (Fig. 6c), due to shortage of precipitation and higher temperatures, H 2 remained at lower values (9% of H u ) and VWS was maximum reaching most of the time a mean value of 0.9. The reductions in H 2 implied a reduction in R (mean value of 0.69), lower mean values of T and BSE, 11% and 4% of DET respectively, in spite of higher values of PET in this season. 25 In autumn (Fig. 6d), precipitation increases and temperature decreases slowly, so the system was recovered. Despite of the fact that H 2 increases to 34% of H u and mean VWS decreases from 0.9 to 0.6, the vegetation reactivate its growth with mean value of R equal to 0.71 and consume more water in transpiration process reaching

EGU
T a mean value of 20% of DET. This dynamic does that system records a moderated VWS in this season. In respect to BSE, since that H 2 increases, BSE increases too, reaching 10% of DET.

Conclusions
This paper presents a conceptual vegetation-soil model based in a tank type schema-5 tization. Assuming vegetation biomass must be a state variable instead of a fixed parameter, the objective was to develop a simple model to represent the soil-vegetationatmosphere dynamic in arid and semiarid zones, which means water-limited ecosystems, in order to reproduce the possible interactions in both ways between vegetation and soil moisture. To do this, the biomass dynamic (relative biomass R) was linked 10 to water demand (transpiration T ) and water-soil availability (H 2 ), using the vegetation water stress (VWS) as indicator of plant survival conditions to the environmental restrictions and T as an indicator of biomass growing. The proposed dynamic vegetation-soil conceptualization results in a parsimonious model (CDVSM), with low computational cost and which can be easily linked with others 15 complete hydrological or land surface models. It can be proved CDVSM reproduces well the biomass dynamic based on soil water balance, considering VWS as an index of its dynamic in semiarid and arid zones, due to -a great extent -VWS determines the growing season, water-uptake dynamic and can help to understand the adaptations strategies of the vegetation to shortage of water. Respect to bare-soil evaporation 20 (BSE), the model reproduces this dynamic in agreement with Wythers et al. (1999) but using a lower level of parameterization.
For a Mediterranean semiarid slope with Quercus coccifera L. T is more sensible to changes in precipitation (P ), effective soil depth (z e ) and maximum available water (H u  EGU and R do not show high sensitive to these parameters and inputs. I.e., the Quercus coccifera L. changes evapotranspiration fluxes and soil moisture content in order to maintain more stable values of VWS and R. In the long term, R shows variations along the year, recording its maximum values in spring (mean value 0.73) and minimum in autumn (mean value 0.68); these results 5 are agree with the reality, because in spring the vegetation experiments the optimal conditions of temperature and H 2 (23% of H u ) for growing. In summer, the higher temperatures and shortage of water (H 2 =8% of H u ) implies regulation in all vital processes like T (11% of DET ) and losses of R (from 0.73 to 0.69) to keep itself in optimal conditions. And in autumn, the vegetation exhibits accumulative response of all processes 10 of leaf shedding (k), extreme temperature in summer and shortage of water; in this season R decreases slowly from 0.69 to 0.68, but recording reductions in VWS (from 0.89 in summer to 0.62) and increments in H 2 (from 9% of H u in summer to 34%).  Figure 5. Analysis of sensitivity to maximum available water (H u ), effective root depth (z ), annual precipitation (P) and annual potential evapotranspiration (PET). The Y axis Fig. 5. Analysis of sensitivity to maximum available water (H u ), effective root depth (z e ), annual precipitation (P ) and annual potential evapotranspiration (PET ). The y-axis is the percentage of transpiration T , bare-soil evaporation BSE, vegetation water stress VWS and relative biomass R; the x-axis is the percentage of parameter reference value showed in parenthesis for H u , z e , P and PET. Introduction  Fig. 6. Box and whisker chart of relative biomass (R), vegetation water stress (VWS), transpiration (T ), bare-soil evapotranspiration (BSE) and available water (H 2 ) along the mean year. The variables T and BSE are scaled with daily mean evapotranspiration (DET =3.43 mm day −1 ) and H 2 is scaled with maximum available water (H u =51.99 mm).