4. Modeling Process
5.3 Model Data
5.3.3 Land Cover
The model used in this study is set up using two different land-cover related modules. Non-irrigated vegetation is modeled using the more physically based routines originally incorporated in SWAT, whereas irrigated agricultural areas are modeled using the more conceptually based CropWat approach (see section 5.1.2.2). Therefore the required input data is presented in the following distinct sections.
5.3.3.1 Non-irrigated Areas
The following parameters have been determined in field studies conducted within the IMPETUS framework or by literature review, as described in the following paragraphs:
Minimum and Optimal Growing Temperature: T_BASE, T_OPT[°C]
These temperatures determine the plants growth and its seasonal variation of other plant related parameters.
Minimum and Maximum Leaf Area Index: ALAI, BLAI [m²/m²]
These parameters govern transpiration and interception and its seasonal variation.
Canopy Height: CHTMX [m]
The Canopy height affects wind speed at the ground level, hence evapotranspiration.
Maximum stomata conductance GSI [m/s]
Stomatal conductance is a driver of transpiration.
Rooting depth: SOL_ZMAX, RDMAX [m]
The rooting depth affects the plants ability to extract water from the soil.
Curve Number: CN2 [-]
The Curve Number determines the surface runoff as a fraction of precipitation that reaches the ground. It is set based on land cover and the hydrological soil group determined in the last section 5.3.2.
Arid and semi-arid regions are under-represented in the SWAT crop database. Arid rangelands, grass and brush steppes are all represented by little bluestem (Schizachyrium scoparium) a North American prairie grass; they differ by varying Leaf Area Index (LAI) only. SWAT applications in arid regions have rarely been conducted; therefore only little information on required plant parameters exists and specifics of contrasting biomes are not captured by the actual database. Nevertheless, the remaining parameters, which are predominantly necessary for modeling the plants nutrient cycle and provide results that are not considered within this study, have been taken from the original SWAT model database (Schizachyrium scoparium, for details see Neitsch et al. 1999).
Within the IMPETUS project a vegetation map has been developed (Finckh 2008). The classification bases on elevation, climate data and soil data (sources, as described in section 5.3.2), Landsat ETM+ satellite imagery (NASA 2003) and a habitat model (Staudinger &
Finckh 2005). 26 different land cover types have been classified. Since this classification is too detailed for a direct use in SWAT-MAROC, the land use map has been simplified according to Table 5-6. The following features have been dropped and resulting gaps have been filled using the Euclidian Allocation routine from ArcGIS:
Agricultural areas, because they are modeled separately as described in the following section 5.3.3.2
Linear features linked to streams such as stream beds, meadows and swamps, due to their minor areal extent
The reservoirs, as small reservoirs are not considered in this study and the reservoir Mansour-Eddahbi is modeled separately as described in section 5.3.8
Table 5-7: Aggregation scheme for the LU/LC map used in SWAT-MAROC. Original data provided by Finckh (2008)
Biome Area [km²] Simplified class
1 Date palm oases 33 -
2 Mediterranean Oases 199 -
3 Sub-Mediterranean Oases 59 -
4 Rain-fed agriculture 101 Hamada steppe, degraded
5 Swamps 7 -
6 Debris 57 Mountainous vegetation, degraded
7 Dense thorny cushion shrubs 797 Mountainous vegetation
8 Clear thorny cushion shrubs 1289 Mountainous vegetation, degraded 9 Dense thorny cushion shrubs, lowland 225 Mountainous vegetation
10 Clear thorny cushion shrubs, lowland 300 Mountainous vegetation, degraded
11 Juniper trees 67 Mountainous vegetation
12 Woody artemisia steppe 203 Brush steppe
13 Dense artemisia steppe 1998 Brush steppe
14 Degraded artemisia steppe 2313 Brush steppe, degraded
15 Dense rocky Hamada 6 Hamada steppe
16 Dense Hamada 1619 Hamada steppe
17 Degraded Hamada 4376 Hamada steppe, degraded
18 Arid Hamada 105 Hamada steppe, degraded
19 Saharan rock communities 900 Hamada steppe, degraded
20 Meadows (Oleander) 65 -
21 Meadows (Atriplex Glauca) 3 -
22 Active wadis 15 -
23 Reservoirs 29 -
24 Mountain wadis 22 -
25 Sealed surface 19 -
26 Mining site 1 -
The remaining classes have been aggregated to 6 major classes accounting for the most dominant factors that affect land cover in the research area: altitudinal zonation and human interference (Finckh 2008). The three altitudinal zones (mountainous vegetation, brush steppe and Hamada steppe) have been split in degraded and non-degraded conditions, thus resulting in 6 classes.
Temperatures that affect plant growth in a semi-arid to arid environment have been compiled by White et al. (2008). Along an altitudinal gradient in the Northern Chihuahuan desert, grass and brush steppe as well as mountainous vegetation have been characterized. Minimum growing temperature is set to 5°C; optimum growing temperature is 27°C for the steppes and 22°C for the mountainous vegetation.
Leaf Area Index (LAI) may be most simply described as leaf area (m²) divided by ground area (m²). This interpretation is also adopted by the SWAT model. Several other definitions exist; Asner et al. (2003) provide an overview. LAI can be determined using different methods: measurement (e.g. Asner et al. 2003), estimation based on the degree of ground cover (e.g. Diekkrüger 1996) or remote sensing techniques (e.g. Baret et al. 2007). All of these techniques except remote sensing have been applied in this study (Table 5-8 provides an overview). Baumann (2009) determined specific leaf areas (leaf area/leaf mass [m²/ kg]) for dominant plant types and then harvested and weighed leaves of 120 sample plots in the research area. Within the catchment plant coverage degrees have been estimated for all vegetation units as displayed in Figure 5-13 (Fritzsche 2011).
Figure 5-13: Plant coverage in the Upper Drâa catchment. Median, error bars envelope the 2nd and 3rd quartile, i.e. the middle 50% of observations (Fritzsche 2011)
Though these values are highly variable, a clear increase of coverage with altitude can be stated. To account for dense or degraded conditions, the 75th percentile has been assumed as dense vegetation cover and the 25th percentile as degraded vegetation cover. Based on these coverage degrees LAI has been calculated using an empirical relationship, as proposed by steppe: 2.1; Asner et al. 2003). Since Asner et al. define LAI as the leaf area per area covered by the plant (LAIc); the given values have to be multiplied with the coverage degrees as well.
The minimum LAI has been estimated according to the dominant plant types in each vegetation unit. Mountainous vegetation and shrub steppe are dominated by shrubs, their minimum LAI is estimated to 75% of the maximum LAI. Hamada steppes are dominated by annual grasses, therefore the minimum LAI is assumed to be 0.
Plant heights are generally low, not accounting for dispersed trees. Since vegetation degradation affects coverage stronger than height of particular plants (Baumann 2009), the same height has been assumed for dense and degraded conditions. These are in particular 50 cm for mountainous vegetation and 30 cm for dwarf-shrub steppe and the Hamada grasses.
Table 5-8: Leaf area indices for the Upper Drâa catchment
Baumann
Maximum stomatal conductance represents the plants ability to minimize water losses through evaporation. In the mountains, where water availability is generally higher, a value of 0.004 m/s has been assumed (White et al. 2008). In the brush-steppe water-saving
mechanisms are more distinct; a value of 0.002 m/s has been assumed (White et al. 2008).
The Hamada’s annuals grow and blossom quickly after sufficient rain, therefore water-saving mechanisms are not required and a value of 0.005 m/s has been assumed (White et al. 2008).
Rooting depth is the essential parameter for determining to what extent plants can benefit from soil water storage. Haase et al. (1996) conducted tracer experiments with Retama Sphaerocarpa in semi-arid Southeastern Spain, concluding that their rooting depth exceeded 28 m. Excavated roots of Prosopis juliflora in Southern Arizona exceeded 50 m in length (Phillips 1963). Canadell et al. (1996) provide a comprehensive review of rooting depths for all terrestrial biomes, among these deserts (mean rooting depth 9.5 m) and sclerophyllous shrublands (mean rooting depth 5.2 m). Tamarix species, prevailing in the study area develop roots lengths of 20 m. Hence it can be assumed that roots are generally capable to penetrate the whole soil profile in the study area. In some cases they might as well reach the aquifer which gains importance considering the process of revaporation. Appendix 5 provides an overview on the plant parameterization used.
The Curve Number procedure is used in this model to calculate surface runoff. The Natural Resources Conservation Service provides Curve number values for a variety of semi-arid land cover types (NRCS 1986). The values have been chosen according to the following Table 5-9, accounting for land cover and soil properties. According to these values the tendency to produce surface runoff decreases with altitude, i.e. increase in vegetation cover. Due to the generally low coverage degrees, most parameters related to plant growth and plant cover introduced in this section can be assumed to be rather insensitive. The Curve Number instead, governing the discharge processes in this model, generally is the most sensitive parameter of the model (Veith et al. 2010).
Table 5-9: Runoff Curve Numbers Land Cover types in the research area (based on NRCS 1986)
Land cover SWAT-MAROC (Land cover NRCS)
Hydrological Soil Group
A B C D
Mountainous vegetation
(juniper steppe, good condition) 50 58 73 80 Mountainous vegetation, degraded
(juniper steppe, fair condition) 60 75 80 89 Brush steppe
(sagebrush steppe, fair condition) 40 51 63 70 Brush steppe, degraded
(sagebrush steppe, poor condition) 60 67 80 85 Hamada steppe
(desert shrub, fair condition) 60 71 81 89 Hamada steppe, degraded
(desert shrub, poor condition) 70 80 87 93
Common ranges as provided by the NRCS give only an idea of reasonable values (NRCS 1986), but appropriate parameter values in a specific catchment depend on different variables, such as subcatchment size, climatology and others (Veith et al. 2010; Simanton et al. 1996;
Hawkins et al. 2009). Therefore the adaptation of the values listed above during calibration is clearly needed. Furthermore the conceptual character of the Curve Number and the uncertainty already associated with the soil data needed to determine Soil Hydrological Groups and Curve Number make a further treatment of the Curve Number in the uncertainty analysis compulsory.
5.3.3.2 Irrigated Areas
For the irrigation subroutine the following parameters are required:
Evaporation coefficient: Kc [-]
A factor converting reference evapotranspiration into the actual plant water demand
Soil Storage: SOL_STOR [mm]
Pore space within the soil profile that can store water available to plants
Irrigation efficiency: IRR_EFF [%]
The fraction of irrigation water that is actually consumed by the irrigated plants
Fraction of Discharge available for irrigation: IRR_FRAC[%]
The fraction of discharge that can be diverted for irrigation purposes
Plant water requirement is calculated using the reference evapotranspiration (ET0, calculated using the Penman-Monteith approach as described in section 5.1.1 and 5.1.2.2), which is calculated using the climate data introduced in section 5.3.6.1, which are adapted to the oasis conditions. Furthermore the area of each oasis and the respective crop mix are required. The values are set according to the agricultural census of 1997/98 (MTP 1998). The surface areas of the oases are: M’Semrir 1200 ha, Boulmalene 2553 ha, Kelaat 2673 ha, Skoura 1725 ha, Toundout 2635 ha and Ouarzazate 5215 ha. The crop mix is presented in Appendix 6. The locations of the oasis can be seen from Figure 3-1. The following
Table 5-10 gives an overview on the CropWat parameters used within the simulation. The Kc -values of each culture vary according to the seasonal plant growth cycle. The following perennial (p) or seasonal (s) crops are grown in the catchment: Alfalfa (p), Barley (s), Pulses (s), Vegetables (p), Wheat (s) and Corn (s) as second crop in summer. Furthermore date palms, fruit trees (Apple, Apricot, Almond) and olive trees are grown. A density of 300 per ha is assumed for palms and 400 per ha for other trees (Siebert et al. 2007).
Table 5-10: CropWat parameters for crops grown in the Upper Drâa catchment according to 1) Moroccan authorities (MTP 1998) and 2) CropWat Manual (Allen et al. 1989a)
Stage [days] Kc Value [-]
Since oases are located close to the wadi beds, in general deep alluvial soils can be assumed (Klose 2009). Therefore soil storage is set to 500 mm. This relatively high value also accounts for deep rooting palm trees and groundwater storage that is accessed by pumping, hence not only water stored in the soil profile.
In this study, irrigation efficiency is defined as the fraction of applied irrigation water that is not lost to evaporation. Percolating water is not actually lost from the system, but can be pumped or consumed by deep-rooting plants. A range of reasonable values is given by Bos &
Nugteren (1990) and Foster & Perry (2010). Since flood irrigation in basins is the dominant irrigation type in the study area, an irrigation efficiency of 75% has been assumed (cf. Table 5-11). This value is rather high compared to other studies conducted in the study area (Klose et al. 2010b), but lines in with values estimated by local authorities (ORMVAO 1995). A further parameter required in the irrigation module is the fraction of runoff that can actually be abstracted; it is set during calibration. This fraction reflects difficulties to fully exploit low flow, as the river stage might fall below irrigation channels, and high flow as the irrigation systems capacity might be surpassed.
Table 5-11: Irrigation efficiencies (Foster & Perry 2010)
Irrigation technology
(40-80%) Moderate but variable Moderate (20–30%)
Both latter parameters have to be considered in the uncertainty analysis, as they represent rather conceptual parameters and their effect on total irrigation volumes is high. The site-specific Kc-values have been adapted from local sources (MTP 1998) and their auditability is assumed.