DECLARATION 2: PUBLICATIONS
3. DEVELOPMENT OF AN IMPROVED COMPREHENSIVE CONTINUOUS
3.1 Introduction
3. DEVELOPMENT OF AN IMPROVED COMPREHENSIVE
cover class in good hydrological condition (> 75% plant cover). In the SCS-SA adaptation of the SCS (1956) land cover and soils classification, the concept used to define soils into hydrological soil groups is slightly different and the number of soil groups was increased from four to seven (Table 3.1), in order to accommodate the wide range of soil types found in South Africa (Schmidt and Schulze, 1987a; Schulze, 2012). Group A soils have the highest infiltration and permeability characteristics and vice versa for Group D soils.
Table 3.1 Initial CNs for Row Crops for specific land management practice, hydrological condition, and soil group classes (after Schulze et al., 2004)
Land Cover Class
Land
Treatment/Practice/Description
Stormflow Potential
Hydrological Soil Group A A/B B B/C C C/D D
Row Crops
1 = Straight row High 72 77 81 85 88 90 91
2 = Straight row Low 67 73 78 82 85 87 89
3 = Straight row + conservation tillage High 71 75 79 83 86 88 89 4 = Straight row + conservation tillage Low 64 70 75 79 82 84 85
5 = Planted on contour High 70 75 79 82 84 86 88
6 = Planted on contour Low 65 69 75 79 82 84 86
7 = Planted on contour + conservation
tillage High 69 74 78 81 83 85 87
8 = Planted on contour + conservation
tillage Low 64 70 74 78 80 82 84
9 = Conservation structures High 66 70 74 77 80 82 82
10 = Conservation structures Low 62 67 71 75 78 80 81
11 = Conservation structures +
conservation tillage High 65 70 73 76 79 80 81
12 = Conservation structures +
conservation tillage Low 61 66 70 73 76 78 79
The most comprehensive land cover classification available for use with the ACRU model is the COMPOVEG database (Schulze, 1995; Smithers and Schulze, 2004). The COMPOVEG database contains default assigned parameter values required by the ACRU model to represent five land cover categories, namely urban land uses, agricultural crops, natural vegetation, aquatic systems and commercial forests, as classified by Schulze and Hohls (1993) and depicted in Figure 3.1.
Figure 3.1 The four-level structure of the land cover/land use classification developed for the ACRU model (Schulze, 1995)
The land cover classification does not explicitly account for land management practices associated with agricultural crops, as accounted for in the SCS-SA classification (Table 3.1).
Since the ACRU model is a daily timestep CS model, the land cover classification does account for different crop development stages, i.e. from planting to harvest, and accounts for regional differences in planting dates for specific dominant crops cultivated extensively in different parts of the country, such as maize and wheat (Figure 3.1). The classification also distinguishes between commercial and subsistence crops, however, does not explicitly represent the land management practice and hydrological condition for each. In terms of natural vegetation, the classification includes classes to represent good, fair and poor hydrological condition for selected land cover classes such as veld (grassland). This, however, is not consistently represented for all natural land cover classes. Furthermore, Rowe et al. (2018) identified that the ACRU model is insensitive to the parameters adjusted and used to represent the different
hydrological condition classes, in terms of design flood estimates. This was particularly concerning, based on the comparative changes in stormflow response simulated by the SCS- SA model for similar changes in hydrological condition for similar land cover classes. Rowe et al. (2018) therefore, performed a sensitivity analysis of the ACRU model to selected parameters to identify which parameters to use, to more adequately represent the change in stormflow response for different land management practices and hydrological conditions. It is important to reiterate that, in this study, the assumption has been made that the hydrological responses simulated by the SCS-SA model, through the CN, for the range of land cover classes defined in the SCS-SA land cover classification, are reasonable and representative of these land cover classes. The reliance on the SCS-SA model and associated CNs is attributed to the absence of observed data on hydrological responses from land cover classes and soil combinations as defined in the SCS-SA land cover classification. Further justification for the use of the results simulated by the SCS-SA model, i.e. as a surrogate for observed data to simulate similar magnitudes and changes in stormflow response in ACRU, is gleaned from the fact that the CNs adopted in the SCS-SA model were at least calibrated using observed data for a range of land cover / soil conditions (Mishra and Singh, 2003). Rowe et al. (2018) identified two parameters to represent land management practice and hydrological condition in the ACRU model, namely:
(i) the Quick Flow Response Coefficient (QFRESP) which partitions stormflow into a same day response fraction and a subsequent delayed stormflow response, and (ii) the Critical Hydrological Response Depth of the Soil (SMDDEP). These parameters are currently generally set to recommended default values, however, some guidance on the selection of SMDDEP is provided in the ACRU Theory Manual on the basis of vegetation density, soil conditions, climate and rainfall intensity (Schulze, 1995). Rowe et al. (2018), however, developed a consistent methodology to parameterise these two parameters using SCS-SA CNs.
Consequently, linking both of these parameters to physically measurable soils and land cover characteristics of a catchment, including land management practices and hydrological conditions. For context, a summary of the methodology applied by Rowe et al. (2018) is provided in the section to follow.
The objectives of this chapter are to: (i) build on the investigations and results of Rowe et al.
(2018), and (ii) to incorporate these developments into a comprehensive CSM system for DFE in South Africa. The idea is to start with a simple system similar to, and based on, the SCS-SA model (Schulze et al., 2004), in order to facilitate migration from the SCS-SA approach to the
ACRU CSM approach in practice. The objective is in line with recommendations from the international literature as reviewed in Chapter 2, e.g. the United Kingdom and Australia, of simplicity and user friendliness, while still providing accurate results. It is hypothesised that a system that incorporates the valuable information calibrated into the CN along with explicit soil water budgeting will provide the most accurate results when simulating flows for different land cover and soil combinations. Additional motivation lies in the realisation that the SCS-CN method is still widely used (Brocca et al., 2011; Grimaldi et al., 2012; Rossman, 2015; USACE, 2016).