**Hong Hanh Nguyen \*, Friedrich Recknagel and Wayne Meyer**

Department of Ecology and Environmental Sciences, University of Adelaide, Adelaide SA 5005, Australia; friedrich.recknagel@adelaide.edu.au (F.R.); wayne.meyer@adelaide.edu.au (W.M.)

**\*** Correspondence: hanh.nguyen@adelaide.edu.au; Tel.: +61-044-919-2658

Received: 7 November 2018; Accepted: 11 December 2018; Published: 13 December 2018 -

**Abstract:** Urbanization poses a challenge to sustainable catchment management worldwide. This study compares streamflows and nutrient loads in the urbanized Torrens catchment in South Australia at present and future urbanization levels, and addresses possible mitigation of urbanization effects by means of the control measures: river bank stabilization, buffer strip expansion, and wetland construction. A scenario analysis by means of the Soil and Water Assessment Tool (SWAT) based on the anticipated urban population density growth in the Torrens catchment over the next 30 years predicted a remarkable increase of streamflow and Total Phosphorous loads but decreased Total Nitrogen loads. In contrast, minor changes of model outputs were predicted under the present urbanization scenario, i.e. urban area expansion on the grassland. Scenarios of three feasible control measures demonstrated best results for expanding buffer zone to sustain stream water quality. The construction of wetlands along the Torrens River resulted in the reduction of catchment runoff, but only slight decreases in TN and TP loads. Overall, the results of this study suggested that combining the three best management practices by the adaptive development of buffer zones, wetlands and stabilized river banks might help to control efficiently the increased run-off and TP loads by the projected urbanization of the River Torrens catchment.

**Keywords:** SWAT; urbanization; nutrient loads; constructed wetlands; buffer zones; river bank stabilization

#### **1. Introduction**

Urbanization is the most common trend in land use changes worldwide, with approximately half of the global population residing in disproportionately small areas of land [1–3]. The urbanization of catchments is associated with sealing, compaction, degradation, and mixing of natural soils with imported soils [4,5], and requires informed sustainable management. Increased runoff and erosion rates, degraded water quality, reduction in biodiversity, wetland loss, and eutrophication are some of the consequences of rapid urbanization [6,7]. Analysis of 106 river catchments worldwide found that the proportion of catchments with streamflow being fragmented and disturbed by dams in urban areas is projected to increase to 70% by 2050 [7]. In Australia, natural catchments have been drastically altered since European settlement by land clearing and development of cities. Approximately 90% of the Australian population is living in urban areas [1] and many catchments face the risk of elevated nutrient loads and substantial algal blooms [8–10]. Thus, studies that allow quantitative evaluation of effects of urbanization are of great importance for the future water-sensitive design of Australian cities [11]. Catchment modeling has been defined as an important tool to assist this target [11,12].

The Soil and Water Assessment Tool (SWAT) is a widely used catchment modeling tool that allows to predict streamflow and non-point source pollutants under varying soil, land use and management conditions worldwide [13–15]. The SWAT model was originally developed to simulate rural catchments, but algorithms describing urban processes were later incorporated in the model [15]. Results from many studies by the SWAT have suggested that urbanization causes significant alterations in the water budget of catchments by increasing surface runoff and decreasing baseflow in streams [16–20]. Some studies have also reported a linear relationship between the speed of urbanization and the increase in sediment and nutrient loads [21,22]. In the study by Lee et al. [23], the projection of urbanization for 2030 suggests increases of total nitrogen and total phosphorous in many catchments by up to 24% and 111%, respectively. As stated by Wang and Kalin [22], substantial urbanization on forest lands is expected to cause higher peaks for sediment and total phosphorous loads during wet seasons, whereas rapid urbanization may even have a stronger effect on nitrogen and phosphorous than projected climate change [24]. In the case of Australia, most studies on catchment urbanization have focused on hydrological impacts [25] whilst studies on nutrient loads have been applied to agricultural catchments [26–28].

The River Torrens catchment covers an area of 200 km<sup>2</sup> and is located in the central part of Adelaide, the capital of South Australia. It supplies drinking water, environmental flow, and fulfills recreational and conservational purposes for the capital city [29]. Urban development is affecting water quality of its tributaries and creeks [8].

This study focuses on modelling effects of urbanization on streamflow and nutrient loads in the Torrens catchment using the SWAT model. The study also examines the effectiveness of potential mitigating control options in response to future catchment urbanization, and improves understanding of this issue for urban catchment managers and policy makers. This case study may also of interest to modelers working on similar environmental problems around the world.

#### **2. Materials and Methods**

#### *2.1. Study Area*

The urban section of the River Torrens catchment below Gorge weir (hereafter called the Urban Torrens catchment) was used throughout this study for the SWAT model application. The study area includes the First to Fifth Creek and the River Torrens, which pass through the Adelaide Central Business District (CBD). The catchment lies between latitude −34◦51'23" and −34◦56'53" S and longitude 138◦32'55" to 138◦43'52" E. The altitude of this area ranges from 9 to 681 m with an average value of 214 m. The Mediterranean climate of the study area is characterized by a low average annual rainfall of 600 mm that is mostly concentrated in sporadic storm events in summer or during the wet winter.

#### *2.2. Input Data*

#### 2.2.1. Soil Data

The soil inputs required for the SWAT model comprise of soil maps and soil attribute data (Figure 1). The soil maps of the study area include the map of South Australia, which was provided by the South Australian Department of Agriculture, and a map of the urban area which was extracted from a project on mapping soils around metropolitan Adelaide by the Department of State Development [30]. This project was carried out to explore the properties of Metropolitan Adelaide soils, which include some reactive soils and clays that are sensitive to seasonal and human-induced changes and have caused severe failure of masonry buildings in many urban regions around Adelaide. Both maps were provided at the resolution of 1:100,000 and were clipped to prepare unique raster map using a geographic information system (GIS) tool.

**Figure 1.** Soil maps of the study area.

For the SWAT database, the major soil information was provided by the Australian Soil Resource Information System (ASRIS) [31], while the attributes of soils in the missing information area (Figure 1) were constructed on the basis of data available from the Drill Core Reference Library, published literature [30] and expert knowledge. Information from 27 data points drilled to 10 m depth [28] was combined to develop eight major soil classes: black earth (BE), brown solonized (BS), estuarine sediments (EM), podzolic (P2), red-brown earths (RB3, RB5, RB6), and solodic (SK) soils. These soil classes were further characterized by soil attributes comprised of soil layers, soil hydrological groups, plant root depth, soil dry bulk density, soil organic content, and percent of clay, silt, sand, and rock fragments. Some soil parameters were estimated using the following functions [32,33]:

$$
\theta\_p = 0.132 - 2.5 \times 10^{-6} \times \text{e}^{0.105 \times \text{@sand}} \tag{1}
$$

$$K\_{\rm sat} = 750 \times \left(\frac{\theta\_{\rm sat} - \theta\_d}{\theta\_d}\right)^2 \tag{2}$$

where *θ<sup>p</sup>* (m<sup>3</sup> H2O/m3 soil) is the soil available water content (SOL\_AWC) (mm H2O/mm soil), *Ksat* is the saturated hydraulic conductivity (SOL\_K) (mm/day), *θsat* is the upper limit of water content that is possible in a soil of known bulk density and *θ<sup>d</sup>* (m<sup>3</sup> H2O/m<sup>3</sup> soil) is the volumetric drained upper limit water content.

*θsat* was calculated from Soil bulk density (*ρd*) (g/cm3) using the following formula:

$$
\theta\_{sat} = 1.0 - \frac{\rho\_d}{2.65} \tag{3}
$$

*θ<sup>d</sup>* was calculated from the gravimetric drained upper limit *wd* (kg H2O/kg Soil) and *ρd*:

$$
\theta\_d = w\_d \times \rho\_d \tag{4}
$$

$$w\_d = 0.186 \times \left(\frac{sand}{clay}\right)^{-0.141} \tag{5}$$

These equations have been successfully applied to derive soil characteristics for a range of soils in south eastern South Australia [34]. The soil erodibility (USLE\_K) parameter (0.013 (metric ton m2 h)/(m3-metric ton cm)) was estimated from relative proportions of sand, silt, and clay in each soil layer using the method provided in the SWAT manual [15].

The resulting attributes of the soil profile of the Urban Torrens catchment are provided in Table 1. These include average data for two soil layers of soil classed which are estimated from drill hole information and more detailed data on five soil layers of soil classes provided by the ASRIS source.


(mm/h); SOL\_CBN: organic carbon content (%); CLAY: clay content (%); SILT: silt content (%); SAND: sand content (%); SOL\_ALB: moist albedo; USLE\_K: Universal Soil Loss Equation(USLE)equationsoilerodibility(K)factor(0.013(metrictonm2 h)/(m3-metrictoncm)).

#### *Water* **2018** , *10*, 1846
