While it has been established that the tools reviewed cannot be used for this exercise, the methods and criteria used for determining suitability of NBS in these tools have a good potential for use in allocation of large-scale NBS. A novel method and tools for mapping suitability of four large-scale NBS: floodplain restoration, detention basin, retention pond, and river widening are developed by: (1) determining criteria for spatial allocation of NBS; (2) development of a conceptual model; (3) development of toolboxes for spatial allocation of large-scale NBS; and (4) applying the toolboxes in a RECONECT study area.
Determination of Criteria for Spatial Allocation of NBS
The general criteria were determined by further analysing the seven tools in the literature review. Based on this analysis, the criteria include site conditions, catchment characteristics such as topography, streams, and water bodies, land use and land cover, roads and infrastructure, and performance requirements such as risk reduction and water management (Table 2). To reduce overlap and repetition of the criteria, they are divided into three distinct and broad categories: (1) Biophysical (natural aspects of the landscape including topography, hydrology, soil, water bodies, etc.); (2) Planning and governance (different types land use such as agriculture, industry, etc.), urban fabric, and infrastructure); and (3) Performance Requirements (service requirements such as reduction of natural hazards, water management, etc.).
In addition to utilising similar input data, the tools also use similar criteria to determine site suitability for NBS. Based on the score, the most commonly used criteria for spatial allocation of NBS in the analysed tools, with a score ≥ 4 (for a simple majority of tools analysed), are slope, soil type/class, imperviousness, distance from stream, land use type/zone, urban land use and road buffer. Imperviousness is directly related to changes in land use (Federal Interagency Stream Restoration Working Group (FISRWG) 1998), and urban land use can be characterised as land use type over larger scales, these two criteria can be represented in land use type/zone.
Slope, soil type/class, distance from stream, land use type/zone, and road buffer are used as general criteria for spatial allocation of large-scale NBS as existing tools demonstrated their applicability for spatial allocation over different catchment scales and a range of NBS.
Using the general criteria required for spatial allocation as slope, soil type/class, distance from stream, land use type/zone, and road buffer, a conceptual model (Fig. 1) is developed to illustrate the input and processes needed to produce a suitability map as an output. The base maps needed for the model, i.e., input data, are defined by each criterion. The slope can be derived from a digital elevation model (DEM), soil class/type can be produced using soil maps, distance from stream can be estimated using maps of the streams and rivers, land use type/zones can be determined from land use and land cover maps, and road buffers can be determined using maps of road infrastructure.
The derived maps are transformed using conditions for each criterion to produce maps which show areas where each condition is met. These maps are then combined to produce a general suitability map that delineate areas where all base conditions or general criteria are map. The general suitability map is coupled with NBS specific criteria to create maps that show suitability for each NBS.
Setting Up Toolboxes for Mapping Suitability
The conceptual model (Fig. 1) is utilised to create a toolbox in ESRI ArcMap 10.5, its model builder, and spatial analysis tools. ArcMap provides a powerful GIS interface where the conceptual model can be implemented using tools that can be used for a variety of functions including determining hydrological characteristics in a catchment, raster calculations, interpolation, and estimation of areas and volumes. The conceptual model is utilised to define the base maps i.e., input data needed for the model and the processes used to derive, transform, and combine GIS data to produce suitability maps.
Base maps are used to derive maps using slope and Euclidean distance (i.e., the shortest distance between two points) calculations for slope rate and distance from rivers and roads, respectively. The derived maps are raster images conforming to resolution of the input DEM, which are then transformed to Boolean maps to show areas that meet the conditions; (1) slope rate ≤ 5 %, (2) distance from river ≤ 1 km, and (3) distance from for ≥ 50 m. The land use map is reclassified to a Boolean scale, to distinguish between areas that are permeable to land use change (e.g., agricultural land, parks, natural land) and those impermeable to land use change (e.g., urban areas, industrial zones). Utilising a Boolean scale allows combination of several layers using simple logic operations as all derived maps are on the same simplified scale. The transformed Boolean raster maps are converted to vector layers and a geometric intersection of all the four layers is computed to produce discrete polygons that show areas in the watershed where all the spatial allocation criteria are met. This produces the general suitability map of the methodology that is used as general criteria for spatial allocation.
The general suitability map is combined with measure specific criteria, which are dependent on the type of NBS, to locate regions where these measures can be applied. The Flow length tool is used to produce a raster of upstream and downstream distance, along the flow path of the catchment. This is divided into three sections to delineate upstream, midstream, and downstream sections of the catchment along the river for determining areas best suited for floodplain restoration, detention basins, and retention ponds. A buffer zone from the banks of the river is similarly used as a measure-specific criterion for river widening.
The suitability map for each NBS consists of vector layers consisting of polygons for floodplain restoration, retention pond/detention basin and lines for river widening, which constitute the output of the tools.
Verification of Approach for Mapping Suitability
The methodology only maps suitability and does not select and size specific locations for the NBS. In lieu of hydrological and hydraulic models, the suitability maps produced are assessed using two approaches:
Comparing suitability maps with satellite or digital images, using Google Earth Pro, to identify areas where the suitability may not match the actual or existing conditions of the study area. The suitability maps are laid over digital images of the area to identify where produced suitability maps may encroach on urban areas and other spaces that are not permeable for land use change;
Flood maps and suitability maps are utilised to generate interpolated surfaces using the Inverse Distance Weighting (IDW) tool. The Surface difference tool in the 3D analyst tools is used to calculate displacement between the generated surfaces and the catchment surface (DEM) to estimate volume between the two surfaces. This gives an estimate of the volume of floodwater and the storage capacity available for the NBS. The suitability maps can be used to select NBS sites that have enough potential to store the estimated volume of floodwater.
The case study area used to test the methodology is from the EC-funded H2020 RECONECT project, Tamnava River basin, in Serbia (Fig. 2). Tamnava River is a tributary of the Kolubara River; an 87 km long river that flows through western Serbia, which itself is a tributary to the river Sava. The Tamnava basin stretches over an area of 746 km2, and houses the rivers Tamnava and its main tributary Ub. The Tamnava River flows into the Kolubara River about 7 km south of the town of Obrenovac, lying on the confluence of the Kolubara River with the Sava River. The Kolubara basin houses considerable lignite reserves that supply approximately half of the national thermal energy. Lead, zinc, rare antimony, are some of the minerals mined in the region (Serbian Environmental Protection Agency (SEPA) 2014).
The Kolubara River basin’s hydrological and geomorphological characteristics make the area susceptible to flood waves. The Tamnava River basin suffered from devastating floods in 1999, 2006, 2009, and 2014. Despite mitigation efforts, flooding continues to endanger lives, agriculture, infrastructure, and private and industrial properties in this area (Babić-Mladenović and Kolarov, 2016). In May 2014, record-breaking rainfall of more than 200 mm occurred over western Serbia within a week due to low-pressure system “Yvette”. This rainfall equivalent of 3 months led to extensive flooding in urban and rural areas over whole Serbia, but especially in the Kolubara River basin (Plavšić et al. 2014). High intensity flash floods devastated houses, parts of the roads, and bridges, while the casualties were also reported. The overall estimated damage and losses exceeded EUR 1.5 billion (Government of the Republic of Serbia 2014).