Nature-based solutions for flood–drought mitigation using a composite framework: a case-based approach

It has been estimated that by 2050 more than two-thirds of the world's population will live in cities (Nações Unidas 2019; Gupta & Bharat 2022b). The increased pace of urbanization with global population growth demands additional land (Gupta & Bharat 2022a), creating more built up areas with increased impervious surfaces. Global water demand accounts for an increase of 20–30% in the current level of water usage. This is due to the rising demand in the industrial and domestic sectors (UNESCO 2018). According to a report by the World Meteorological Organization (WMO) (2021), 3.6 billion people had inadequate access to water for at least 1 month per year in 2018. By 2050, this is expected to rise to more than 5 billion. An estimated $651 billion (USD) in flood damages occurred globally from 2000 to 2019 (Tellman et al. 2021), whereas from 1998 to 2017, droughts caused global economic losses of roughly USD 124 billion (Daniel et al. 2022).

India, being a land of varied geographies and micro-climates, is exposed to extreme climate risks (Rao 2019). The endless undulations from extreme heat and drought to extreme rain and floods have affected India, a country of 1.3 billion people (Masson-delmotte 2018; Rao 2019). The total flood-prone area of India is around 40 million hectares which is equivalent to 12% of the total area of the country (Joshi 2020) whereas over a fifth of India's land area (21.06%) is facing drought-like conditions, according to recent data released by Drought Early Warning System (DEWS), a real-time drought-monitoring platform (Shagun 2021). Urbanization has modified the working of natural systems which includes the built cover and the natural hydrological cycle (Sharma & Bharat 2011). Impermeable surfaces, compacted soils, reduction in vegetation cover, and increased built-up areas have reduced the ability to intercept and infiltrate stormwater and surface runoff (Whitford et al. 2001). As a consequence, the risk of flooding has risen. The increase in impermeable surfaces has reduced the water percolation capacity leading to a situation of drought (Sharma & Bharat 2011).

To benefit and mitigate the double-edge extremes of flood and drought, the concept of ‘NbS’ is considered which is an umbrella concept that uses nature to provide economic, social, and environmental benefits (Maes & Jacobs 2015; United Nations Global Compact 2019; Baldwin et al. 2022). They underpin the Sustainable Development Goals (SDGs) (United Nations 2019), supporting vital ecosystem services, biodiversity, and access to fresh water, improved livelihoods, healthy diets, and food security from sustainable food systems (United Nations Global Compact 2019; Gupta & Bharat 2022b). They value harmony between people and nature, as well as ecological development (Yanfen 2020). NbS represents a holistic and people-centric response to climate change (Seddon et al. 2020). They are effective and globally scalable (United Nations Global Compact 2019). For instance, measures for water harvesting can be developed to satisfy a dual purpose of flood prevention in addition to water harvesting. Such dual/multi-functional measures may range from local and city/district scale (e.g., green infrastructure, such as green roofs, green walls, and rain gardens) to whole catchment scale (e.g., using natural and constructed wetlands for flood control) (Kalantari et al. 2018; Baldwin et al. 2022).

In the study, flood is considered a water surplus situation with a heavy concentration of rainfall, un-tappable runoff due to inadequate drainage, and an increase in impervious surfaces (Wright 2007), whereas drought is considered a condition of water scarcity with a lack of adequate rainfall, the surface and the groundwater levels being low and reduced soil moisture. To understand the effect of the bipolar events, the study is conducted at the district level for Banaskantha district, Gujarat, and validated in the Muzaffarpur district. This study proposes a novel rubric in its own way that can be applied as an initial study for districts affected by both flood and drought. It aims for a mutual solution towards flood–drought through NbS which can be technically applied in affected areas.

A set of indicators were identified through a rigorous literature review to study flood and drought and also to identify potential zones for groundwater recharge. These include: rainfall (in millimetres) (Gautam & Bana 2014) – excess rainfall causes overflowing leading to a flood like situation whereas when the frequency and intensity of rainfall is less, it tends to a drought-like situation; soil type (Ochoa et al. 2019b) – different soils have different water holding capacity, moist soil relieves the deficit situation but in turn decreases the capacity to infiltrate excess water, which in the long run might lead to a drought-like situation; elevation (in metres) (Ochoa et al. 2019a) – the higher the elevation, lesser is the situation of water accumulation and more are the chances of water deficit as water does not accumulate and infiltrate whereas when the elevation is low, the tendency of water accumulation increases which in turn creates a situation of water surplus; slope (in %) (Ochoa et al. 2019a) – lower slope has less runoff capacity, therefore more water accumulates leading to a more flood like situation and vice versa, land use/land cover (LULC) (classified from the satellite data) (Ehsani et al. 2017) – different land uses have different roughnesses which are different towards the flow of water and its accumulation and the drainage density varies with it (Gautam & Bana 2014) – the higher the drainage density, the higher the water accumulation and therefore the higher the chances of flood and vice versa.

To understand the importance of the selected indicators compared to each other, a comparison matrix was prepared with the help of the Multi Criteria Decision-Making (MCDM) technique of the Analytical Hierarchy Process (AHP) (Gupta & Bharat 2022c). The AHP, which is a general theory of measurement, is used to derive ratio scales from both discrete and continuous paired comparisons (Saaty 1987). The method is being used in varied applications like multicriteria decision-making, planning and resource allocation and in conflict resolution (Saaty 1987; Gupta & Bharat 2022c).

To address the issue of flood–drought for the pilot case of the Banaskantha district, the Soil Conservation Service-Curve Number (SCS-CN) method (Satheeshkumar et al. 2017; Kumar et al. 2021) for rainfall–runoff and Topographic Wetness Index (TWI) (Grabs et al. 2009; Ballerine 2017; Mattivi et al. 2019; Meles et al. 2020) were applied for flood analysis. For drought analysis, the Normalized Differential Water Index (NDWI) (Liu et al. 2004; Gu et al. 2007; Shashikant et al. 2021) and the Soil Water Index (SWI) (Robinson et al. 2008; Leeuwen 2015; Esit et al. 2021; Syawalina et al. 2022) were applied. Groundwater recharge potential zones (Yeh et al. 2016; Senthilkumar et al. 2019; Kaewdum & Chotpantarat 2021) were mapped using the selected indicators. The SCS-CN method has been widely used to estimate runoff (Khaddor et al. 2015; Pathan & Joshi 2019; Al-Ghobari et al. 2020; Patel 2020; Kumar et al. 2021). After overlaying the flood severity map, drought severity map and groundwater recharge zones, an integrated flood–drought severity map was computed. A notional 3 × 3 rubric was then developed to provide probable best solutions through ‘NbS’ to flood–drought severities and were validated on the study case of the Muzaffarpur district. The following methods were used to compute flood, drought, and groundwater recharge zones in the following ways.

The SCS-CN method (Cornell University 2008; Satheeshkumar et al. 2017) was used to calculate runoff. It relies on only one parameter, CN (Curve Number) (Ara & Zakwan 2018).

For determining the CN value, the LULC classification map and the hydrological soil group map (HSG) were overlayed in ArcGIS. After computing, the LULC soil map in ArcGIS, along with developing the CN lookup table, the CN values for the study case were determined. With the computed CN values, S (potential maximum retention) was computed using the relationship as [S = (1,000/CN) − 10]. Runoff was calculated using Q = (P − 0.2 S)²/(P + 0.8 S), where Q represented actual runoff in mm, P was precipitation in mm, and S was potential maximum retention.

The TWI (Mattivi et al. 2019; Kopecký et al. 2021) was applied to detect surface water accumulation zones, which helped in understanding the potential of accumulation that causes floods (Ballerine 2017). It helped in identifying wet areas (Meles et al. 2020) which were at high risk of flooding and vice versa. Typically, the TWI indicates a range from −3 to 30 (Ballerine 2017). A lower index value has low accumulation, whereas higher values represent areas with increased accumulated runoff potential which may cause floods. It was derived using the digital elevation model (DEM) (Deenik 2021).

The NDWI was used for water analysis (JRC European Commission 2011; Das 2017). It highlighted surface water. Remote sensing data from Landsat 8 were used in the study to calculate the NDWI. The formula is as follows: NDWI = (Band 3 – Band 6)/(Band 3 + Band 6), where Band 3 = Green band; Band 6 = Short-Wave infrared (SWIR) band (Landsat 8 imagery) (Özelkan 2020). The spatial analyst tool in ArcGIS helped in performing raster calculations. The resultant raster image has a value between −1 and 1. By reclassifying the resultant raster image (for value ≥0), the surface water body count was generated. Surface water body area then can be calculated by multiplying the count with the generated image cell size (Sahu 2014).

The SWI was applied to quantify the moisture condition of the soil (Robinson et al. 2008; Richaud 2019). It helped in understanding the water stress levels (Robinson et al. 2008; Richaud 2019). The relationship between land surface temperature (LST) (Syawalina et al. 2022) and the normalized difference vegetation index (NDVI) gave the SWI (Leeuwen 2015). It directly helped in indicating drought severity for the study (Dobriyal et al. 2012; Richaud 2019; Zbiri et al. 2019). The values ranged from 0 to 1 (Leeuwen 2015), where values near 1 have high vegetation with low surface temperature and a higher level of soil moisture (Leeuwen 2015; Syawalina et al. 2022). The values near 0 have less vegetation with high surface temperature and lower levels of soil moisture (Leeuwen 2015; Syawalina et al. 2022). Here, as the soil moisture increases, the chance of drought decreases and vice versa (Esit et al. 2021; Zeri et al. 2022).

With the help of the selected six indicators, weighted overlay analysis in GIS was conducted to identify groundwater potential recharge zones (Yeh et al. 2016). These zones could be used to reduce a water surplus situation. By directing the excess water into these recharge zones, it could help in mitigating water stress by reviving the groundwater aquifers. By adding all the weighted map layers of the selected indicators, a resultant map was generated which contained groundwater potential recharge zones. Input factors were given value through a robust literature review and accordingly, weights were assigned through the AHP. The weightages given in percentage were as follows: Drainage density – 35%, Rainfall – 12%, Slope – 9%, DEM – 18%, Soil – 16%, and LULC – 10%.

After computing the flood and drought scenario and the potential groundwater recharge zones, the generated maps were overlayed using weighted overlay in ArcGIS. Weightage and ranks were given through the AHP. After overlaying, an integrated flood–drought severity map was computed. It was divided into nine different flood–drought severity scenarios. To provide solutions to the flood–drought severity scenarios, a notional 3 × 3 rubric was developed. It was developed with specific NbS derived through a robust study of literature. This study was conducted on the pilot case of the Banaskantha District, Gujarat. A ground reality check was done for the pilot case and an applicability rubric was developed and later, the solutions were validated in the study case of the Muzaffarpur district by computing a similar applicability rubric.

Due to varied geological formations from Archean to Alluvium (Goyal & Sharma 2015), only Central Banaskantha has a high potential for groundwater development and is suitable for deep tube wells (up to 30 m), while the eastern, western, and northern parts of the district have negligible recharge capacity due to the presence of hard rock formations and saline aquifers (Government of India and Ministry of Water Resources Central Ground Water Board West Central Region Ahmedabad 2011; Gujarat State Disaster Management Authority (GSDMA) 2016).

The availability of water in the system, i.e., using the excess from the floods to reduce the deficit of others, i.e., drought, is the intent behind the study. Floods due to excess stormwater runoff and droughts due to scarcity of soil moisture are considered for the study.

To compute the runoff, the average annual rainfall (P) of 0.09 mm for the year 2021 was used as the major input. Different CN values ranging from 71 to 100 were computed by overlaying the LULC map (Figure 8) and the hydrologic soil group map (Ross et al. 2018) (extracted from the Global Hydrologic groups) in ArcGIS for the pilot case of the Banaskantha district. Different retention values (S) were then computed and inputted in the runoff equation. After applying the equation, different values of runoff (mm) (Table 1) were computed.

The NDWI used remote sensing data (Landsat 8 Collection 1 Level 1) from the archive of USGS for the past 7 years for pre-Monsoon (January–March) and post-monsoon (October–December) period. It was developed using near infrared and SWIR bands. The resultant values helped to compute existent varied areas of surface water bodies over the selected years (Table 2). The areas were computed for all 7 years for both pre- and post-monsoon period in km².

A process of weighted overlay in ArcGIS was carried out for the demarcation of potential groundwater recharge zones in the district. Six different maps were used to compute the map using a function of Map algebra in ArcGIS. The soil map was developed using the World soil data. The slope map was developed in ArcGIS using the DEM data. The DEM was directly extracted from Aster data (Earth Data) and masked according to the district boundary. IMD gridded data were used to develop the rainfall map. A drainage density map was developed using the DEM data. The LULC map was developed through supervised classification in ArcGIS using the Landsat 8 imagery downloaded from Earth explorer (USGS Archive – 15 October 2021) to demarcate different land uses in the district.

Five major zones were identified for groundwater recharge with different areas in km² (Figure 8).

The computed flood severity map, drought severity map, and groundwater recharge potential map of the pilot case were overlayed to develop an integrated flood–drought severity map. Nine scenarios of flood and drought were selected for the pilot case. Later, a notional rubric was developed using the NbS as the rubric for the best probable solutions for the nine scenarios. It was overlayed with ground reality to check its feasibility to develop an applicability rubric. The solution rubric was then applied to the study case to check its validity by developing a different applicability rubric. To capture water through the recharge zones, their percolation capacities should be maximum and unobstructed. Incorporating solutions that were nature-friendly would help in creating absorption zones – natural or artificial and in turn increase infiltration. It was anticipated that the issue of water stress would be resolved only if the excess water from the runoff was utilized to balance the hydrological cycle.

The selected six indicators which were compared using the AHP achieved a consistency ratio of 0.035 (Figure 5). The highest weights were computed for rainfall, drainage density, and soil type. The SCS-CN method for rainfall–runoff computed different runoffs which varied from 0.09 to 5.14 mm (Table 1). The generated values for the TWI ranged from a high of 24.6 to a low of 2.9 (Figure 6) (Ballerine 2017). When 7 years of satellite data were analysed using the NDWI, it ranged from −1 to +1 and variations could be seen in surface water bodies which were achieved from 64 to 179 km² pre-monsoon to 56–191 km² post-monsoon (Table 2). The SWI computed values ranged from 0 to 1 (Figure 7) (Saha et al. 2018). The resultant map depicting groundwater recharge potential zones (Figure 8) identified five categories of zones with maximum potential for groundwater recharge having a total area of 88 km² in comparison to the total district area of 10,743 km².

The best probable solutions for the existing nine situations of the pilot case of the Banaskantha district (Figure 9) were notionally depicted in a 3 × 3 rubric (Figure 10). The solutions provided were based on a robust literature study for NbS. These solutions were conceptual in nature and they needed checks and validation.

To check and develop the applicability ratio of the suggested best probable solutions, they were overlayed on the ground to check their validity and feasibility through the use of Google Earth (Akanbi & Agunbiade 2013). Google Earth image was added in ArcGIS with UTM coordinates (to match it with the coordinate system of ArcGIS – Tiff images) to validate the solutions by geo-referencing and auto-adjusting the image underneath the suggested probable solutions (Akanbi & Agunbiade 2013; Brian Klinkenberg 2019). The rubric (Figure 10) was then cross-validated and overlaid with the identified groundwater recharge potential zones, different soil types (Figure 8), and the DEM (Figure 8) to develop an applicability rubric (Table 3). By validating the nine situations from the solution rubric on ground, there were variations found.

The notional solutions from the rubric (Figure 10) were overlayed to validate their effectiveness. When the solutions provided were overlaid with ground reality of the district, an applicability rubric (Table 4) was developed. The applicability rubric was found not applicable to the solutions of the solution rubric (Figure 10).

As the study focuses on mutual flood–drought mitigation, minimizing stormwater runoff and maximizing water percolation is the best possible solution for the balanced working of the natural hydrological system. A lack of groundwater recharge capacity is a major concern. The groundwater resources have been overused in the Banaskantha district (Goyal & Sharma 2015). There is a loss of 4,094 × 10⁷ gallons (total runoff of 12.23 mm on an area of 12,703km²) of water (Table 1) and to fulfil the needs of the 31.21 lakhs of people in the district (Government of Gujarat no date), 1,173 × 10⁵ gallons of water is required.

Preventive and curative measures through NbS were proposed. To prevent the event from occurring or if it is in the initial phase, a preventive measure would be effective but curative measures were needed when the event had already occurred and a solution was required. Low-flood and low-drought severity areas could incorporate preventive measures so that the events could be prevented from becoming severe; medium severity areas could incorporate preventive and curative measures where if in case one event had low severity and the other had moderate severity, a combined measure could be applied; and in the case of high severity areas, curative measures could be applied if the event had already happened and needed a solution to mitigate it.

Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) (IPBES 2019) highlights the need to integrate biodiversity considerations in global decision-making. NbS, which are ‘nature-friendly’ solutions, are cost-effective ways of meeting SDGs (Gupta & Bharat 2022c). They are the ‘actions to protect, sustainably manage, and restore the natural or modified ecosystems, effectively and adaptively, simultaneously providing human well-being and benefits to the biodiversity’ (International Union for Conservation of Nature (IUCN) 2017; Partnership for Environment and Disaster Risk Reduction (PEDRR) and Friends of Ecosystem-based Adaptation (FEBA) 2020).

Banaskantha, which is a land with both the severities was checked for the solutions provided. Here, four cells (F1D2, F1D3, F2D3, and F3D1) were seen in full applicability with solutions, i.e., the solutions which could directly be applied as an initial process, three cells (F1D1, F2D1, and F2D2) were in partial applicability, i.e., solutions could be partially applied and two cells (F3D2 and F3D3) had no applicability, i.e., none of the proposed solutions were possible. This indicated that the solution rubric (Figure 10) developed could be used as an initial basis for the study as it was 44.44% (4:3:2) (Table 3) accurate on the ground. Later, in the study, to validate the solutions in the other case of the Muzaffarpur district, it was computed that two cells (F1D1 and F2D1) were in full applicability of incorporation, five cells (F1D2, F2D2, F2D3, F3D1, and F3D2) were in partial applicability and two cells (F1D3 and F3D3) had no applicability of incorporation of the solutions. This indicated the solution rubric (Figure 10) developed could be used for Muzaffarpur with a ground accuracy of 22.22% (2:5:2) (Table 4).

The study makes it possible to notionally overlay the solutions and check their effectiveness on the pilot case and validate it on the other case. In any area or region which has a similar situation of flood and drought, similar soil conditions and similar elevation profile, the proposed solutions (Figure 10) could be directly incorporated as an initial process to achieve mutual flood–drought mitigation. It will help in understanding the possible options as solutions and later customization would be required to achieve an integrated mitigated situation of flood and drought so that there is balanced working of the hydrological system.

This study, based on achieving mutual mitigation of floods and droughts through the incorporation of NbS, sees variations in the results for the pilot and the validation cases of Banaskantha and Muzaffarpur, respectively. They vary in spite of having nine similar flood–drought severity conditions as mapped in Figures 9 and 11. The probable solution rubric which is a technical solution rubric varies with the change in ground realities. Banaskantha has an applicability ratio of 4:3:2 (44.44% ground accuracy) (Table 3), whereas Muzaffarpur has a ratio of 2:5:2 (22.22% ground accuracy) (Table 4).

The applicability rubric which is developed as a contingency plan towards a situation of flood and drought uses of 3 × 3 rubric (Figure 10), which can be used directly as an initial study tool for any region having similar flood–drought conditions. These solutions help in maintaining the water availability of soil which can in turn reduce water stress conditions. It can be converted into an exhaustive study at a later stage after its direct application as an initial process. The choice of vegetative species, enhancement in the soil-bearing capacity, and creation of artificial recharge zones can be worked out in further studies. The use of NbS not only helps achieve mutual flood–drought mitigation but could also help in reducing the effect of climate variations.

We want to thank our institution ‘Maulana Azad National Institute of Technology, Bhopal’ for providing us with resources and opportunity for conducting this research. Each author contributed equally to the submitted article.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.