1 Introduction

Kolkata, a flood-prone city, faces increasing urban flooding due to a combination of topographical challenges, rapid urbanization, and climate change [1]. The city's geographical location makes it highly vulnerable to tropical cyclones, storms, and tidal surges. Its natural slope, declining from the Hooghly River towards the east, further hampers drainage efficiency. Rapid urban development and severe pollution, have impacted the natural flood buffers—lakes, canals, wetlands (specifically, the East Kolkata Wetlands, or EKW), ponds, and parks, exacerbating the risk [2]. Additionally, the aging stormwater drainage system is frequently overwhelmed by the effects of climate change [3], with grey infrastructure solutions proving inadequate during periods of intense rainfall. Despite the pressing need, there is a lack of localized studies evaluating the feasibility and socio-economic impact of Nature-based Solutions (NBS) in such challenging urban contexts, making this research an important contribution to filling that gap.

The study was conducted at two sites in Kolkata—Maheshtala and Rabindra Sarovar Lake area—where contingent valuation (CV) surveys were carried out to assess the willingness of people to adopt NBS in these areas. A ‘‘Room for River’’ program was proposed for Maheshtala, while restoration of the wetland ecosystem was suggested for Rabindra Sarovar Lake. The rationale for the ‘‘Room for River’’ initiative in Maheshtala lies in its network of canals directly connected to the Hooghly River. These canals historically mitigated waterlogging and ensured free flow of water during high tides. However, encroachment, extreme siltation, waste dumping, and extensive concretization in Maheshtala have drastically reduced the canals'effectiveness. Simply restoring the canals to their original state through this program could significantly improve water management in the area. Rabindra Sarovar, on the other hand, inherently serves as an NBS. Restoration, conservation, and preservation of the lake and park area would enhance its capacity to act as a flood buffer. Currently, the lake suffers from siltation and pollution, while the park area faces ongoing concretization. Addressing these issues would not only improve flood resilience but also bolster the ecological and recreational value of the site. Furthermore, Benefit Transfer of the lake ecosystem to the wetlands of Maheshtala was also proposed.

The Room for the River Program was first introduced by the Netherlands in 1995 with the goal of improving flood safety and spatial quality along the Rhine and Meuse Rivers. This comprehensive approach involved various projects such as deepening the main river channel, creating water storage areas, relocating dikes, creating secondary channels, lowering floodplains, lowering groynes, and reinforcing dikes to create more space for rivers [4]. However, a study by De Groot and De Groot [5] found that residents living by Dutch rivers perceived these measures as unnatural interventions in contrast to the environment. In response, this study proposes a modified Room for River program using NBS, focusing on the restoration and conservation of canals connected to rivers. This approach aims to enhance flood mitigation, water quality, and groundwater recharge in a more sustainable and eco-centric manner.

Wetlands have played a crucial role in human civilization. They are recognized for their ecological functions, including flood mitigation. Acting as green infrastructure, wetlands and floodplains store and slow floodwaters, reducing downstream flood risks [6]. However, human activities have led to significant wetland losses and river-floodplain disconnection, exacerbating flooding risks [7]. Quantifying the total economic value of wetlands'flood mitigation services can influence regional planning decisions. The economic approach to valuing environmental changes focuses on measuring people's willingness to pay (WTP) for environmental gains or willingness to accept (WTA) compensation for losses. The total economic value (TEV) includes both use and non-use values [8]. Studies like Ndebele and Forgie [9] have used contingent valuation methods (CVM) to estimate the economic benefits of wetland restoration programs. These valuations help assess the economic efficiency of public policies aimed at restoration. Benefits transfer, applying monetary values from prior studies to new contexts, is another valuable tool in to calculate TEV [8].

Batzias, Zoupanidou, Kopsidas, and Siontorou [10] conducted a study using Contingent Valuation Method (CVM) to assess citizen’s WTP for lake restoration. They found that WTP was higher among citizens who were influenced by factors such as proximity to the lake, income, and perception of the lake's condition. Sherif [11] conducted a study on park conservation and identified significant WTP values. The study revealed that factors such as marital status, income, and livestock ownership had a positive influence on WTP, while age, distance from the park, and initial bid amount had negative associations. Both studies highlight the importance communities place on environmental resources and their willingness to.

We undertook Contingent Valuation (CV) studies in Maheshtala to gain insights into willingness to pay (WTP) and in Rabindra Sarovar Lake to evaluate willingness to contribute (WTC). By applying a Bivariate Probit model for our analysis, we were able to effectively assess these dimensions. Following this, we conducted a comprehensive Cost–Benefit Analysis (CBA), which allowed us to quantify the benefits derived from the WTP and WTC values. Additionally, we gathered cost information through stakeholder interactions and leveraged various verified sources from both governmental and non-governmental entities. This approach ensured a well-rounded understanding of the economic implications involved.

2 Data and methodology

2.1 Description of study site 1: Maheshtala

Maheshtala is a municipality area located at the south of Kolkata, in south 24 Paraganas district, and is one of the largest municipalities in West Bengal, with an area of 44.18 sq. km. [12].The population of as per census 2011 is 449,423, and the total number of households in is 89,596. The municipality is characterized by significant urban development, covering nearly 50% of its area, with vegetation and water bodies occupying the remaining space [13] (Refer Fig. 1). Rapid urbanization has led to severe waterlogging issues, exacerbated by the area's low elevation of 2–6 m above sea level. The municipality features 10 irrigation canals draining into the Hooghly River, serving as the main sewerage system [14]. However, these canals face encroachment, waste accumulation, and lack of maintenance, reducing their water-carrying capacity. Additionally, houses 700 bleaching and dyeing units, whose effluents contaminate both surface and groundwater. Consequently, the canals'water quality is very poor, with high levels of pollutants including Total Dissolved Solids (TDS), Biochemical Oxygen Demand (BOD), and heavy metals, making it unsuitable for various uses year-round [15]. The area's conditions make it ideal for implementing a Room for River program to restore and conserve the canals, addressing both flood management and water quality issues.

Fig.1
Fig.1
Full size image

Source: Google Earth Pro

Map of Maheshtala.

2.2 Description of study site 2: Rabindra Sarovar Lake

Rabindra Sarovar Lake, located in southern Kolkata, is a significant urban water body covering 38% of a 73-acre area [16]. Figure 2 shows the study area. It serves as a natural CO2 sink and hosts diverse flora and fauna, including numerous bird, fish, insect, and plant species [17]. The lake offers various recreational facilities and is popular for morning walks. Despite being a successful example of urban ecological restoration, the lake is currently facing ecological stress due to poor water quality, declining water levels, and unnecessary concretization projects. The lake's dissolved oxygen (DO) profile indicates potential hypoxic conditions, while its high pH levels are considered unacceptable for a natural water body [18]. There is an urgent need for sustained maintenance of water quality and levels, controlled development, and extensive tree planting. The lake's ecosystem restoration presents potential for public–private partnerships (PPP) [19], given the presence of commercial clubs and high-end real estate in the vicinity. Additionally, there's potential for applying the Benefit Transfer method to the wetlands [20, 21] in Maheshtala.

Fig.2
Fig.2
Full size image

Source: Google Earth Pro

Map of Rabindra Sarovar Lake Area.

2.3 Conceptual framework

A primary survey based CVM was conducted in the two earlier mentioned study sites, respectively. In the CVM survey, the respondent’s WTP for study site 1, and WTC for study site 2, for three proposed improvement scenarios from the status quo was determined [20]. The proposed improvement scenarios are in the context of restoration and preservation of the canals (room for river program) of Study site 1, and restoration and preservation of the lake and park area of study site 2. For each improvement scenario, the method of elicitation of the WTP of the respondent was a double-bounded dichotomous choice study (DC) [22, 23]. The payment vehicle for study site 1 is an increment in the compulsory payment of the annual municipality tax, on yearly basis, for the next 25 years, and the payment vehicle for study site 2 is voluntary contribution by the visitors on yearly basis, for the next five years.

Three hypothetical scenarios were developed for both study sites, focusing on a 25 year program for the restoration and conservation of the canals in Maheshtala and Rabindra Sarovar Lake. A 25 year timeframe was chosen to assess the optimal effectiveness of NBS under given constraints, as such initiatives typically require long-term implementation to yield measurable outcomes. The three improvement scenarios are, scenario A, or small improvement, scenario B, or medium improvement, and scenario C, or large improvement. The attributes defining these scenarios mainly revolve around the idea of mitigation of urban flooding. Study site wise improvement scenarios and their attributes are given in the following Table 1 and 2.

Table 1 Attributes and CV scenarios of study site 1: Maheshtala
Full size table
Table 2 Attributes and CV scenarios of study site 2: Rabindra Sarovar Lake
Full size table

In the above given hypothetical scenarios, describing the attributes, scenario A defines a small improvement situation where planting of trees and their maintenance is considered, which will help in reduction of flow and storage of extra water. Scenario B defines a medium improvement situation where cleaning and dredging of the ten canals is considered every alternate year, which will help in reduction, storage and discharge of the extra water. Finally, scenario C defines a large improvement situation where installation of Rainwater Harvesting (RWH) in the households of Maheshtala is considered, which will help in storage and discharge of the extra water. Here extra water is defined by the quantity of stormwater that exceeds the present carrying capacity of the canals.

In the above given hypothetical scenarios, describing the attributes, scenario A defines a small improvement situation where planting of trees and their maintenance is considered, which will help in improvement of flora, fauna and water level. Scenario B defines a medium improvement situation where cleaning and dredging of the lake is considered every year, which will help in improvement in flora, fauna, water level, and help in storage of extra water as well. Finally, scenario C defines a large improvement situation where installation of RWH in the residential space around the lake (ward 87), is considered, which will help in storage and discharge of the extra water, and increase and keep the water level consistent in the lakes. Here extra water is defined by the quantity of stormwater that exceeds the present carrying capacity of the lake and the drainage system of ward 87 of Kolkata.

The data collected from both study sites is predominantly qualitative, particularly variables such as environmental awareness, perception, and opinions, which were measured on a Likert scale. Using dummy variables for these qualitative factors could lead to the"curse of dimensionality"and multicollinearity. To address this, Principal Component Analysis (PCA) was employed. To analyse the willingness to shift from Scenario A to B and B to C, data on WTP or WTC for Scenarios B and C was modelled using a Bivariate Probit Model for both study sites. This approach enables the joint estimation of the probabilities of willingness to pay for Scenarios B and C, accounting for the potential interdependence between these two shifts. It also effectively models the relationship between the two dependent variables, capturing shared unobserved factors. For this analysis, the dependent variables (willingness to pay for Scenarios B and C) were treated as binary variables. Lastly, an ex-ante CBA was conducted, encompassing all three improvement scenarios for each study site over the 25-year timeframe.

2.4 Primary survey: sampling strategy and data collection

Primary survey of both the study sites were conducted in the month of March and April of 2024. In study site 1, 101 face-to-face household interviews were conducted, where a multi-stage sampling strategy was employed, starting with the selection of specific wards that contained at least one of the 10 irrigational canals discussed earlier in Sect. 2.1. Within these wards, areas were further narrowed down to those located within approximately 3 kms of the Hooghly River, identified as flood plains prone to severe waterlogging. Data on waterlogged areas was obtained from the Maheshtala Municipality. Finally, households in these areas were selected using simple random sampling. The survey covered the following wards: 33, 32, 31, 28, 20, 19, and 7. On the other hand, in study site 2100 face-to-face interviews were conducted, with most respondents being visitors and around 10 from the households of Ward 87. Initially, the survey aimed for an equal split between visitors and Ward 87 residents, based on pilot survey findings from February–March 2024, which indicated that most visitors were not local residents. However, during the primary survey, it was observed that a majority of visitors were actually from Ward 87, leading to a reduction in household interviews. Visitors were selected through convenience sampling, while Ward 87 residents were chosen using simple random sampling. The simple random sampling was conducted using the lottery method.

2.5 Principal component analysis

PCA is used to extract core elements from large datasets, reducing dimensionality while preserving maximum variance [24]. For both the study sites, environmental variables, like, problems, awareness, and interest, is analysed as PCA 1; while future perceptions regarding the mitigation measures is analysed as PCA 2. Environmental problems and awareness variables have been captured on a Likert scale and are ordinal and categorical in nature, while the interest variable is continuous ordinal in nature. Refer to appendix for the description of the variables. After the PCA with varimax rotation was conducted, the most significant factors were identified, only those principal components (PCs) whose eigenvalues are equal to or exceed the value of 1, are used for further analysis [25]. For study site 1, three components from PCA 1 and two components from PCA 2 were selected, explaining cumulative variances of 62% and 61%, respectively. For study site 2, two components each from PCA 1 and PCA 2 were selected, explaining cumulative variances of 56% and 75%, respectively.

2.6 Empirical model

Balakumar and Das [26] in their CVM study uses a bivariate probit regression to examine two distinct outcomes—the household's willingness to move from Scenario B to Scenario C, and the household's willingness to move from Scenario C to Scenario D of their study’s CV scenarios. This method allows for the simultaneous modelling. The Bivariate Probit model is particularly useful in this context because it recognizes that unobserved factors affecting the decision to move in one scenario may also impact the decision to move in the other scenario. Since both the study sites in this study employed a double-bounded DC study, a bivariate probit model was used by examining the respondent’s willingness to move from scenario A to scenario B, and from scenario B to scenario C.

Table 3 gives the list of variables and their description defining the bivariate probit model of study site 1. The list of variables of study site 2 is given in the appendix. They are similar to and less compared to study site 1. The sources of the variables include the literature referenced in the preceding sections, the current section, and insights gained from the pilot survey.

Table 3 List of variables and description of study site 1
Full size table

2.7 Cost–benefit analysis

CBA is a technique used to assess the economic feasibility and societal value of projects [27]. It considers all costs and benefits to society, assigning monetary values to consequences. This study has conducted an ex-ante CBA, since it helps determine if a project would provide positive net social benefits, aiding in resource allocation decisions. The costs of the proposed NBS projects are given in Table 4 for each site along with the respective sources.

Table 4 Costs of the NBS projects for each study site and their sources
Full size table

The benefits are compiled using the survey data on the WTP and WTC for the CV scenarios of each study site respectively. We calculate the average WTP and WTC for each study site respectively and multiply it with the total number of households willing to pay in study site 1, and total number of households and visitors willing to contribute in study site 2, respectively, for each CV scenario. The sum of multiplied values for each CV scenario gives us the total benefits. The results of the calculation of the benefits are given in Table 5. The details of the calculations of total number of households and visitors willing to pay and contribute from each study site respectively, for each CV scenario, is given in the appendix.

Table 5 Calculation of the benefits
Full size table

The benefits of study site 1 derived above, is proposed to be extracted from the households of Maheshtala for the entire duration of the proposed project, that is, 25 years, from 2025 to 2050, since its payment vehicle is an increment in municipality tax. However, the benefits of study site 2 derived above is proposed to be extracted from the households of ward 87, and from the visitors of Rabindra Sarovar Lake for the duration of five years of the project, that is from 2025 to 2029, since the payment vehicle is voluntary contribution.

Although revenue stream is not included in a social CBA, it is proposed in this research, as part of the NBS project implementation, that the revenue generated by KMDA from study site 2 can be exclusively used for the restoration and conservation of the Lake ecosystem. Thus, only for study site 2, the revenue generated is also included in the calculations of the total benefits, and this revenue is proposed to be extracted for the purpose of the fulfilment of the NBS project solely for the entire duration of the project, that is, 25 years, 2025 to 2050. Table 6 gives us the revenue streams considered and the source of the data.

Table 6 Revenue streams for study site 2
Full size table

3 Results and analysis

3.1 Social, economic and demographic characteristics of the sampled respondents of study site 1: Maheshtala

The Maheshtala dataset (101 respondents) shows an average family size of 3–4 members (range: 1–10), with 2 dependents on average (range: 0–7). The average age is 51 years (range: 17–86) with a standard deviation of 17, indicating a diverse demographic profile of the sample. It is to be noted that the number of observations for the variable “Age” in the dataset is 94, less than the total sample size of 101, as certain interviewees declined to disclose their age. The statistics are represented in Table 7 below.

Table 7 Descriptive Statistics of study site 1
Full size table

Further, snippets of the provided results in Table 8 outline proportions, standard errors, and confidence intervals for various demographic and socioeconomic variables. The sample is predominantly male (63.9%). Occupations are mainly private service (19.6%) and business (22.7%). Education levels are mostly graduate (36.1%) and 12 th standard (22.7%). Household expenditure is primarily Rs.15000–20000 (19.5%).

Table 8 Proportion estimation of study site 1
Full size table

The data collected on frequency of visit to Hooghly River reflects that most of the respondents visit Hooghly River yearly or once a year (32%), followed by monthly or once a month (23%). Fewer participants engage in the activity daily (16%) or weekly (15%). The least common frequency is less than once a year, with 14 respondents. This suggests that while the respondents most commonly visit on an annual basis, a significant number of people also participate on a monthly or more frequent basis.

3.2 Environmental awareness characteristics of the sampled respondents of study site 1

Regarding environmental problems, respondents consider frequent flooding, drain clogging, and water body degradation as severe issues (55%, 71%, and 53%). Surprisingly, solid waste management is not seen as important (68%), despite awareness of canal dumping. This is because they are mostly concerned whether the solid waste is timely taken out of their individual households, which are evidently done by the municipality. Sanitation is viewed as a moderate issue (39 respondents). Municipal water supply and pollution are not considered important issues by the majority (58% and 36% respectively).

The survey results regarding respondents'opinions on the condition of the canals highlight several areas for improvement. A significant portion of participants rated the current condition as"Very Bad"(36%) or"Bad"(24%), indicating a widespread perception of the canals being in need of attention. Notably, no respondents selected"Very Good,"suggesting that there is ample opportunity for enhancement. Overall, these findings indicate a public desire for better maintenance and cleanliness in the canals, signaling a clear pathway for constructive improvements moving forward.

The"Interest in Environmental Issues of the Canals"rating scale reflects respondents'interest in the environmental problems of the canals, where 1 indicates"not interested"and 5 indicates"very interested."The results show varying levels of interest: 33% of respondents rated themselves as"very interested"(5), 27% at rating 4, 20% at rating 3, 13% at rating 2, and 7% as"not interested"(1). This indicates that a majority of respondents are highly interested in the environmental issues concerning the canals.

The survey on the perceived future benefits of the proposed Nature-Based Solution (NBS) – the Room for River program – reveals varied opinions across three key areas. Flood reduction is viewed as the most significant benefit, with 80% of respondents anticipating a positive impact, likely due to the wide canals with sufficient depth providing a large carrying capacity. Water quality improvement is the second most recognized benefit, with 58% of respondents perceiving positive changes. However, opinions on groundwater recharge are more divided: only 26% see benefits, 39% consider it unrelated, and 26% remain unsure, likely because groundwater recharge is less immediately visible.

Regarding potential negative impacts, as seen in the result of the survey, a substantial majority (79%) believe the project will not have adverse effects. However, 15% of respondents express concerns, primarily about possible land displacement for residents due to canal widening. This minority view, while not predominant, is significant enough to warrant consideration in project planning.

3.3 Social, economic and demographic characteristics of the sampled respondents of study site 2

The survey of 100 respondents from study site 2 reveals diverse family structures, with average size of 3–4 members and 1–2 dependents. Ages range from 19 to 85, averaging early 50 s, indicating a broad demographic representation. It is to be noted that the number of observations for the variables “Family size”, “No. of dependent members”, and “Age” in the dataset is 97, 97, and 98 respectively, less than the total sample size of 100, as certain interviewees declined to disclose the information. This is represented in Table 9.

Table 9 Descriptive Statistics of study site 2
Full size table

Snippets of the provided results in Table 10 outlines a male majority (67.7%), diverse occupations with private service leading (25%), high education levels (87.5% graduates and postgraduates), and varied household expenditures, reflecting a broad socioeconomic spectrum.

Table 10 Proportion Estimation of study site 2
Full size table

The data on visit frequency to Rabindra Sarovar Lake shows that the largest proportion of respondents visit daily (39%), followed by weekly visitors (25%). A smaller group visits once or twice a year (17%), while 10% visit monthly. The least frequent visitors, those visiting less than once a year, make up 9%. This suggests that most respondents have a high level of awareness regarding the lake's environmental issues.

3.4 Environmental awareness characteristics of the sampled respondents of study site 2

Regarding environmental problems, the surveys reveals diverse levels of concern for environmental issues from the respondents. Decreasing water levels (55%), sanitation (44%), and lake/park degradation (31%) are perceived as the most severe problems. Solid waste management (31%), water pollution (30%), and further lake/park degradation (36%) are seen as moderately severe. There's notable uncertainty regarding wastewater dumping (27% don't know) and water pollution (18% don't know). Overall, decreasing water levels and environmental degradation emerge as top priorities, with solid waste management and pollution as secondary concerns.

Survey results regarding the status of the lake and park area show diverse opinions on the lake's current condition. 35% responded"Don't Know,"indicating significant uncertainty. 27% rated it"Fair,"suggesting average conditions. Negative ratings ("Very Bad"and"Bad") totalled 27%, while positive ratings ("Good"and"Very Good") accounted for 17%. The data highlights a tendency towards neutral or negative perceptions, reflecting concerns about the lake's water quality, water level, and biodiversity.

Interest in the environmental issues surrounding Rabindra Sarovar Lake highlights a strong commitment among respondents, as demonstrated by a rating system with five response options ranging from 1 (not interested) to 5 (very interested). A significant 33% of participants rated their interest as a 5, showing a high level of engagement. Furthermore, 27% rated their interest as 4, indicating considerable concern for these issues. Rating 3 received 20% of the responses, reflecting a moderate level of interest, while 13% of respondents chose rating 2. The smallest group, at 7%, rated their interest as 1, meaning they were not interested at all. Overall, these results suggest a robust enthusiasm for the environmental issues associated with Rabindra Sarovar Lake, providing a strong foundation for further awareness and action.

The survey on the perceived future benefits of restoration and conservation projects for Rabindra Sarovar Lake reveals positive expectations across three key areas. Improvement in water quality is seen as the most significant benefit, with 77% of respondents believing the projects will have a positive impact. Groundwater recharge is the second most recognized benefit, with 64% perceiving positive changes. Flood reduction is also viewed positively, with 55% of respondents expecting improvements. Notably, very few respondents (4–14%) consider these projects unrelated to the three assessed areas, indicating a consensus on their relevance. However, some uncertainty exists, particularly regarding flood reduction benefits.

Regarding potential negative impacts, a substantial majority (82%) believe the projects will not have adverse effects, while only 8% anticipate negative outcomes. This overwhelmingly positive perception suggests strong community support for the lake ecosystem restoration and conservation efforts.

3.5 Principal component analysis

Before carrying out the PCA, Kaiser–Meyer–Olkin (KMO) test and Bartlett’s test of sphericity were conducted. KMO checks for the partial correlation values among the considered variables for PCA. If the values come significantly high, that is between 0.5 and 1, it indicates that PCA would be useful. Bartlett’s test, on the other hand, examines the hypothesis that whether the underlying correlation matrix is an identity matrix, which would suggest that the variables are uncorrelated. Thus, a p-value at significance level below 0.05 indicates that PCA is appropriate for the data, since it tells us that the variables are correlated [31]. The results of KMO and Barett’s test for each variable, are provided in Tables 11, 12, 13 and 14 below. It can be observed that other than PCA 2 of study site 1, PCA 1 of study site 1 and PCA 1 and 2 of study site 2 have KMO values within the acceptable range of 0.5–1, indicating sufficient sampling adequacy. Although the KMO value for PCA 2 of study site 1 is 0.47 (slightly below the threshold) it was included because Bartlett’s test yielded a p-value of 0.006, suggesting strong correlations among the variables. Moreover, PCA 1 of study site 1 and PCA 1 of study site 2, as well as PCA 2 of study site 1 and PCA 2 of study site 2, respectively, capture similar environmental parameters, such as perception, responsiveness, and interest. Thus, to ensure consistency in PCA variables for meaningful cross-site comparisons, it was necessary to include PCA 2.

Table 11 Results of Principal Component Analysis (Varimax Rotated Component Matrix) (PCA 1) of Study site 1
Full size table
Table 12 Results of Principal Component Analysis (Varimax Rotated Component Matrix) (PCA 2) of Study site 1
Full size table
Table 13 Results of principal component analysis (Varimax rotated component matrix) (PCA 1) of Study site 2
Full size table
Table 14 Results of Principal Component Analysis (Varimax Rotated Component Matrix) (PCA 2) of Study site 2
Full size table

In the subsequent regression analysis, the regression models incorporate scores of each PC, that is, loadings (“comp”) described in the varimax rotated component matrix Tables 11, 12, 13 and 14. The highest loading in each column is represented in bold. These PCs are named according to the variable with the highest loading on each PC. For example, in Table 11, variable “clogging of drains” of the environmental problem parameter has the highest loading of 0.464 on PC 1 (“comp1”). This naming convention helps in identifying the primary characteristic represented by each PC, making it easier to interpret their impact within the regression model. By using the component scores, the regression model effectively summarizes and utilizes the essential features of the original variables, thus simplifying the data while retaining its most significant information. This approach enhances the model's explanatory power and helps in uncovering the relationships between the principal components and the dependent variable.

3.6 Results of the empirical models

Before proceeding, it is important to note that in this model, a dependent variable value of 0 indicates the respondent is willing to shift scenarios ("Yes"), while a value of 1 indicates they are not willing to shift ("No"). A negative coefficient for an independent variable means that as the variable increases, the likelihood of responding"No"(1) decreases, thus increasing the likelihood of responding"Yes"(0). Conversely, a positive coefficient suggests that an increase in the independent variable leads to a higher likelihood of saying"No"(1), decreasing the likelihood of a"Yes"(0) response.

Numerous iterations of the Bi-Variate Probit Model were carried out for both study sites using different combinations of variables to identify the best-fitting model. The models with the lowest Akaike Information Criterion (AIC) values were selected as the best versions. AIC is a measure used in statistical models to compare different models and assess their relative quality. It is particularly useful for model selection, helping to identify the model that best explains the data with the minimum number of parameters [32]. The empirical models displayed below were the ones with the lowest AIC, across all the models that were tested, and hence are the ones that are chosen to interpret the data.

3.6.1 Results of the empirical model of study site 1

Table 15 provides the result of the Bivariate Probit model for study site 1, Maheshtala. The dependent variables are Willingness to shift from scenario A to scenario B (wtp_sb) and Willingness to shift from scenario B to scenario C (wtp_sc).

Table 15 Result of the Bivariate Probit model for study site 1
Full size table

The analysis for the dependent variable (wb_sb) shows significant relationships between willingness to pay (WTP) for Scenario B and several independent variables. For the variable sex, the negative coefficient (−1.841) and significance at the 0.01 level (p = 0.004) indicate that being female is associated with a higher likelihood of saying"Yes"to WTP for Scenario B, compared to males (refer Table 7). The variable occu shows a positive coefficient (0.342) and a significant p-value at the 0.01 level (p = 0.001), suggesting that as occupational status increases (refer Table 7), there is a higher likelihood of saying"No"to WTP for Scenario B. Additionally, freq demonstrates a negative coefficient (− 0.346) and is significant at the 0.05 level (p = 0.047), which implies that individuals with more frequent visits are more likely to say"Yes"to WTP for Scenario B.

The variable exp_h, representing household expenditure, has a weak positive association (coefficient of 0) with a “No” response to WTP for Scenario B, with marginal significance (p = 0.070). Another factor, pc1 (clogging of drains), which is the highest-loading environmental issue in the first principal component, shows marginal significance (p = 0.096) and a negative coefficient (− 0.269). This suggests that as respondents perceive the severity of this problem as higher, they are more likely to say"Yes"to WTP for Scenario B. The future environmental variable c2,"the restoration and conservation project will have negative impacts,"loaded onto the second principal component, shows a significant positive association with the dependent variable (coefficient = 0.749, p = 0.03). Respondents who perceive the project as having potential negative impacts (responses from"Maybe"to"Not related") are more likely to say"No"to WTP for Scenario B. Other variables, including age, education, waterlog, pc1, pc3, c1, and c2, did not exhibit significant effects on (wb_sb).

Similarly, for the dependent variable (wb_sc), certain variables show significant relationships with WTP for Scenario C. sex has a negative coefficient (− 1.355) and is significant at the 0.05 level (p = 0.019), suggesting that females are more likely to say"Yes"to WTP for Scenario C. For occu, the positive coefficient (0.274) and significance at the 0.01 level (p = 0.004) indicate that higher occupational status is linked to a higher likelihood of saying"No"to WTP for Scenario C. Additionally, freq has a negative coefficient (− 0.353) and is marginally significant at the 0.10 level (p = 0.067), implying that more frequent visitors are more inclined to say"Yes"to WTP for Scenario C. The second principal component, c2, associated with future environmental perceptions, also has a significant positive relationship (coefficient = 0.622, p = 0.018) with a"No"response to WTP for Scenario C, reinforcing that perceived negative impacts of the project lead to lower WTP. Other variables, including age, education, waterlog, exp_h, exp_p, pc1, pc2, pc3, and c1, do not show significant effects on (wb_sc).

3.6.2 Results of the empirical model of study site 2

Table 16 provides the result of the Bivariate Probit model for study site 2, Rabindra Sarovar Lake. The dependent variables are Willingness to shift from scenario A to scenario B (wtp_sb) and Willingness to shift from scenario B to scenario C (wtp_sc).

Table 16 Result of the Bivariate Probit model for study site 2
Full size table

The results reveal significant associations between the independent variables and the dependent variable for (wb_sb). For the variable educ, a negative coefficient of −0.226, significant at the 0.05 level (p = 0.024), suggests that higher education is associated with a greater likelihood of saying"Yes"to contributing for Scenario B. For freq, the positive coefficient (0.406) and 0.05 level significance (p = 0.013) indicate that more frequent visits are associated with a higher likelihood of saying"No"to contributing for Scenario B. Additionally, pc1, representing the environmental problem variable"water pollution"with the highest loading on the first principal component, has a positive coefficient of 0.386 and is highly significant (p = 0.009). This suggests that respondents perceiving this problem as more severe are more likely to say"Yes"to willingness to pay for Scenario B.

Similarly, significant associations were found between independent variables and the dependent variable (wb_sc). For the future environmental perception variable c2 ("the conservation project has negative impacts"), the coefficient is −0.395, marginally significant at the 0.10 level (p = 0.063). This suggests that respondents who perceive the project as not having negative impacts are more likely to say"Yes"to paying for Scenario C. The variable educ also shows a significant effect, with a coefficient of −0.322 and high significance (p = 0.007), indicating that higher education is associated with a higher likelihood of saying"Yes"to contribute for Scenario C.

For freq, the coefficient is positive (0.75) and highly significant (p = 0.001), implying that frequent visits correlate with a greater likelihood of saying"No"to contribute for Scenario C. pc1, with a coefficient of 0.534 and high significance (p = 0.004), suggests that higher values on the first principal component are associated with a higher likelihood of saying"No"to contribute for Scenario C. Lastly, the constant term in the model is marginally significant (p = 0.088), indicating that other factors not captured in the model may also influence willingness to contribute for Scenario C.

3.7 Cost benefit analysis

The viability of the proposed NBS, that is, room for river program (restoration and conservation of canals) in study site 1 and restoration and conservation of lake and park area of study site 2, was checked with the use of ex-ante social CBA. The costs, benefits, and revenues were adjusted for inflation using World Bank data from 2003 to 2024, while a 4% inflation rate, based on the RBI's 2–6% tolerance band, was assumed for the 2025–2050 project period. Furthermore, two discount rates for the two study sites have been applied respectively to future costs and benefits to calculate their present values (PV), accounting for opportunity costs and time preferences, with the net present value (NPV) determined as the difference between PV(Benefit) and PV(Cost). It was found that both projects were coming to be viable.

In Study Site 1, the total benefits exceed the total costs, and the NPV is positive, Rs. 22,427,989.10, when social discount rate of 2% is considered, denoting the viability of the projects. The internal rate of return (IRR) is calculated to be 3%. A low social discount rate of 2% suggests that the project places a high value on future benefits and costs. This means that the future impacts of the project are almost as important as the present impacts, indicating a long-term perspective, prioritizing sustainability for future generations. The table with the detailed calculations for 25 years is provided in the appendix.

In Study Site 2, the total benefits (benefits and revenues are considered together) exceed the total costs, and the NPV is positive, Rs. 3,291,292.853, when social discount rate of 19% is considered, denoting the viability of the projects. The internal rate of return (IRR) is calculated to be 18%. A high social discount rate of 19% implies that the project heavily prioritizes immediate benefits over future ones. The future impacts are heavily discounted, meaning they are considered much less important than the present impacts, indicating a short-term perspective, where immediate gains are given more weight. The table with the detailed calculations for 25 years is provided in the appendix.

4 Discussion

Climate change is exacerbating urban flooding issues through more extreme weather events and rapid urbanization. Traditional grey infrastructure is proving inadequate and expensive to manage these growing challenges. In response, NBS are emerging as sustainable and cost-effective alternatives. These solutions work with natural ecosystems to manage water through absorption and controlled dispersal, mimicking nature's own water management processes. The successful implementation of NBS heavily relies on public acceptance and involvement. It's crucial to raise awareness about climate change impacts and the limitations of grey infrastructure. Community support is essential, as NBS often require space, initial investment, and ongoing maintenance. Public involvement promotes the long-term sustainability of these projects, as people who feel ownership and see tangible benefits are more likely to take an active role in stewardship. The study conducted a contingent valuation survey to spread knowledge about urban flooding challenges and how NBS can mitigate these issues sustainably. Strong public support and advocacy are vital for creating the political will to prioritize and fund NBS as part of climate adaptation strategies. This includes vocal advocacy from citizen groups, participation in public consultations, and voting for leaders who champion green solutions.

An astounding 71% from study site 1, that is Maheshtala were willing to pay, and 60% from study site 2, that is Rabindra Sarovar Lake were willing to contribute for the NBS projects proposed for each of these sites. This shows the awareness people have regarding the eminent threats of annual urban flooding, and their enthusiasm to mitigate the issue. This also shows the awareness and knowledge among the people regarding the efficiency and sustainability of NBS, like, restoration and conservation of canals, wetlands, lakes, planting of trees, RWH, etc. at mitigating urban flooding. The Cost Benefit Analysis shows the viability of the proposed NBS project for both the study sites, as the NPV for study site 1 is calculated to be 22.4 million INR at a social discount rate of 2%, and the NPV for study site 2 is calculated to be 3.3 million INR at a social discount rate of 19%. The internal rate of return for study site 1 is 3%, while for study site 2 is 18%. A low social discount rate of 2% suggests that the project places a high value on future benefits and costs. This means that the future impacts of the project are almost as important as the present impacts. While a high social discount rate of 19% implies that the project heavily prioritizes immediate benefits over future gains. A reason for this could be that study site 2 being a national lake and being in the city proper, is in quite good condition, compared to alarming condition of study site 1, which is a municipality area, adjacent to the city of Kolkata. Moreover, study site 1, which is a municipality, compared to study site 2, which is a ward of KMC, is a huge area of study. This significantly influences the internal rate of return, NPV, and hence derivation of the social discount rate. It is also to be noted that the result of the CBA for the two study sites might have been influenced by the timeline of flow of benefits. While benefits for study site 1 was considered for the entire timeline of the project, the benefits for study site 2 was considered only for first 5 years of the entire timeline of the project. However, it is to be noted that for study site 2, the cash inflow of revenue is considered for the entire timeline of the project, along with the benefits, which is not the case for study site 1.

5 Conclusion

This study demonstrates the critical importance of NBS in addressing the escalating urban flooding challenges in Kolkata. With climate change driving more frequent and severe extreme rainfall events, including 100-year storms occurring at shorter intervals, cities like Kolkata face increasing risks of inundation, loss of infrastructure, and disruption to daily life. The findings from Maheshtala and Rabindra Sarovar Lake reveal that NBS, such as canal restoration, tree plantations, and rainwater harvesting, offer sustainable and cost-effective solutions to mitigate these challenges while securing significant community support. Importantly, respondents from Rabindra Sarovar Lake also recognize the area’s role as an existing NBS that mitigates flooding in its vicinity, further emphasizing the community’s understanding of its ecological and functional significance.

The WTP in Maheshtala and the WTC at Rabindra Sarovar Lake reflect an encouraging level of public awareness about the urgent need for sustainable flood mitigation strategies. The CBA further emphasized the viability of NBS projects, with Maheshtala showing a long-term positive NPV of 22.4 million INR at a 2% discount rate, and Rabindra Sarovar Lake demonstrating immediate gains with an positive NPV of 3.3 million INR at a 19% discount rate. These results highlight the potential of NBS to offer a balanced approach to urban resilience, combining environmental conservation with economic sustainability.

However, the study also recognizes several challenges, including small sample size, limited survey participation, particularly among women, and difficulties in obtaining accurate cost estimates due to the nascent implementation of NBS projects in Kolkata. Addressing these gaps in future research will help refine strategies and provide a stronger foundation for decision-making.

Given the intensifying impacts of climate change, traditional grey infrastructure alone cannot provide adequate protection against increasingly severe weather events. NBS, which work in harmony with natural processes, present a sustainable alternative by mitigating flood risks, enhancing groundwater recharge, and improving water quality. To make NBS an integral part of urban planning, a multifaceted approach is necessary. Community engagement through awareness campaigns, workshops, and volunteer programs can foster shared responsibility and ownership. Public–private partnerships can provide the necessary funding and management capacity for large-scale projects, while government policies should incorporate clear climate resilience goals and financial incentives, such as tax benefits for adopting rainwater harvesting systems. Additionally, urban planning must prioritize the protection of natural flood buffers like wetlands, floodplains, and riparian zones while avoiding excessive concretization. Integrating NBS into land use planning, stormwater management, and resilience strategies, as well as evaluating them during environmental impact assessments, will reinforce their adoption. Regional collaboration among municipalities and bioregions will also be crucial for knowledge sharing and coordinated implementation.

Ultimately, this study reinforces the idea that NBS are not just an option but a necessity for cities like Kolkata. As climate change accelerates, adopting NBS offers a pathway to transform urban landscapes into greener, more adaptive, and resilient spaces. This transition not only addresses the immediate threats of flooding but also secures a sustainable future for generations to come, ensuring that cities can withstand the growing challenges posed by a changing climate.