How local governments avoid floodplain development through consistent implementation of routine municipal ordinances, plans, and programs

Abstract

Avoiding floodplain development is critical for limiting flood damage, yet there is little empirical evidence of how local governments effectively avoid floodplain development. We conduct a mixed-methods study to explain how local floodplain management influences floodplain development in New Jersey, a state with high development pressure and flood risk. We find that 85% of towns developed relatively little in the floodplain from 2001 to 2019, and they achieved this with commonplace land use management tools and modest levels of local government capacity. One hundred twenty-six New Jersey towns put none of their new housing in the floodplain 2001–2019. Our findings run counter to common reports of rampant floodplain development requiring legal innovation and capacity-building campaigns and suggest alternative approaches for promoting floodplain avoidance. We find multiple paths to floodplain avoidance, weak support for previously identified drivers, and strong evidence that the keys to avoidance include having a few high-quality tools that are well-implemented, requiring consistency, coordination, and commitment of local officials. The multiple paths and importance of maximum, rather than average, quality might explain the mixed evidence in prior research connecting floodplain management actions and development outcomes. A lack of attention to towns that limit floodplain development impedes our ability to learn from and disseminate their successes. Contrary to our expectations, we show that floodplain avoidance can be and is achieved through routine municipal practices. Our findings underscore the importance of sustained commitment to development management as a core tool for limiting flood risk.

Graphical Abstract

Open in new tabDownload slide

Introduction

Despite billions of dollars of investments and widespread mitigation efforts, the costs of disasters in the United States have grown dramatically [1, 2]. Floods are the most prevalent and expensive US disaster, and while climate change plays a role, the primary reason for rising costs is the concentration of people, infrastructure, and economic activities in hazardous areas: an “expanding capital stock” that creates an “expanding bullseye” for disasters [3–5].

For this reason, in 1975, Gilbert White recognized that avoiding floodplain development is possibly “the single adjustment most likely to lead to a decline in national flood losses” [6]. Yet a 2018 analysis by Climate Central and Zillow found that eight US coastal states are building faster on flood-prone lands than elsewhere [7]. From 2001 to 2019, Florida alone built 398 000 new residential properties in the Federal Emergency Management Agency (FEMA) designated Special Flood Hazard Area (SFHA, herein “floodplain”) [8]. In North Carolina, over 75 000 acres of vacant floodplain land in 2017 were zoned for development [9]. Since the SFHA is known to underestimate flood hazards, it is probable that even more development is occurring in areas that experience floods outside these designated regulatory floodplains [10, 11]. Such floodplain development may seem unsurprising given the financial incentives for local governments to develop and the challenges inherent in balancing hazard exposure with economic growth [12–14], but a national analysis of municipal floodplain development finds that only 26% of US municipalities build in floodplains more than would be expected by random chance alone, given the size of their floodplains and rates of development [8]. This invites questions into how municipalities limit floodplain development and how such avoidance might be improved.

The conceptual model underpinning research on floodplain management is that contextual factors shape state and local actions, which in turn shape floodplain development and flood damages (see Fig. 1) [15–22]. Contextual factors studied in prior works include geographic factors like the share of land in the floodplain and nature and value of the waterfront, socioeconomic factors such as wealth and population density, and government capacity elements like financial resources, knowledge, motivation, and personnel [19]. Indicators related to floodplain management actions include the quantity, quality, and implementation of state and local plans, regulations, policies, and programs, although much research has focused specifically on plans [15–22]. Management actions may seek to influence whether or where building occurs in floodplains (i.e. avoidance) or how susceptible that development is to flood damage (e.g. through elevation and building design).

The conceptual model is often implicit in studies. Making it explicit enables critical examination of under-studied connections (green lines, on the left, in Fig. 1). For example, numerous studies have researched the contextual factors that influence local jurisdictions to adopt more numerous, advanced, or better-quality floodplain management actions [15–19, 21, 23–26]. However, whether more numerous, advanced, and better-quality actions lead to less floodplain development is less established because few studies have directly assessed floodplain development outcomes, and among these the evidence is mixed [18, 27]. Burby and French found an “extremely weak” correlation between the number and complexity of land use management actions adopted by towns and the percent of floodplain development permits issued [20]. Brody and Highfield and Brody et al. found implementation but not quality correlated with fewer wetland development permits in Florida [28, 29]. Deyle et al. found “marginal effects” that some but not all measures of high-quality floodplain management in Florida correlated with lower development rates in designated hazardous areas [27]. Burby and French hypothesized that some of this weak correlation may be due to a “land use management paradox,” in which the same factors that predict local floodplain management action (e.g. wealth, population density) also predict increased floodplain development. Burby and French also hypothesized that additional management actions become increasingly less effective as the floodplain is developed. Forty years of development have passed since their study, so the question arises whether the contextual factors and management actions that predicted limited floodplain development in 1981 continue to deliver today, or whether legal innovations and alternative approaches are required.

In this study, we assessed patterns of municipal floodplain development and floodplain management in New Jersey from 2001 to 2019 to understand how towns limited recent floodplain development. We chose New Jersey because it represents an extreme case [30]. Climate Central and Zillow found that housing growth in the New Jersey coastal risk zone was 3.4 times higher than surrounding areas [7]. Five New Jersey cities made their top 10 list for most risk zone development, and Ocean City, NJ, built more flood-exposed homes than any other city in the country [7]. New Jersey faces high development pressure. Yet New Jersey also has a tradition of local autonomy that could be expected to produce variation in both local floodplain management and development [31]. Understanding which New Jersey towns have managed to limit floodplain development allows us to study success as well as failure, to identify not just “weak correlates” of floodplain avoidance but suites of practiced tools, and to avoid the common adaptation research challenge of limiting theoretical insights by selecting solely on the dependent variable [32, 33].

What we find complicates common narratives about floodplain development. Limited floodplain development is the norm, not the exception, in New Jersey. Municipalities routinely limit floodplain development using land use management and hazard mitigation tools that would have been familiar to Burby and French [20] and do so with modest and existing levels of local government capacity. There are multiple paths to limit floodplain development, rather than one best practice, and having a few strong floodplain management tools may be more important than having many tools, potentially explaining the mixed evidence in previous studies. Consistent with prior work, we find that the implementation of floodplain management tools is crucial, and strong efforts taken early can reduce implementation burdens over time, which has implications for efforts to improve future floodplain avoidance. Overall, our results indicate a need for alternative research sampling and analysis methods and more targeted advocacy strategies and policy incentives to improve floodplain management, and we conclude with several recommendations.

Methods

Our analysis took place at the municipal level in three tiers: statewide, across four counties, and in four towns (Fig. 2). First, we examined the relationship between two indices measuring floodplain development and possible geographic and socioeconomic drivers, including wealth and proximity to the coast (see Methods and Supplementary material S1A). Second, we analyzed how floodplain development relates to municipal capacity and management actions in 128 municipalities across four counties (Supplementary S1B). Finally, we conducted in-depth case studies in four towns representing four common floodplain development patterns: Rural River Avoidance (Lumberton), Suburban Coastal Avoidance (Aberdeen), Wealthy Waterfront Development (Weehawken), and Multiple Priorities with Mixed Outcomes (Woodbridge) (Supplementary material S1D). By combining state-wide statistical measures with key informant interviews and legal analyses (Fig. 2), we contribute new empirical evidence linking specific plans, ordinances, and practices to resulting floodplain development.

A full description of our data sources, coding rubrics, and analytical methods is available in the Supplementary material. In brief, this paper analyzes connections across context, local flood management actions, and floodplain development outcomes at state-wide (496 of the 565 New Jersey municipalities), county (128 municipalities in four counties), and local levels (four municipalities representing four development patterns, Fig. 2). We use two new measures of floodplain development. The Floodplain Housing Index (FHI) and Floodplain Development Index (FDI) measure new housing development (FHI) and increased impervious surface (FDI) from 2001–2019 in the 100-year FEMA-designated floodplain (herein ‘floodplain’) (see Supplementary material S1A for details) [8]. Although people experience floods outside the FEMA-designated floodplains, these areas have law and policy implications and are therefore the most likely to be subject to municipal floodplain management efforts. Low FDI and FHI scores (below 1) mean that fewer homes and less infrastructure were built in the floodplain than would have occurred if such construction was sited randomly. Of the 565 New Jersey municipalities, 69 had either no floodplain, no new housing built 2001–2019, or incomplete floodplain maps; our state-wide analyses include the 496 municipalities with complete data. Although New Jersey municipalities include cities, boroughs, towns, townships, and other designations, we will refer to all herein as towns or municipalities unless the distinction is relevant. All land in New Jersey falls within a municipality, so our analysis does not address unincorporated areas within counties that might be relevant in other states.

We used floodplain development outcomes (FDI and FHI) and geographic and socioeconomic context factors to identify four common floodplain development patterns in New Jersey municipalities (see Supplementary material S1D). State, county, and national practitioners helped us select one New Jersey township to represent each pattern: Rural Inland Avoidance (Lumberton), Coastal Suburban Avoidance (Aberdeen), Coastal Urban Development (Weehawken), and Multiple Priorities with Mixed Outcomes (Woodbridge). We avoided outliers in population and floodplain size and selected four townships, rather than villages or cities, to standardize our governance analysis (see Supplementary Table S1) [34]. In each municipality, we visited floodplain development sites and interviewed key informants. Semi-structured interviews provided quantitative rankings and qualitative insights (protocol in Supplementary material S1D.2). We analyzed municipal codes to identify floodplain-relevant ordinances and scored 23 floodplain management tools (ordinances, policies, programs, and practices) for quality based on the nature of the requirement, strictness, and spatial coverage (see Supplementary material S1D.3 for process and rubric). Key informants verified our scores and rated the implementation of tools in practice.

We gathered government capacity and floodplain management data for 128 municipalities in the four counties where our case study towns are located (Burlington, Hudson, Middlesex, and Monmouth). New Jersey County Hazard Mitigation Plans include a “Capability Assessment Worksheet” for each municipality that records the presence or absence of elements such as specialty personnel (e.g. hazard experts, grant writers), technical supports (e.g. early warning systems), fiscal tools (e.g. impact fees), plans (e.g. transportation plan), regulations (e.g. post-disaster recovery ordinance), and other actions (e.g. education and outreach). Each county asks a unique set of questions (from 58 in Monmouth to 99 in Hudson), with only 23 questions in common, so we categorized elements as Personnel, Financial, Plan, Regulation, or Other Action and calculated the percent of positive responses in each category (see Supplementary material S1B, Figs S1 and S2).

Results and discussion

Most New Jersey municipalities limit floodplain development to some extent, even in challenging contexts, and do so with traditional land use management and hazard mitigation tools and with modest levels of government capacity.

Most New Jersey municipalities limit floodplain development more than prior research would suggest given their geographic and socioeconomic context

From 2001 to 2019, 422 New Jersey municipalities (85% of sample) limited the development of new housing in floodplains more than would be expected based on the extent of the floodplain within the municipality and rates of new housing construction (Floodplain Housing Index, FHI < 1) (Fig. 3; Supplementary Figs S3 and S4). Three hundred thirty-five municipalities (68%) limited increases in impervious surface in the floodplain (Floodplain Development Index, FDI < 1). New Jersey municipalities have a mean FHI of 0.45 (median 0.17, range: 0–6.38) and a mean FDI of 0.83 (median 0.5, range: 0–6.98). Scores below 1 indicate that less development occurred in the floodplain than would be expected based on the amount of a town’s developable land in the floodplain and the rate of development. Low FDI and FHI scores do not necessarily reflect “successful” floodplain avoidance (herein “avoidance”) since they are relative measures and therefore do not capture the absolute number of homes or size of infrastructure built in the floodplain. There is a good argument that a town with 10% of its land in the floodplain should place 0% of its new development in that floodplain to be “successful” (rather than <10%). Nevertheless, statewide medians at or below 0.5 indicate New Jersey towns are developing their floodplains less than half as much as might be expected based on the size of the floodplain and rate of development, and some degree of avoidance is the norm (contrary to our expectation, Fig. 4: Result 1 [R1]).

Burby and French found that towns with large floodplains build more in the floodplain [20]. We also find that as the amount of a town’s land in the floodplain increases, they put disproportionately more housing into the floodplain (Fig. 4: R2; Pearson’s, FHI, r = 0.22, P = 5.7e-7; but not overall impervious surface: FDI, r = 0.07, P = .13). Similarly, coastal waterfronts are hypothesized to attract development, and we find higher relative floodplain development rates in coastal towns (n = 92) than state averages (Fig. 4: R3, coastal towns mean FHI = 0.73, all NJ towns mean FHI = 0.45, Welch’s t-test, P = 3.2e-05; coastal towns mean FDI = 1.04, all NJ FDI mean = 0.83, P = .046). Towns with high median property values might develop more in their floodplains to reap property tax revenues, or their wealth might enable them to seek alternative revenue sources [20]. Perhaps reflecting these competing hypotheses, we find no significant relationship between median property value and floodplain development (Fig. 4: R4, Pearson’s, FHI r = 0.02, P = .73; FDI r = 0.06, P = .14). However, in coastal towns there is a significant relationship between median property value and increased floodplain development (FDI r = 0.33, P = .0014), suggesting that a combination of wealth and coastal waterfront may make avoidance more challenging (see Supplementary Table S4).

Indeed, the 126 New Jersey municipalities (25% of sample) that placed none of their new housing in the floodplain (FHI = 0) tend to be less coastal (3% compared to 18.5% overall, Welch’s t-test, P = 1.6e-10) and less populous (11 100 people on average compared to 17 000 across the state, P = .003) and to have 8.5% of their land in the floodplain on average relative to 18% statewide (P = 7.9e-12) (see Supplementary Table S5). Conversely, the 69 New Jersey municipalities (14% of sample) that concentrated new housing in the floodplain (FHI > 1) are more likely to be coastal (39% coastal compared to 18.5% overall, Welch’s, P = .001), with a higher average amount of land in the floodplain (33% compared to 18% statewide; P = .0002) (see Supplementary Table S6).

However, many coastal towns and towns with large floodplains limited floodplain development. Sixty (65%) of New Jersey’s 92 coastal towns have an FHI < 1 (mean = 0.73), and 43 (47%) have an FDI < 1 (mean = 0.47). Four coastal towns placed none of their new housing in the floodplain (FHI = 0), and one fully avoided any new impervious surface in the floodplain (FDI = 0). Similarly, of the 48 New Jersey towns with more than half their developable land in the floodplain, 25 (52%) have an FHI < 1 (mean = 0.85), and 17 (35%) have FDI < 1 (mean = 0.6). Avoidance may be more challenging in towns with coastal shores or large floodplain shares, but it is not uncommon.

Local government capacity predicts an increase in floodplain management actions but is a weak explainer, at best, for floodplain development outcomes

Previous studies have shown that increased local government capacity is positively correlated with the adoption of more floodplain management actions, use of tools that are politically controversial or challenging to implement, higher-quality plans, and reduced flood damages [15–19, 21, 24, 26, 35, 36]. More planning staff, financial resources, and technical expertise have been positively correlated with more numerous and higher quality flood mitigation plans [16, 19, 20, 22, 24, 28, 37, 38]. High-quality plans have been shown to correlate with reduced flood damages [17, 28, 39, 40]. Based on these studies, it is not surprising that higher New Jersey municipal Capacity Scores (mean = 57%, range: 23%–87%, see Supplementary Fig. S2) correlate with an increased number of local plans (Pearson’s, n = 128, r = 0.41, P = 1.9e-6), regulations (r = 0.37, P = 2.1e-5), and other actions (r = 0.67, P = 2.2e-16) (Fig. 4: R5).

However, it is surprising that we do not find statistically significant correlations between local government capacity and floodplain development outcomes (Fig. 4: R6). This holds whether we use the Capacity Score, types of specialty personnel, or financial resources or test individually for nine specific capacity indicators (Supplementary material S2D). In fact, although not statistically significant, an increase in types of specialty personnel present is related to increased floodplain development (FHI r = 0.06, P = .5; FDI r = 0.02, P = .8), and a few elements, such as mitigation planning committees and mutual aid agreements, have statistically significant correlations with increased floodplain housing (Supplementary Table S7). An ordinary least squares (OLS) model using the presence or absence of nine capacity elements did not significantly explain floodplain development (Supplementary Table S8). Some capacity elements may be necessary for floodplain avoidance but are so prevalent across our sample that their presence is insufficient to explain variation in outcomes (e.g. emergency managers and professional engineers are present in 98% of towns).

More local floodplain management actions do not predict increased avoidance of floodplain development, and some actions correlate with increased floodplain development

Local floodplain management actions are expected to reduce floodplain development unless constrained by context, limited capacity, or poor implementation (see Fig. 1) [15–19, 21, 24, 26, 35, 36]. The number of floodplain management actions taken by a town has shown a weak positive correlation with floodplain avoidance in past studies [20]. However, we find that the number of floodplain management actions taken by a town (as reported in County Hazard Mitigation Plans) is not a significant predictor of floodplain avoidance (Fig. 4: R7, Pearson’s, n = 128, FHI: r = −0.05, P = .6; FDI: r = −0.07, P = .45). Much of the existing literature has focused on plans specifically, and we find an increased number of local plans relates to floodplain housing avoidance (Pearson’s, FHI r = −0.22, P = .014) but is not statistically significant with respect to overall floodplain avoidance (FDI, r = −0.15, P = .1). An ordinary least squares (OLS) model using the presence or absence of seven types of local plans to explain FHI is significant but has little explanatory power (F-statistic 3.2, Adj. R² 0.11, P = .007, Supplementary Table S9); plan presence has even less explanatory power for FDI (Adj. R² 0.01, Supplementary Table S9). Intriguingly, as the number of regulations and other actions (e.g. outreach campaigns) that might be expected to implement plans increases, so does floodplain housing development, although the relationship is not statistically significant (Pearson’s, regulations: FHI r = 0.05, P = .6; FDI r = −0.009, P = .9; actions: FHI r = 0.67, P = .45; FDI r = 0.01, P = .9). Emergency Operations Plans (P = .014) and Mitigation Maintenance Programs (P = .0007) show statistically significant correlations with more floodplain housing development (Supplementary Table S10), perhaps reflecting Burby and French’s [20] floodplain management paradox.

Participation in the Community Rating System (CRS) has been previously used as an indicator of local action since the program incentivizes floodplain avoidance [37]. In New Jersey, however, communities that participated in CRS in 2005 had higher rates of floodplain housing development 2001–2019, on average, than those that did not participate (Welch’s t-test, mean FHI in 53 participating communities = 0.88; mean FHI in 443 non-participating communities = 0.4, P = 1.3e-06; Supplementary Fig. S5). The relationship with overall development had a similar direction but was not significant (mean CRS FDI = 0.86; mean non-CRS FDI = 0.83, P = .7). Better CRS scores, which reflect more town actions, do not predict more floodplain avoidance (Fig. 4: R7; Pearson’s, n = 53, FHI r = 0.04, P = .8; FDI r = −0.12, P = .4; Supplementary Fig. S6). Towns whose CRS scores improve over time have been shown at national scales to be more likely to avoid floodplain development [8]. However, in our sample the 40 towns that improved their CRS scores from 2005 to 2019 had statistically similar floodplain development patterns to those that did not improve (Welch’s, P = .24, see Supplementary material S2e). Fig. 4 summarizes our key results, and Supplementary Table S11 summarizes the tests that support each result.

A few high-quality tools, matched to context and consistently implemented over time, are key. Consistent implementation is supported by cooperation and local commitment to flood management

We examined municipal actions in four townships representing four floodplain development patterns: Rural River Avoidance (Lumberton), Suburban Coastal Avoidance (Aberdeen), Wealthy Waterfront Development (Weehawken), and Multiple Priorities with Mixed Outcomes (Woodbridge), and our results indicate that the combination of management actions that is most effective for floodplain avoidance depends on context (e.g. geography, socioeconomics, capacity, and timing). This could explain why statistical analyses in this and previous works find only weak or counterintuitive correlations between tools and development outcomes [18, 20, 27, 28]. For example, Aberdeen, representing the development pattern of suburban coastal avoidance, relied primarily on a conservation/recreation district that covers riverbanks and wetlands, open space acquisitions, and affordable housing development that involved revitalizing brownfields rather than expanding into greenfields. Lumberton, representing rural river avoidance, used a transfer of development rights program to preserve low-density farmland. Woodbridge, multiple priorities with mixed outcomes, used property acquisitions to undo past floodplain development and a mix of zoning regulations that allow commercial development in floodplains but limit housing (Fig. 5; Supplementary Table S11).

No single plan, regulation, policy, or program emerged as a consistent driver of floodplain avoidance. That said, key informants consistently emphasized the value of tools such as zoning or acquisitions that made land legally undevelopable (Aberdeen Interview #1, #4, Woodbridge Interview #1, #3, Lumberton Interview #2, Weehawken Interview #1). Permitting requirements (such as site suitability studies) and building codes (including elevation requirements) were considered useful for reducing the consequences of floods but not for avoidance (Aberdeen #1, #2, #4; Lumberton #2; Woodbridge #1).

Innovative legal tools (such as life estates or conditional permits requiring future decommissioning) and politically sensitive tools (such as eminent domain) were not present in our case study towns and were generally considered unnecessary or even undesirable by interviewees. Woodbridge Interview #1 noted that a controversial tool like eminent domain “could set you back 10 years” by turning public opinion against floodplain management.

Tool quality and implementation have been previously shown to be key, and our results support this (Fig. 4: R8) [18, 28, 35]. The average quality (type, strictness, and spatial coverage) of floodplain management-relevant ordinances across our case study towns does not vary much (Supplementary Table S11) because ordinances often adopted state baselines (e.g. New Jersey Model Flood Damage Prevention Ordinance), which we defined as moderate strictness and therefore drove scores towards the middle. However, averages mask important patterns. Aberdeen (suburban coastal avoidance) had two high quality tools, and both Lumberton (rural river avoidance) and Woodbridge (mixed outcomes) had one, while Weehawken (wealthy waterfront development) had no high-quality flood-relevant municipal ordinance or regulation (see Fig. 5 and Supplementary Table S12). Interviewees stressed the importance of having a few good tools (Aberdeen #1, #3; Woodbridge #1).

Individual leaders championed the adoption of new ordinances in Woodbridge and Aberdeen, but interviewees described implementation as a collective action requiring broad and consistent commitment to compliance across silos and over time. This requires coordination (Aberdeen #1, #3, #4; Woodbridge #1, #3) [17, 41]. One Aberdeen official described the way they worked with others in local government as “a well-oiled machine. I think that’s what protects and saves this community”. One mechanism by which CRS participation influences flood management is through shared information and expertise. Local officials described their CRS cohort (other municipal officials from participating communities in their county) as a crucial source of expert knowledge and a means of educating new floodplain managers to enable continuous implementation (Aberdeen #1; Woodbridge #1).

Local officials’ commitment to hazard mitigation has been previously emphasized as a key factor in implementation, and our informants agree, although they emphasized limiting variances more than punishing violations [17, 19, 26, 42]. In Aberdeen, officials felt that high development demand gave them power to hold businesses and developers to a higher standard: “It’s real simple: you’re doing it this way, or you can go do it in another town… I have news for you. There’s someone else who wants your spot” (Interview #1). A Woodbridge official described the Mayor’s emphasis on flood management as providing both motivation and justification for strict enforcement (Woodbridge #1). Weehawken, on the other hand, preferred to give “a lot of deference to developers” (Weehawken #1) and is currently in litigation for allegedly engaging in spot zoning to allow riverfront development in violation of the town’s master plan. Weehawken officials described floodplain management as a state, not a local, responsibility (Weehawken #2).

Local floodplain avoidance may be unrecognized or be lost in aggregations

In conversations with six federal and seven state officials, we asked for recommendations on places to study avoidance, and all recommended towns that had experienced damage during Hurricane Sandy. Even when we provided a short list of eight towns with limited floodplain development, Lumberton and Aberdeen were largely overlooked because “they’re not doing anything special” (State Interview #5) or their work is “not above and beyond” (Meeting #3). However, the fact that these towns are not doing anything exceptional might be the most interesting result of all: they limit floodplain development with moderate government capacity through the committed, coordinated, and consistent use of a few mundane but high-quality municipal tools.

Conclusion

This study complicates prior work in ways that may indicate a need for different strategies in floodplain management research and practice. We highlight a few here to spark discussion. First, contrary to a narrative in which legal innovations or government restructuring are needed to overcome coastal wealth and development pressures, we find a picture of floodplain management in which most towns are already limiting floodplain development, to little fanfare, using the tools and moderate capacity available to them. A full 126 New Jersey towns (25%) put 0% of their new housing in the floodplain between 2001 and 2019. This reminds researchers to be cautious in aggregating development or damage statistics at levels that mask local successes. Practitioners might consider focusing their reform efforts and attention on the minority of towns rampantly developing in floodplains, rather than promoting statewide changes.

Similarly, efforts to improve existing tools and implementation may be more effective than the adoption of new tools. Calkins described “new plan syndrome” as the tendency to prefer adopting new plans over fixing the implementation of existing ones, but our work reinforces the idea that efforts to improve the quality and implementation of current tools may be more effective than adopting new ordinances, plans, or practices [43]. Moreover, utilizing existing tools may limit “reform fatigue” that can increase resistance to additional regulations [44]. That said, building the commitment and capacity for consistent, long-term implementation is, in many ways, more difficult than disseminating new model plans or ordinances, and it will require concerted effort. Tools that ease implementation burdens (i.e. involve fewer and less-technical decisions, require fewer procedures and less documentation) may improve compliance. Conversely, tools that defer or increase implementation burdens into the future, such as “triggers” or “threshold language” may be politically expedient today but pose challenges for long-term implementation if they rely on future commitment or coordination or expertise [45, 46]. It is worth noting that our informants praised past leaders who set high standards, “so all I have to do now is enforce it!” (Aberdeen #1), as this was seen to reduce burdens on current staff and reduce the level of specialized expertise required. If Burby and French are correct, as our results suggest, that early avoidance eases long-term floodplain management, then aggressive action today may be more beneficial than a plan for increased action tomorrow [20].

Second, if towns can limit floodplain development with a few high-quality tools, and the optimal tool or toolset differs by context, as our study suggests, this might explain why correlation studies focused on individual tools, average quality, or number of tools have found only “extremely weak” or “marginal” connections between floodplain management actions and development outcomes [20, 27]. A potentially fruitful area of research may be to use set relations or other non-correlational research methods to identify policy suites of effective tools and factors driving avoidance and to better understand the political, social, and geographic factors that might make toolsets more or less appropriate in different contexts [47]. For example, Samoray et al. found that new floodplain housing is more commonly built in the most or least affluent US towns, rather than middle-income communities, and different regulatory approaches are likely suited to one or the other extreme but not both [48]. Different approaches are also likely necessary to address different political contexts or to address social justice and equity concerns, and this will be an important area for future research.

Third, this study raises questions about what is being avoided in a floodplain ‘avoidance’ strategy. Woodbridge, like other mixed outcome towns, is building industry but not residences in the floodplain. In case of flood, this strategy might limit residential suffering but still result in significant economic losses and employment, pollution, and debris management challenges. Similarly, Aberdeen, like many coastal suburban towns, is limiting new development but allowing redevelopment, sometimes at a larger or more expensive scale, albeit with higher base elevations, vents, and other on-site mitigation elements. Such floodplain development might be safer or might displace water to neighboring homes or communities (which also raises potential equity questions) [49, 50]. Redevelopment might also contribute to gentrification and housing displacement [51]. Our analysis focused on new development—the conversion of green space to impervious surface or residential structure—as the clearest example of floodplain development, but an increase in the size and value of buildings in the floodplain is a different type of floodplain development and is regulated very differently. These nuances, and their implications for people and their experiences during and after floods, need to be further addressed in future work to grasp the full exposure of flood-prone towns. We focused on tools to limit floodplain development, but further research is needed on how floodplain development affects communities and might disproportionately affect communities of color or less affluent communities [52]. Both new developments and redevelopments might be elevated high enough, armored safely enough, or designed to such a future-proofed standard that neither they nor the roads, utilities, or other services that connect them to the town, will increase the town’s overall flood risk in the coming century. We think this unlikely (and even less likely to be equitably implemented), but towns might be making a calculated decision that the benefits of a limited increase in exposure outweigh the costs. Such calculations presumably include the costs of a municipal budget to pay for future resistance, accommodation, or retreat measures, or they could be reasonably expected to do so. If avoidance is the norm, and concentration the exception, state and federal policies might consider providing less support for the minority of towns that have chosen to continue to increase their exposure.

All of which leads to our overall conclusion: If most towns are able to limit floodplain development—even with moderate capacity and in challenging contexts, using standard municipal ordinances, plans, and programs—then we should wonder not what has gone right in towns that limit development but what has gone wrong in towns that develop their floodplains.

Acknowledgements

We thank Elena S. Hartley for the artwork and design of the illustrations and graphics.

Author contributions

A.R. Siders (Conceptualization [lead], Formal analysis [lead], Funding acquisition [equal], Methodology [equal], Project administration [lead], Supervision [lead], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal]) Jennifer Niemann-Morris (Data curation [equal], Formal analysis [equal], Investigation [lead], Validation [supporting], Writing—original draft [supporting]) Miyuki Hino (Conceptualization [equal], Formal analysis [equal], Funding acquisition [equal], Methodology [lead], Project administration [equal], Supervision [equal], Writing—original draft [supporting]) Elizabeth Shields (Formal analysis [supporting], Investigation [supporting], Writing—review & editing [supporting]) Lidia Cano Pecharroman (Formal analysis [supporting], Investigation [supporting], Writing—review & editing [supporting]) Tess Doeffinger (Formal analysis [supporting], Investigation [supporting], Writing—review & editing [supporting]) Logan Gerber-Chavez (Formal analysis [supporting], Investigation [supporting], Writing—review & editing [supporting]) Ju-Ching Huang (Formal analysis [supporting], Investigation [supporting], Writing—review & editing [supporting]) Alexandra Lafferty (Formal analysis [supporting], Investigation [supporting]) Salvesila Tamima (Formal analysis [supporting], Investigation [supporting]) Caroline Williams (Formal analysis [supporting], Investigation [supporting]) Armen Agopian (Data curation [supporting], Validation [supporting]) Christopher Samoray (Data curation [supporting], Validation [supporting]) Katharine Mach (Conceptualization [equal], Formal analysis [supporting], Funding acquisition [equal], Investigation [equal], Project administration [equal], Supervision [supporting], Visualization [equal], Writing—original draft [supporting], Writing—review & editing [equal]).

Supplementary data

Supplementary data is available at Oxford Open Climate Change online.

Conflict of interest: Katharine J Mach holds the position of Editorial Board Member for Oxford Open Climate Change and has not peer reviewed or made any editorial decisions for this paper. The other authors continue to declare no conflicts of interest.

Funding

A.R.S., M.H., and K.J.M. acknowledge the support of the National Science Foundation (awards 2034308, 2033929, and 2034239). T.D. acknowledges support from NSF award 2104600. The article processing charges were funded by NSF award 2034239.

Data availability

Floodplain Development Index (FDI) and Floodplain Housing Index (FHI) processed data and code are available through the DesignSafe Data Depot repository (designsafe-ci.org, PRJ-4662). Code for the R analyses used standard libraries available from R-CRAN. Interview transcripts are not made publicly available, in line with the human subjects protocol approved by the University of Delaware, University of Miami, and University of North Carolina Chapel Hill. Data that may be available will be provided upon reasonable request to the corresponding author.

Datasets

Multi-Resolution Land Characteristics Consortium (MRLC). National Land Cover Database, “Land Cover (CONUS).” 2019, https://www.mrlc.gov/data/nlcd-2019-land-cover-conus.

MRLC. National Land Cover Database, “Urban Imperviousness.” 2019, https://www.mrlc.gov/data/type/urban-imperviousness.

US Geological Survey (USGS). Protected Areas Database of the United States (PAD-US) 2.1. 2022, https://www.usgs.gov/programs/gap-analysis-project/science/protected-areas.

LANDFIRE. LANDFIRE 2016 Remap (LF 2.0.0). 2021, https://www.landfire.gov/data/lf2016.

References

Author notes