Using the past to understand future property acquisitions: an examination of historic voluntary and mandatory household relocations

As climate change continues to exacerbate hazard losses across the USA, many communities are seeking to permanently reduce their hazard exposure by relocating households out of high-risk areas. In the USA, governments have traditionally relocated households using two mechanisms: buyout programs and eminent domain. Although both programs acquire private properties for public ownership and use, buyout programs are defined as voluntary by government entities, whereas eminent domain is compulsory. For decades, Harris County, Texas, has used both voluntary buyouts and eminent domain to relocate households out of flood-prone areas for different uses; buyouts produce open space, whereas eminent domain is used for flood risk reduction projects Although the literature on buyouts continues to grow, especially related to managed retreat, little research considers mandated relocation in response to hazard in the USA despite the increasing potential and calls for mandated relocations due to sea level rise and other climate-induced hazards. To address this gap in the literature, we analyze county records spanning approximately 100 years to examine the geographic patterns related to Harris County’s use of eminent domain and buyouts programs to mitigate flood hazards. Integrating these records into a Geographic Information System, this novel study is the first to compare these two programs and consider the implications for future managed retreat.


Introduction
With increases in the frequency and severity of hazard events related to climate change, many communities are considering available options for facilitating permanent relocation away from areas at risk for repeated flooding. Though relocation may be considered an inevitable response to these adverse circumstances, questions of how, when, and whom to relocate, and where to relocate them to, continue to pose considerable challenges for all stakeholders. Here, we are primarily concerned with how these relocations are and will be facilitated and, more specifically, the degree to which relocation will be voluntary.
In the USA, relocations are accomplished through two primary mechanisms: eminent domain and buyouts. The history of eminent domain dates back nearly 150 years, and it provides a legal mechanism for the government to forcibly acquire private land for public purposes provided that the owners are provided "just compensation" (Rikon 2018, np). Buyouts, conversely, have been in use since the late 1970s and have only seen widespread use since the mid-1990s (Greer et al. 2022). Buyout programs create open space and therefore, are most often used to advance risk reduction goals, such as removing homes from areas of high flood risk.
Although both eminent domain and buyouts result in a return of private land to public use, federally funded buyouts, such as those implemented using funds provided by the Federal Emergency Management Agency (FEMA), are required to be voluntary, meaning that property owners may be offered a buyout but cannot be forced to sell their properties. While this distinction is clear on paper, several previous studies have questioned whether buyouts can truly be considered voluntary, as social, economic, and political pressures may leave property owners feeling as though they have no option but to relocate (Binder and Greer 2016;de Vries 2017;De Vries and Fraser 2012).
In recent years, the onset of climate-induced relocations in the USA and the prospect for broader retreat and relocations as a climate adaptation tool, policy makers, practitioners, and, to a lesser extent, researchers, have begun to explore the use of forced relocation for efforts that historically would have been considered to fall under the purview of voluntary buyouts (Craig 2019;Koslov 2016). This discussion is treated by scholars and the media as relatively new in the USA, though other countries, including Canada, are actively engaged in mandatory climate relocations (Flavelle 2019), and there is evidence of similar approaches being used in the USA (Lubben 2022). Importantly, however, the potential transformation of voluntary buyouts into mandatory relocation programs reflects a new spin on a long history of attempts to address institutionalized social issues (here, socially constructed exposure to hazard risk) with technocratic interventions.

Study site
Located on the Gulf Coast, Harris County, Texas, is the third most populous county in the USA and has a development pattern that reflects its history of resisting zoning and land use planning. Riverine flooding from coastal storms and hurricanes has led to considerable flood losses for the city (Fulton 2020). Due to development patterns and flood exposure, Harris County has the largest number of severe repetitive loss properties in the nation (NRDC 2020). To mitigate these hazards, the county has utilized property acquisitions to reduce flood-related risks. The buyout program, currently overseen by the Harris County Flood Control District (HCFCD), was scaled up significantly following 2001's Tropical Storm Allison (FEMA 2021).
Since beginning their program in the 1980s, Harris County has acquired over 3000 properties (Greater Houston Flood Mitigaition Consortium 2018). These buyouts have been funded by a number of programs, including the Hazard Mitigation Grant Program (HMGP), Community Development Block Grant (CDBG), Flood Mitigation Assistance (FMA), the City of Houston, Harris County Community Services Department, Harris County Engineering Department, and the US Army Corps of Engineers (USACE) (HCFCD 2021). Published policies by the HCFCD identify five criteria that are applied when considering properties for buyouts: (1) flood risk (considering the source, depth, and risk of flooding); (2) benefit-cost of acquiring the properties; (3) potential for acquired property to contribute to other HCFCD projects, such as drainage projects or retention basins; (4) size of area for acquisition; and (5) owner and community support for buyouts (HCFCD 2020). In addition to the voluntary buyout program, the county has also utilized eminent domain to acquire properties for structural flood control projects dating back to the 1920s.
Given this long-standing history of voluntary and eminent domain property acquisitions, Harris County serves as an ideal case study to examine how both programs have operated historically and the implications for future property acquisitions. Furthermore, with the county being one entity operating both eminent domain and buyout programs, this case allows for a unique and direct comparison between the two approaches and property acquisitions. This exploratory study, which is the first to our knowledge to examine the relationships between buyouts and eminent domain acquisitions, engages in a historical and spatial analysis of all voluntary and eminent domain property acquisitions conducted in Harris County. From this analysis, we draw critical insights needed to shape on-going conversations and future policy related to mandatory relocation as a climate adaptation strategy. Specifically, we examine the spatial relationships (e.g., location within the 1% flood zone) and patterns over time between eminent domain and buyout properties along with property characteristics such as the value of acquired properties, type of property (e.g., household or commercial), and property size (e.g., full or partial property acquisition). Finally, we examine if any of the properties acquired through eminent domain fit within the HCFCD criteria for selecting buyout properties.

Literature review
Property acquisitions have a long-standing, and often controversial, history in the USA and have been implemented for a variety of reasons, including hazard risk reduction (e.g., Nellermoe 2016; Sinclair 2019). As climate change continues to impact coastal and lowlying communities, property acquisitions are increasingly used to relocate residents living in areas at high risk to chronic flooding and/or coastal inundation (Mach et al 2019;Siders 2019). While both the researcher and practitioner communities identify that relocation is an unavoidable climate adaptation strategy for many locations, the question remains as to how to best conduct just acquisitions. In the USA, buyout programs, funded in part by federal grants, are the tool of choice for many communities seeking to relocate households out of high-risk areas (Siders 2019). These programs are described as voluntary, meaning the homeowner can decline to participate in the program. Other countries, however, rely on mandatory relocations for managed retreat including Canada (Flavelle 2019), Australia (Reale and Handmer 2011), and Vietnam (Lindegaard 2020). Increasingly in the USA, 1 3 conversations and practices are shifting from voluntary to mandatory relocations including relocations conducted by USACE (Flavelle 2020b) and other federal entities (Flavelle 2020a). There are a limited number of historic US examples present in the literature, however, namely in Connecticut following the Hurricane of 1938, that detail mandatory property acquisitions using eminent domain with the stated goal of hazard risk reduction (Zavar and Hagelman 2018). This suggests there may be a longer history of mandatory property acquisitions in the USA than presently recorded in extant scholarship.
Eminent domain, the oldest form of property acquisition in the USA and one that is protected under the 5th Amendment of the US Constitution, allows government entities to acquire private property for public use with "just compensation" provided to the landholder (Rikon 2018). In Texas, over 90 different types of government entities have the power of eminent domain including water management boards like the HCFCD (GAO 2006). Comparatively, buyout programs were not federally codified until 1993, when an amendment to the Robert T. Stafford Disaster Relief and Emergency Assistance Act (1988) increased federal support for long-term mitigation projects, including buyout programs, through FEMA's Hazard Mitigation Grant Program (HMGP) (Kihslinger et al. 2017). Although buyouts occurred before the amendment in 1993, this amendment increased the funding and provided a formal mechanism for funding buyout programs.
When local and state entities mandatorily acquire a property using federal funds, they are subject to the procedures and payment amounts outlined in the URA (GAO 2006). Specifically, the URA establishes guidelines for pricing fair market values of acquired properties, relocation payments, and types of assistance available to relocated residents. Although all acquisition programs seek to compensate the property owner, the URA offers guidelines for relocation assistance for eligible tenants, including renters, who must relocate due to a federally funded project (FEMA 1998;HUD 2021). Per interviews conducted with government officials following Hurricane Harvey in 2017 (Binder et al. 2020a, b), Harris County follows the URA guidelines for all types of property acquisitions, including voluntary buyouts. However, it is less clear in the literature, and in practice, the extent to which other communities enforce URA guidelines for buyouts or other voluntary acquisitions.
Furthermore, while the literature on property acquisitions has increased in recent years (Greer et al. 2022), research on eminent domain and hazard management remains sparse in the social science literature. Much of the literature on mandatory relocation is concentrated in law reviews and considers the relevant policy components associated with property acquisitions (e.g., King 2018;Napolitano 2016;Pettus 2019;Supekar 2020). Although this work provides vital contributions to climate adaptation conversations related to property acquisitions, the law reviews frequently lack empirical research thus limiting the depth of knowledge related to mandatory relocations. The nascent studies from the social sciences that engage in empirical research related to mandatory relocations provide a few insights on participant experience, largely aligning with buyout literature.
Issues of recreancy, or the mistrust of government agencies related to the misuse of regulatory powers, are prevalent in both bodies of the literature. For example, Lynn (2016,2017), engaging with an environmental justice frame, interviewed 53 residents of Kashmere Gardens in Houston, Texas. Through these interviews, Lynn identified the presence of recreancy in a structural mitigation project consisting of both forced and voluntary relocations of a portion of a primarily lower income African American and Latino neighborhood. Despite the stated goal of community involvement in the planning process for the flood control project, residents disagreed with government officials on the risk of flooding in the neighborhood as well as the selection criteria for properties for acquisition-both those selected for relocation and those owners who wanted to relocate but were excluded from the program. Furthermore, residents questioned the effectiveness of the flood control project to protect their neighborhood from future flood events. Residents suggested that the flood control project was located in their neighborhood due to the lower value of their properties, making it easier to acquire more properties for the project. Residents also criticized the length of time consumed by the planning process-about a decade-while they remained exposed to flooding.
Issues such as trust in government and government actors (Greer et al. 2019;de Vries 2019;Hanna et al. 2021;Dachary-Bernard et al. 2019), the length of time for program implementation and relocation (Binder et al. 2020a, b), and the uneven experiences of those selected to relocate and those who remain in the periphery (Binder et al. 2020a, b) are also common in the buyout literature. Specifically, de Vries and Fraser (2012), in a study of the perceived voluntariness of four buyout programs across the USA, found that while buyout offer acceptance was generally high, only 65% of homeowners thought the programs were voluntary. They also found that mistrust of program officials was a common sentiment among the 10% of households that refused to participate. When considering the role of buyout administrators, the study found an inherent contradiction in their role, where they must simultaneously represent the interests of the municipality (i.e., high participation rates to signal a successful buyout program) and the homeowner (i.e., the right to refuse to participate). Likewise, they suggest that perceived voluntariness would likely lead to a lack of participation and run counter to an impetus to expedite the process.
Both literatures also highlight the impacts that relocations have on affected populations. In a case of forced relocation, Sanders et al. (2004) examined the experience of older African American adults forced to relocate 30 miles away from their pre-storm housing after Hurricane Andrew impacted Miami-Dade County, Florida, in 1992. All study participants reported decreased mental and physical wellness that was exacerbated by relocating away from social networks and support structures. The majority of participants expressed a desire to return to their pre-hurricane homes, and most intended to relocate again when possible. Similarly, Barile et al. (2020) compared health-related quality of life outcomes across three communities, including one where most residents relocated through a post-hurricane buyout program. Taking a longitudinal approach that tracked changes in key indicators over a period of years, they found residents who relocated after Hurricane Sandy reported significantly lower satisfaction with life compared to a control community where most residents rebuilt in place. Importantly, residents of the community immediately surrounding the relocated community (or the peripheral community, as defined by Binder et al. 2021), fared even worse, experiencing significant declines in general health and satisfaction with life and increases in stress compared to the control community.
Despite these similarities, there are key differences between buyout programs and eminent domain relocations. The primary difference is in the ability of a homeowner to choose to participate in the program. In the case of eminent domain, homeowners are forced to participate, while buyout program are required by federal policy to maintain homeowner choice (FEMA 1998), though scholars have raised questions about the extent to which buyout programs can be considered truly voluntary (Binder and Greer 2016;Hanna et al. 2021;de Vries and Fraser 2012). In addition to impacting the role of homeowners in relocation decision-making, this difference across policy tools represents a significant difference in the practitioner's toolbox. For example, practitioners tasked with implementing a large-scale mitigation project (whether structural or nonstructural) need to acquire the land necessary for achieving that outcome. As voluntary programs, homeowners living in the footprint of a proposed project have the option of not participating, and by remaining in their home, can bring the implementation of the proposed mitigation project to a halt.
The slow-moving nature of buyout funding further complicates these situations, as buyout funding timelines that are misaligned with residents' recovery timelines may lead to buyout program attrition and exacerbate "checkerboarding," where heterogeneous attrition interferes with buyout program goals (Binder et al. 2020a, b). With complications such as these accommodating buyout programs, it is easy to see why practitioners may prefer to avoid the issue of resident choice though the use of eminent domain.
Further, eminent domain is only applicable in the case of structural projects that serve the public good such as the construction of bayous or levees. As such, there is more flexibility or fluidity in determining the physical boundaries of buyout programs. This issue is at the heart of several challenges associated with buyouts, including decisions of where buyouts are implemented and who is (or is not) included in a given buyout program. After Hurricane Sandy, for example, the state of New York implemented a home buyout program in several neighborhoods in New York City. Decisions about which communities were eventually included in the program, and even which households within the selected communities were included, were deeply controversial (Binder and Greer 2016). With buyouts, while some individual funding mechanisms have broad requirements, there are no set federal requirements for determining which properties to include, leaving individual implementing agencies to choose to prioritize properties based on flood loss history, future flood risk, and properties that will maximize the efficient use of limited funds, or other factors that are considered locally important. While this flexibility allows for local input, it also raises significant concerns about the equity implications of such subjective efforts ( Given these disparate challenges and concerns, a critical next step in understanding relocation policy in the USA is to improve our understanding of similarities and differences in the historical application of these two policy tools, which share a common policy history. Specifically, we need to better understand where eminent domain and buyout programs have been implemented, for what purpose, in what social context, and, importantly, what trends have emerged in these areas over time. In this paper, we address these questions using historical and spatial data from Harris County, Texas. We anticipate two different spatial patterns to emerge regarding use of eminent domain compared to buyouts. Historically, urban areas in the USA rely on flood control projects such as dams, channels, and levees to reduce periodic flooding (Montz and Gruntfest 1986;White 1945). During the first half of the twentieth century, these structural projects allowed development in flood-prone areas, which overtime contributed to more development in flood zones. These building practices also reduced ecological infrastructure, such as wetlands, while increasing impervious surfaces resulting in more frequent floods with less open space for flood water to occupy. At present, communities are facing the challenges of these historical land development practices and many are relying on buyouts to remove the homes built in these high-risk areas (Siders 2019). Therefore, buyouts often occur in areas that were historically deemed safe due to structural mitigation projects. To achieve these structural mitigation projects, Harris County often used eminent domain to acquire the needed properties. Based on this history, we hypothesize that historic eminent domain properties will occur in the same areas as buyouts with buyouts serving as the most recent flood mitigation strategy. We also anticipate that eminent domain acquisitions will decrease starting in the 1990s with the rise of federal funds for voluntary buyouts (Peterson et al. 2020). Starting at this time, we also expect to see a shift in the geographic distribution of sites acquired through the two programs, such that buyouts and eminent domain will be implemented in different geographic areas given that HCFCD has specific criteria for buyouts (HCFCD 2020) and utilizes each program to achieve different flood risk reduction goals (e.g., eminent domain results in structural mitigation projects and buyouts create open space). Given these different selection criteria, we hypothesize that eminent domain properties are more likely to occur outside of a flood zone with minimal to no previous flooding since HCFCD prioritizes the use of buyouts for properties in the flood zone (HCFCD 2020). Relatedly, the literature suggests that proximity to flood zones reduces property value (Bin et al. 2008), and therefore, we expect that the single-family properties acquired through a buyout program are likely lower in property value due to their frequent flood history and location within the flood zone. By understanding how eminent domain and buyout programs have operated in the past, we will then be able to develop future studies that will allow for a full and nuanced comparison across these two policies tools, including studies that examine the process and impacts of these programs from the perspective of other key stakeholders, including affected residents and communities.

Methods
We used three distinct datasets to identify trends in Harris County's property acquisitions over time. The oldest database, obtained from the county, is a complete record of all properties acquired through eminent domain by the county since the 1920s. This dataset provides spatial information, project details, size of property, and instrument type. The second dataset, also obtained from the county, is the list of properties the county acquired through a buyout program dating to 1995. This dataset also contains spatial information and property values. We created the third dataset using public records available via the online Harris County Real Property Search Portal by searching deed transactions to identify properties acquired by the county through both eminent domain and federal buyout programs. Using "Harris County Flood Control District" in the grantee search field, searches were performed in early 2021 with the earliest available documents dating to 2000. This provides a 20-year record of property instruments and captures the increase in property acquisitions following 2001's Tropical Storm Allison.
The public records database contains significantly more detail than the other two datasets including location, date, funding source, acquisition price, and demographic data on the seller. Eminent domain properties also include the justification for the acquisition and, in most cases, a description of the resulting project. These public records represent nine instrument types: general warranty deeds, special warranty deeds, deeds without warranty, easements, dedication easements, judge, Lis Pendens, order, and assignments (Table 1). Since assignment documents transfer property from one governmental agency/department to another, with no monetary exchanges, we excluded these documents from analysis since this study is focused on the transfer of private property to public entities. The remaining instruments comprised a total of 2350 property acquisitions by the county (Table 1).
Both types of warranty deeds, the dominant instrument type, contained information on buyout and eminent domain transactions including seller names, price paid, date of deed transfer, type of seller (e.g., individual, government entity, and organization.), and property location; most of this information was also contained on the other instrument types. On the warranty deeds, the buyout transactions list the federal funding source (e.g., HMGP, PDM, CDBG-DR, SRL, or FMA), and the eminent domain records state the reason for the acquisition (e.g., detention basin). For the Judge and Lis Pendens instruments, the records detailed whether the sellers contested the order and a brief reason for contesting. The instruments also identified whether a full or partial tract was acquired. Partial tracts were common with easements and dedications, but not exclusive to these transactions. Property location was included in all instrument types; however, the location format varied from physical addresses to subdivision parcel numbers. To perform spatial analysis in a Geographic Information System (GIS), the physical addresses were geocoded in ArcPro. For the properties that only contained the subdivision parcel numbers, we used property record unit identification numbers and mapped those properties to the subdivision rather than parcel level. Information from all instrument types were recorded for each property transaction in a database housed in Microsoft Excel.

Data analysis
The Excel database was linked to a GIS in ArcPro Desktop for spatial analyses. Data from the county (e.g., NFIP flood zones) were added to identify potential patterns and relationships between acquisitions and the effected neighborhoods. Spatial analyses in GIS included calculating the number of eminent domain properties located within the 1% flood zone and mapping their distribution. We also mapped both eminent domain and buyout properties by acquisition year to identify spatial trends over five-and ten-year time periods. Additionally, we mapped full properties versus partial properties acquired through eminent domain. To complement the spatial analyses, we performed descriptive statistics on the three databases to identify trends and patterns in property acquisitions across time and by funding sources.

Results
The descriptive statistics used in this study identify some general trends related to Harris County's property acquisitions across two distinct time periods. During the 2000-2020 timeframe, which includes the databased with the most detailed property records, we found that the county acquired more properties through federal buyout programs (67%) than through eminent domain (32%). Across a 100-year history of property acquisitions, however, the total number of properties acquired through eminent domain (12,635) exceeds the total number of completed buyout properties (2259) by over fivefold. This increase in buyout program implementation in Harris County follows the timeline of expanded federal support for buyout programs (Kihslinger et al 2017).

Insights from the 2000-2020 record
Focusing on the 2000-2020 acquisition records, we observe that Harris County utilized five different funding sources to conduct buyouts in addition to eminent domain (Table 2). FEMA's HMGP is the most frequent source of funding for their buyouts and spans the longest timeframe; this reflects national trends in funding buyout programs (Wiley and Kousky 2020). The ability to utilize such a range of funding sources suggests that Harris County is effective and adept at identifying and securing available funding for flood control projects.
Looking at the characteristics of the acquired properties during 2000-2020, the average price of properties acquired through eminent domain was higher than those acquired through a buyout program, $293,469, and $152,980, respectively. However, more detailed analysis provides a better picture of property values for these two programs. Since eminent domain acquisitions in Harris County included commercial properties, whereas most federal buyout programs focus on residential households, we removed the commercial properties to analyze the average acquisition price for residential homes. This brought the average property value for residential households acquired through eminent domain to $181,714. We also noted that Harris County has used eminent domain to acquire partial properties in addition to full properties. The partial properties are acquired for improvement projects, especially channelization and drainage, and therefore, do not displace the property owner. Of the residential households acquired through eminent domain, 142 represented partial property acquisitions. Since buyout programs focus on acquiring entire properties, we removed the partial acquisitions and identified that the average price of full, residential properties acquired through eminent domain was $150,710. Therefore, this analysis indicates that the average acquisition price for residential properties through buyout and eminent domain mechanisms are close in value.
Comparatively, on average, the county paid $548,767 to acquire commercial properties. Of note, we found that the county used eminent domain to acquire not only residential and commercial properties but also parcels owned by government entities (e.g., school districts), estates, and religious organizations. They frequently acquired partial properties using eminent domain, often for drainage improvement projects according to the property deed transactions. The deeds for properties acquired through eminent domain identified 55 different structural engineering projects as the reason for the acquisition; the majority of which were detention basins. Of the 1007 eminent domain properties acquired from 2000 to 2020, 389 (38.6%) were partial acquisitions. This is significant since partial acquisitions are used differently for flood risk management and do not displace residents. Mapping the distribution of these partial and full eminent domain records does not show any clear spatial patterns (Fig. 1). Two implications can be drawn, however, from this analysis of partial property acquisitions. First, the commercial properties targeted for partial acquisitions had higher appraised values than recorded in the deed transactions since these public records only show amount paid for the portion of the property acquired. This may suggest that HCFCD did not have the federal funds to buyout these entire properties or that acquisition of the full property was unnecessary for the intended future land use thus using eminent domain to acquire only the portion of the parcel needed for the project. Secondly, the county avoids full property acquisition if possible and has historically used eminent domain to acquire only the land required for flood control projects. This suggests that HCFCD historically avoids buyouts and resident relocation if possible. Rather, HCFCD uses eminent domain to develop structural mitigation projects aimed at reducing flooding without relocating residents when possible. However, we did observe that the county has used buyout funds recently to acquire partial properties. In 2019 and 2020, the county acquired a total of three partial properties using HMGP and PDM funds. This appears to be a new approach for the county to use federal buyout funding for partial acquisitions despite the long-established history of using eminent domain to acquire portions of properties for flood control projects. The deed transactions recorded whether the sellers contested the eminent domain proceedings. For the vast majority of the eminent domain acquisitions (almost 91%), across all property types, sellers accepted the initial offer, avoiding a court case. Therefore, approximately 9% of the sellers rejected the initial offer resulting in a lawsuit by the county to acquire the property. In Texas, mediation and a commissioner's hearing occur after the lawsuit but prior to full litigation for eminent domain disputes (Ashley et al. 2017). Of the 9% of sellers who rejected the initial offer, 22% of those sellers went on to contest the eminent domain in court with all but one objecting to the original award sum (the owners of one commercial property cited insufficient time to relocate). For those who contested the purchase price, the majority ultimately received a higher sum-in many instances, significantly higher-for the acquisition from the county. Since the 20 properties included both full and partial acquisitions, as well as single family and commercial, it is difficult to draw comparisons with this small sample regarding offer values and adjustment costs. The lowest original offer was approximately $17,000, and the highest was over $9 million with no discernable patterns between commercial status or full/partial properties. Of note, most of Note not all records indicated full or partial property acquisition 1 3 these contested acquisitions were for partial property acquisitions. One seller included in their contestation the right for future development on the property they retained, and the county consented to dock and marine facilities developed on the remaining property. Half of these contested properties received an additional award from the county ranging from $3000 to over two million dollars. For the owners request for more time to relocate, the county agreed to an additional 90 days to vacate. Based on the court records, it appears that most of the owners who contested the proceedings received a higher settlement; however, the process for contesting the proceeding requires resources such as time and money thus limiting who is able to contest.

Insights from the 1920-2020 record
The 100-year eminent domain record provides more context on the historic use of property acquisition for flood control projects. The average size of all eminent domain properties was 2.67 acres. Given the average lot size in Houston, TX, is 0.16 acres (Gatea 2021), the acquired properties are larger tracts of land than most residential properties. Beyond parcel size, this historic record also indicates that eminent domain was used to construct structural engineering projects across 22 distinct watersheds including the San Jacinto, Greens, and White Oak Bayou; these are also watersheds where buyout have occurred. The vast majority of these structural engineering projects were related to channelization projects (93%) with the remaining 7% related to basin projects. The high frequency of channelization projects is reflected in the dominant deed type, an easement, which accounted for almost 53% of all deed transactions in this record. Of note, 8.3% of the historic eminent domains were contested in court as represented by judgment deeds; a higher percentage than that recorded in the 2000-2020 record but still low compared to the percentage of deed transfers (34%; including special, general, and no warranty).
When comparing the county eminent domain and buyout records across time and space, we observe the longer history of eminent domain in Harris County along with the fluctuations in property acquisitions (Fig. 2). Although eminent domain acquisitions occurred across the 100-year record, the highest number of acquisitions occurred in 1950 and the surrounding years. Buyouts did not appear in the record until 1995, with the highest numbers of buyouts occurring in 2001 and 2019. When comparing these years of high acquisitions to historic flood-related disasters, we observe a distinct pattern of event-driven acquisitions. A hurricane in 1949 and severe riverine flooding in 1950 accompanied some of the highest rates of eminent domains in the record (HCFCD 2022), whereas Tropical Strom Alison in 2001 and Tropical Storm Harvey in 2017 correspond to increases in buyouts.
When mapping the distribution of eminent domain properties by decade (Fig. 3), we observe that many properties cluster together along the bayou, suggesting that acquisitions were driven mainly by targeted flood control projects that required expansion and updates overtime. Despite this clustering, it is clear that many of the county's bayous have experienced several project iterations requiring acquisitions over time. For example, the northwest corner of the county, Cypress Creek hosts some of the oldest eminent domain projects dating to the 1920s and 1930s (Fig. 4). However, we can also see that newer eminent domain acquisitions, and therefore newer flood control projects, occurred along both the main branch of Cypress Creek and later along the tributaries, likely as new housing developed. This ever-evolving response to flood risk is not surprising given the high-flood frequency of Harris County and the increased flood risks due to rapid urban/suburban growth coupled with climate change.

Location of acquired properties
The spatial analysis of property acquisitions in Harris County identified several patterns. First, in the 2000-2020 record, we observe that both eminent domain and buyouts predominately occur along the bayous and within the 1% flood zone (Fig. 5). Given that location within the 1% flood zone is a criteria used by HCFCD for selecting buyout properties for acquisition (HCFCD 2021), it is not surprising to see clustering of buyout properties within these areas. Spatial statistics indicate, however, that all of the properties acquired through eminent domain since 1920 are located within the present-day 1% flood zone, suggesting these properties do meet at least some of the buyout selection criteria used by HCFCD. Second, although there are some areas where specific acquisition mechanisms are the dominant tool of choice (e.g., the prevalence of eminent domain in the northwestern part of the county), eminent domain and buyouts occur together, often in the same neighborhoods or along the same sections of the bayou. Not only does this reflect the integrated approach HCFCD uses to acquire properties, and their ability to identify a range of funding mechanisms to complete flood risk reduction projects, but also the interwoven link between urban development and flood mitigation where structural mitigation projects encouraged development of high-risk lands that later required acquisitions and relocations to reduce risk.

Discussion
The findings from this exploratory study contribute to conversations related to property acquisitions and climate adaptation in several important ways. This research identifies that hazard-related mandatory relocations have a longer and more robust history in the USA than previously discussed in the literature, though it is unclear whether or to what extent such acquisitions have been implemented in other US communities. Although a recent literature review revealed that buyout programs remain understudied in comparison with their rate of implementation (Greer et al. 2022), research on mandatory relocation in the USA lags even further behind. This is problematic given the interest to enforce mandatory relocations as part of a climate adaptation strategy.
This historical analysis of property acquisitions contributes needed quantitative data and spatial analyses for understanding future acquisitions and identifying future research needs that can inform policy related to managed retreat. Since 2000, Harris County has acquired more properties through buyout programs than through eminent domain. However, the 100-year record shows that eminent domain has been used much more prominently for flood control projects than previously reported in the literature. The transition from eminent domain to buyouts as the tool of choice for acquisitions reflects the timelines for expanded federal funding for buyout programs (Kihslinger et al. 2017) but also reflects the different and evolving HCFCD goals for each program. Given that our analyses identified that HCFCD effectively used many funding sources for property acquisitions, the expansion of federally funded buyout programs in 1993 offered new approaches for hazard mitigation. This varied use of funding mechanisms showcases their ability to effectively secure federal mitigation assistance: a common characteristic of experienced, urban counties (Seong et al. 2021). This varied use of funding also reflects two different goals driving HCFCD to acquire land; HCFCD uses buyouts to generate open space as a nonstructural mitigation mechanism and eminent domain for structural mitigation projects. Whereas voluntary buyout programs can result in checkerboarding and limit post-buyout land use options (Zavar and Hagelman III 2016), eminent domain provides an avenue for acquisition that guarantees full participation when all the land is needed for a project. Moreover, structural mitigation projects liken channelization and detention basins fall under the provision for the public good, a requirement of eminent domain.
Another indication of the county's experience with property acquisitions is their ability to acquire higher value, often commercial, properties. Although the average value of residential households acquired through buyout programs and eminent domain remain close, our analysis indicates that they used eminent domain to acquire much higher value properties, often over a million dollars. While the limited availability of funds for buyouts may lead communities to prioritize lower value properties in order to maximize mitigative impact (de Vries and Fraser 2012), eminent domain offers an avenue to secure more expensive properties for flood control projects. Property value is particularly relevant when we observe the low rate of homeowners contesting eminent domain (about 8% of all properties and, in recent years, less than 2%). Property owners that did contest the acquisition in court owned higher value properties, often over a million dollars. Moreover, the county uses eminent domain for properties that will be incorporated into structural engineering projects aimed at flood control, most commonly drainage basins. Eminent domain allows more leeway on post-acquisition land use whereas federally funded buyouts limit land use to open space (FEMA 2008). A drainage basin does fall within the open space guidelines (Zavar and Hagelman 2016), but other engineered projects may fall outside the scope of allowable post-buyout land uses.
We observe that buyouts and eminent domain acquisitions occur together, often in the same neighborhood or along the same bayou segment. Furthermore, all the eminent domain acquisitions are located within the current 1% flood zone, a criteria Harris County uses to select properties for buyout programs (HCFCD 2020). We hypothesize this cooccurrence is due to patterns of historic development and flood risk reduction projects. Specifically, the data show HCFCD has used eminent domain for structural mitigation projects since 1920. These structural projects have enabled urban growth in areas previously avoided due to flood risk (Montz and Gruntfest 1986). Development of these flood-prone areas has required continued investment and expansion of flood control projects over the twentieth century as demonstrated by the evolution of eminent domain acquisitions along main stream channels and their tributaries overtime. To combat development in areas at risk of flooding, buyouts were used beginning in the 1990s in neighborhoods that previously relied solely on structural mitigation created through historic eminent domain projects. The rapid urbanization of Harris County, increase in impervious surface cover, reduction in ecological infrastructure, and the effects of climate change have all contributed to flooding in areas once assumed to be protected by structurally engineered projects (Brody et al. 2017;Frazer 2005). Despite HCFCD using these two programs for different goals, historic land use development patterns have resulted in these two programs occurring in the same areas.
Despite the proximity of these acquired properties to each other, there remains a key policy difference that influences household level risk reduction. As a condition of their buyout program, the county requires buyout participants to relocate outside of the 1% flood zone (Binder et al. 2020a, b). This is not a requirement for eminent domain and the county has less control over where people relocate. With these policies differing on relocation guidelines, there is a greater degree of certainty that buyouts will reduce household level flood risk, relative to eminent domain. While a number of the criteria provided by HCFCD are subjective, these data suggest that, in most cases, properties acquired by eminent domain would be eligible for buyouts. They are generally within the 1% flood zone, contribute to other county projects, and are generally larger tracts of land. What seems to have changed, in general, is the funding mechanisms available to Harris County to carry out large-scale flood abatement projects.
Although the results from this exploratory study identify some important considerations to shape future research on mandatory relocations, we see the clear need for more longitudinal, qualitative and mixed methods research that seeks to understand residential and implementing agencies' experiences related to relocation through both eminent domain and buyout programs. Without this empirical research, scholars, practitioners, and policy makers can easily fall into the trap of using technocratic problem solving to address social issues, namely using property acquisition policies to relocate households at risk to flooding while ignoring the suite of historical and structural mechanisms that have influenced where people live and their risk exposure (Elliott et al. 2020;Loughran et al. 2019;Zavar and Fischer 2021). Property acquisitions, among other climate adaptation strategies, are evolving on racialized landscapes already influenced by past housing and hazard mitigation policies (Hardy et al. 2017). As Anguelovski et al. (2016) identify, current climate adaptation strategies tend to either prioritize those with higher socioeconomic statuses through acts of omission or displace communities of lower socioeconomic statuses through acts of commission. Without critical, in-depth, and mixed method analyses, property acquisitions will continue to work as acts of commission thus failing those most vulnerable to the impacts of climate change. Therefore, we emphasize that the findings of this exploratory study should serve as a point of departure to help shape future inquiry and research on property acquisitions as a climate adaptation strategy.

Conclusions
When faced with increased flooding in flood-prone areas and flooding in previously unflooded areas, permanently relocating residents out of floodplains is becoming a more commonly used approach to reduce risk. To move residents, governments have few tools at their disposal, including eminent domain and voluntary buyouts. Although buyouts have a small, but burgeoning literature, studies exploring the use of eminent domain in floodplains are even more limited. In this study, we relied on county records to examine geographic patterns in how Harris County, Texas, has used eminent domain and buyout programs to mitigate flood hazards. Our quantitative and spatial analysis capture a few key findings. When comparing the two acquisition methods, the county has acquired more properties through buyout programs than through eminent domain since 2000, but eminent domain has been used much more prominently and over longer timeframes for what could be interpreted as flood control projects than previously reported in the literature. While the average value of residential properties comparable across both programs, eminent domain has been used to acquire much higher value commercial properties in floodplains. What is notable, however, is how Harris County has funded their acquisitions. Although many buyout programs rely on a single or a few funding sources, the county uses many funding sources for property acquisitions, and a sign of a mature and experienced agency that can effectively secure federal grants. In effect, this indicates that they are using a number of mitigation tools and funding sources to address identified vulnerabilities across time and hazards, which stands in contrast to the traditionally reactionary nature of buyout programs. Whether this technocratic approach to reducing hazard vulnerability is effective or just, however, is a key question that is beyond the bounds of this study to fully address.
Our exploratory work has several limitations that should be acknowledged and that offer avenues for future research. First, based on the limitations of our datasets, we cannot track relocatees over time. This limits our ability to determine whether either tool actually reduces the risk profile for relocatees. For instance, while buyout participants are required to move outside the floodplain, they could move into an area at risk for wildfires or increased crime. While this may fall outside the purview of the county, reducing risk holistically should be a priority with mitigation efforts, particularly when they are federally funded. Future work should address these tools longitudinally at a household level to determine if they are reducing risk at multiple scales. Relatedly, the datasets we used contain limited demographic data. Future research should incorporate census data to provide more context on the spatial distributions of property acquisitions related to socioeconomic factors. Second, absent interviews with relocatees, we cannot capture the differences in the experience of having your property subject to eminent domain as compared to a buyout offer. Although the literature suggests that many buyout participants do not see offers as voluntary (Greer and Brokopp Binder 2017;de Vries and Fraser 2012), there are likely differences in the process that lead to different perceptions by participants and could lead to differential outcomes. Lastly, while we can comment on the spatial patterns within and across the tools, we cannot assign intent in the use of the tools. Future studies should include interviews with eminent domain and buyout administrators to understand their decision-making process.