: Mapping the adaptation solution space – lessons from Jakarta

Coastal cities are under rising pressure to adapt to climate change. They suffer from the severe effects of increased frequencies and intensities of coastal hazards, particularly flooding, while oftentimes continuing to sprawl into hazard-exposed areas and grow beyond the pace of sufficient infrastructure development. Even though these problems have been quite well understood for a while, there is still comparatively little knowledge and scientific assessment of the solution space, i.e., on the options available for adaptation and the ways in which they are being perceived, framed and evaluated in the scientific literature. Focusing on Jakarta, this study presents findings from a systematic assessment of peer-reviewed scientific literature on the adaptation solution space with regard to current and future flooding. Jakarta is chosen as a case study since it is among the cities with the highest flood risk and adaptation pressure globally while also being one of the most heavily researched coastal cities in this regard, certainly in the Global South. Based on a structured keyword search, we assess 339 articles. Results indicate that the perceived solution space is skewed towards hard protection against flooding, while measures to accommodate flooding or retreat from exposed areas are less widely considered in the scientific debate. Soft adaptation measures for the reduction of social vulnerability receive less attention in the literature than those measures targeting the taming of flood hazards, often through engineering solutions. Likewise, hybrid adaptation approaches, which combine soft and hard measures in a complementary way, are only rarely considered. Looking into the future, the findings suggest that despite the importance of hard flood protection as a main adaptation solution in Jakarta, other fields of the solution space deserve increased scientific attention. This relates in particular to urgently needed feasibility and effectiveness assessments of ecosystems-based solutions for flood mitigation and adaptation options targeting social vulnerability. While the empirical results are specific to Jakarta, heuristic observations from research on other coastal cities suggest that similar scoping exercises of the predominantly perceived solution space might be of relevance in many cities beyond Jakarta.

Abstract. Coastal cities are under rising pressure to adapt to climate change. They suffer from the severe effects of increased frequencies and intensities of coastal hazards, particularly flooding, while oftentimes continuing to sprawl into hazard-exposed areas and grow beyond the pace of sufficient infrastructure development. Even though these problems have been quite well understood for a while, there is still comparatively little knowledge and scientific assessment of the solution space, i.e., on the options available for adaptation and the ways in which they are being perceived, framed and evaluated in the scientific literature. Focusing on Jakarta, this study presents findings from a systematic assessment of peer-reviewed scientific literature on the adaptation solution space with regard to current and future flooding. Jakarta is chosen as a case study since it is among the cities with the highest flood risk and adaptation pressure globally while also being one of the most heavily researched coastal cities in this regard, certainly in the Global South. Based on a structured keyword search, we assess 339 articles. Results indicate that the perceived solution space is skewed towards hard protection against flooding, while measures to accommodate flooding or retreat from exposed areas are less widely considered in the scientific debate. Soft adaptation measures for the reduction of social vulnerability receive less attention in the literature than those measures targeting the taming of flood hazards, often through engineering solutions. Likewise, hybrid adaptation approaches, which combine soft and hard measures in a complementary way, are only rarely considered. Looking into the future, the findings suggest that despite the importance of hard flood protection as a main adaptation solution in Jakarta, other fields of the solution space deserve increased scientific attention. This relates in partic-ular to urgently needed feasibility and effectiveness assessments of ecosystems-based solutions for flood mitigation and adaptation options targeting social vulnerability. While the empirical results are specific to Jakarta, heuristic observations from research on other coastal cities suggest that similar scoping exercises of the predominantly perceived solution space might be of relevance in many cities beyond Jakarta.

Introduction
Many coastal cities around the world suffer from chronic flooding, straining their development (Hallegatte et al., 2013). Looking into the future, risks related to flooding in these cities are set to rise sharply (IPCC, 2019a). This rise in risk is driven by climate change effects (e.g., sea level rise and the increasing intensity of heavy precipitation, river flooding and storm surges) but also the effects of urbanization itself (e.g., land subsidence, urban sprawl into floodprone areas, or the accumulation of people and infrastructure) (Tellman et al., 2020;Wolff et al., 2020). As a result, coastal cities are under increasing pressure to adapt over time and in some instances transform fundamentally (IPCC, 2019a;Revi et al., 2020). While this is not a new phenomenon, even well-researched cities like Jakarta, Indonesia, with sound scientific knowledge of their flooding problems and considerable efforts to improve flood risk management and climate change adaptation, keep suffering from flooding year after year.
Persistent flooding hints towards the existence of "adaptation gaps" (UNEP, 2018). In order to better understand and address such gaps, this study looks at scientific research not only on past, current and future trends in flood risk but particularly on adaptation for the case of Jakarta. Research on this city shows a pattern which appears to be typical of coastal cities: while the problem of flooding and its drivers has been quite well researched for Jakarta (e.g., Abidin et al., 2015;Asdak et al., 2018;Budiyono et al., 2016;Latief et al., 2018;Mishra et al., 2018;Moe et al., 2017;Ward et al., 2011a), much less attention has been given to analyzing different potential adaptation options. The main focus of this paper is therefore on examining how the socalled "adaptation solution space" for Jakarta is being covered and framed in the literature. The concept of "solution space" for adaptation to climate change has been receiving increasing attention since the IPCC's Fifth Assessment Report (WGII AR5) (IPCC, 2014). The adaptation solution space can broadly be understood as being made up of potential adaptation options including their synergies and tradeoffs as well as barriers and enablers. Assessing the adaptation solution space, including the feasibility, effectiveness and adequacy of different adaptation options -and their combinations over time -is essential for informing the composition of adaptation pathways. The solution space for climate change adaptation therefore represents a socially constructed, multidimensional space of opportunities for adaptation that determines "why, how, when and who adapts to climate change" (Haasnoot et al., 2020, p. 1), restricted by hard and soft limits to adaptation (Dow et al., 2013). This study assumes that the scientific discourse is one important arena in which the solution space is formed and shaped. The questions of which adaptation solutions are being conceived in science and how they are being scientifically evaluated have great influence on the practical and political debate shaping actual solution decisions (Haasnoot et al., 2020). This paper therefore focuses on the assessment of scientific literature and how adaptation solutions are being perceived and treated in them. Further research might complement this analysis with empirical work on adaptation perceptions among, for example, practitioners, policy-makers and affected actors.
Jakarta represents an example worth analyzing. The city's flood problems are so pressing that it can serve as an early laboratory for current and future adaptation challenges that many other major coastal cities, which are also located in low-elevation coastal zones, e.g., Mumbai, Dhaka, Ho Chi Minh City or Lagos, will also have to deal with (IPCC, 2019a). The combination of continuing urbanization and environmental changes results in increasing risks for the urban population today and in the future, turning Jakarta into one of the most at risk coastal cities globally (Hallegatte et al., 2013;Hanson et al., 2011).
To examine which adaptation options are being considered in the scientific literature and how, the study uses a structured literature review to address the following research questions: -Which flood risk drivers are considered in research on Jakarta?
-To what extent does scientific research consider different hard, soft and hybrid measures for risk management and adaptation, and how are these measures being evaluated?
While the paper is focused on Jakarta, it aims at providing lessons for the assessment and understanding of adaptation solution spaces in other coastal cities facing similar risk and adaptation pressure. The paper is structured into seven parts. Section 2 lays out conceptual considerations around flood risk and adaptation adopted in this study before Sect. 3 briefly introduces the flood context of Jakarta. Subsequently, Sect. 4 describes the methods and data of our analysis. Section 5 presents the results starting with general publication patterns and author affiliations. Subsequently, the results chapter describes drivers of flooding mentioned in the literature before it summarizes the hard adaptation measures discussed. Then, soft and hybrid adaptation measures mentioned in the literature are presented. Section 6 summarizes and discusses the results, relating them to the solution space for flood adaptation for Jakarta. It also presents identified gaps. Section 7 offers key conclusions and an outlook.

Conceptual considerations -flood risk and adaptation options
Climate change adaptation first and foremost means action to limit and reduce climate risks . In order to understand whether and how adaptation is being framed and perceived, it is therefore, we argue, necessary to concentrate on the links between adaptation and risk. In other words, to unpack the adaptation solution space -or rather how it has been discussed in the scientific literature in this case -one therefore has to ask whether and how different adaptation options are considered to have an effect on the different components and drivers of risk . In addition, a key question is which factors of risk might be underrepresented in the current adaptation literature. For these reasons, understanding how risk is being produced and composed is essential. In this study, we draw on a few decades of risk research and understand risk in line with current concepts used in the IPCC to be a function of hazard, exposure and vulnerability (Wisner et al., 2004;IPCC, 2012). According to this, hazards can be defined as (environmental or climate-related) events or processes with the potential to cause damage and harm (IPCC, 2019b). In the case of Jakarta this predominantly means floods. Exposure is understood as the presence of assets or activities (social or environmental) in the spatial, temporal and/or functional reach of hazards. Exposure therefore has a hybrid character as it can be altered by environmental changes (e.g., sea level rise) as well as socioeconomic change (e.g., urban sprawl). Vulnerability refers to the propensity or predisposition to be adversely affected if exposed to a hazard (IPCC, 2019b). For example, the de-gree of susceptibility of livelihoods or infrastructure to suffer harm from floods contributes to vulnerability.
Adaptation links into the causal fabric of risk by aiming to reduce existing as well as future vulnerability (e.g., through health care or other social security programs); exposure (e.g., through planned retreat from hazard-exposed areas); and/or, where possible, hazards (e.g., by limiting flood intensity through retention areas) . Overall, adaptation can hence be defined as the process of adjustment to actual or expected climate change and its impacts in order to moderate harm or, where possible, even exploit beneficial opportunities (IPCC, 2019b). Addressing the different risk components (hazards, exposure and vulnerability) involves assessing and selecting options for policy and action . Such decision-making entails evaluation of the effectiveness, efficiency, efficacy and acceptance of actions . Limits to adaptation apply where available options no longer allow actors to secure valued objectives, functions or assets from intolerable risk (Dow et al., 2013). While coastal cities might reach technical limits of adaptation only rather late (e.g., in terms of the engineering limits theoretically applying to coastal protection), financial, social and institutional barriers and limits are expected to be reached much earlier (Oppenheimer et al., 2019).
Risk analysis is often split into understanding the hazard and understanding other driving forces -including physical and/or social exposure as well as vulnerability. For a long time, scientific research has focused primarily on understanding the hazard with the objective to control nature and protect people from it (Hewitt, 1983;Wisner et al., 2004;IPCC, 2012). In the same vein, the role of "hard" adaptation solutions, which aim at controlling hazards and protecting exposed elements, have grown in importance, becoming a centerpiece of the predominant paradigm for risk reduction in the second half of the 20th century (Hewitt, 1983;Wisner et al., 2004). Hard adaptation measures in the context of flood risk reduction are mostly large-scale engineered human-built structures, e.g., floodwalls or storm barriers (Sovacool, 2011;Oppenheimer et al., 2019;Du et al., 2020). While they often meet their objective to protect people and systems from harmful events and are widely considered an important element within portfolios of risk reduction measures especially in coastal cities (Oppenheimer et al., 2019), they rarely work towards reducing underlying hazard drivers or social vulnerability and they often entail significant downsides. First, they tend to be technologically complex -often prone to failure -and very cost-intensive (Sovacool, 2011). Second, they are comparably inflexible as concrete structures remain for a long time. This can be challenging in the face of high levels of uncertainty regarding climate change and dynamic trends in its impacts, which mean that such structures need constant assessment and sometimes costly updates (David et al., 2016). Third, hard measures in the aggregate generate comparatively few co-benefits and, depending on the planning and implementation process, have even been harmful to local communities and ecosystems (David et al., 2016;Sovacool, 2011). And lastly, infrastructural measures might give people a false sense of security, increasing the overall damage potential in the case of failure (IPCC, 2012). Risk reduction and adaptation regimes centered around hard protection predominantly or exclusively have therefore been problematized for their emphasis on technocratic fixes for solving symptoms rather than causes of risk (Ribot, 2011;Garschagen, 2015;Solecki et al., 2017), paving the way towards addressing the need for changing the protection paradigm towards more holistic risk management approaches (e.g., Viero et al., 2019).
Next to hard adaptation, the importance of soft adaptation measures has therefore also been emphasized for a while, especially for reducing socio-economic vulnerability (Wisner et al., 2004;Ribot, 2011;Solecki et al., 2017) or for absorbing residual risk remaining beyond hard measures (Du et al., 2020). In contrast to hard adaptation, soft adaptation includes an emphasis on ecological and institutional responses, notably ecosystems-based approaches and institutional adjustments, e.g., in terms of land-use planning, building codes, social protection or awareness raising. Soft adaptation is less clearly defined than hard protection, meaning also that the consideration of properties, advantages and disadvantages of measures belonging to soft protection is multifaceted. Yet, a number of overall observations have been suggested in the literature, notably that soft protection is focused on empowering and capacitating local communities to respond to changing hazards and is often based on modular technologies which do not require large outlays of capital or human resources (Sovacool, 2011). However, this is not to argue that certain soft measures also require a large amount of central planning, investment and steering, e.g., in the case of largescale wetland or mangrove restoration.
Hard and soft adaptation measures are often combined and can both be mapped onto the main response types against sea level rise and coastal flooding as used by the IPCC, i.e., protection, accommodation, advance and retreat (Oppenheimer et al., 2019). However, certain clusters can be observed, e.g., in that protection typically relies on hard measures, whereas accommodation typically also requires a stronger integration of soft measures. Assessing coastal adaptation approaches and potential across the globe, the IPCC stresses that hard adaptation is technologically feasible and economically efficient for coastal cities and therefore will continue to play a central role in adapting such cities further (Oppenheimer et al., 2019). However, the IPCC also stresses that hard protection does not come without disadvantages and raises questions of affordability, particularly in poorer regions of the world (Oppenheimer et al., 2019). There is therefore high agreement that hybrid approaches, combining different hard and soft approaches, is a promising way forward in many coastal settings (Oppenheimer et al., 2019). For Shanghai, for instance, hybrid approaches of combin-ing hard storm-surge barriers with wetland development and wet-proofing of infrastructure have been assessed to bring about the highest potential for overall risk reduction (Du et al., 2020). An example from Padua (Italy) shows how proper floodgate operations are ensured by including the end user in designing and implementing control structures and protocols (Mel et al., 2020).
The above considerations mean that measures of both types jointly make up the so-called solution space for climate change adaptation which can be understood as a flexible space spanning multiple dimensions (biological, political, institutional, socio-economic, cultural), scales and actors, containing all potential solutions for climatic risks (Haasnoot et al., 2020). The solution space is confined by soft and hard limits (Dow et al., 2013) and can hence change in form depending on internal and external influencing factors (Haasnoot et al., 2020). This study assumes that scientific research in the field of flood risk management and adaptation represents one of these influencing factors. Through scientific inquiry, scholars assess and evaluate potential flood adaptation options from many different perspectives, creating a diverse and constantly widening landscape of adaptation options, which are readily available for the consideration of decision-makers. Accordingly, scientific perspectives play a vital role in shaping the actual solution space.
3 Brief overview of Jakarta's flood risk and its root causes, impacts and recent management Urban flooding has a long history in Jakarta. The city is naturally prone to coastal hydrological hazards due to its geographical location in a low-lying coastal area facing the Java Sea, and it having more than 13 rivers, including the Ciliwung, flowing through it . Urban flooding in Jakarta is most severe when heavy precipitation, high runoff rates, and storm and/or high tide levels coincide . In the future, climate change is expected to increase Jakarta's "natural" drivers of flood risk, mostly through sea level rise and the increasing potential for heavy precipitation in the entire Ciliwung catchment area Januriyadi et al., 2018). However, besides natural drivers of flooding, there are also multiple human-made causes which significantly contribute to the city's flood problem. First, continuous population growth, urbanization and land-use changes in Jakarta as well as in its surrounding areas, including the upstream area of the Ciliwung, have significantly altered the hydrological system and runoff patterns . Urban densification processes have led to a degradation of the urban drainage system as rivers and canals have been filled and floodplains paved over, reducing the retention and discharge potential . Second, the narrowing of urban waterways due to informal settlements along the banks of the rivers, sedimentation and pollution with waste have further reduced water flow capacities and urban drainage (e.g., Mathewson, 2018). Third, land subsidence is a key driver of the city's flood problem because it exacerbates the impacts of precipitation and sea level rise . Today, 40 % of Jakarta's urban area lies below sea level Salim et al., 2019). Subsidence has four major drivers: excessive groundwater extraction, natural consolidation of soil, increasing infrastructure and building load, and tectonic activities . The impacts of floods, resulting from the interplay of both natural and human-made drivers as described above, have represented a strain on the city's development until today. Major flood events hit the city in the 21st century, i.e., in 2002, 2007, 2013, 2014 and 2020. They resulted in several deaths, up to 500 000 evacuees, and massive direct and indirect economic losses related to infrastructure damage and reduction in productivity as well as business value losses (e.g., Budiyono et al., 2015;Djalante et al., 2017;Octavianti and Charles, 2019). The 2007 floods stand out in their extent and severity. The floods submerged more than 60 % of the city, causing an unprecedented flood extent and unprecedented fatalities, damage and losses (Texier, 2008). They resulted from the confluence of high precipitation in the city, water runoff from upland areas and a strong spring tide pushing water into the city from the sea . This extreme event, driven by compound flood drivers including the seaward intrusion of water, can be described as a demarcation point, triggering a paradigm shift in flood risk management in so far that sea level rise was from then on portrayed to be one of the risk drivers Octavianti and Charles, 2019). While prior to the 2007 event the city government had mostly focused on protecting the city from inland flooding, it then developed the so-called Jakarta Coastal Defence Strategy (JCDS) to also protect the city from coastal hazards . Since then, the city government has concentrated its flood risk management on four key infrastructural measures: namely river and waterway regulation, including dredging and clearance of river banks; canalization; expansion of water reservoirs; and the development of a massive sea wall including land reclamation Octavianti and Charles, 2019). The so-called Great Garuda is one of multiple flood protection measures of the National Capital Integrated Coastal Development (NCICD) plan adopted in 2014. Its shape resembles a Garuda, the Indonesian national bird (see Fig. 1). Developed by a consortium of Dutch and Indonesian planning consultancies, the "giant sea wall" is supposed to protect Jakarta's bay area from the sea. In combination with other infrastructural measures of the NCICD master plan, the bay area will be completely re-invented and developed as a new residential and business district , aiming at marrying coastal protection with urban development while at the same time addressing land subsidence -a problem that started receiving increased attention particularly after major floods in 2013 and 2014.
The question of to which extent the entire project with all its elements will be implemented is the subject of continued debate (cf. Minkman et al., 2021).

Methods and data
This study builds on a two-tier analysis scheme. First, a systematic search for English scientific literature in the citation database Scopus was conducted, aiming at obtaining an overview of the state of research on adaptation and flood risk reduction in Jakarta. While capturing the majority of the international scientific literature, this selection does not capture studies published in other languages such as Indonesian. Nonetheless, the scoping of articles as well as the discussion of our approach and preliminary findings with Indonesian colleagues suggests that our approach is able to capture the ongoing scientific debate quite comprehensively. This is in large part due to the fact that Indonesian scholars today quite actively contribute to the English-language peerreviewed scientific literature, thereby transporting research results of many local to regional studies into that body of knowledge (Djalante, 2018; see also Sect. 5).
As shown in Fig. 2, the Scopus literature database was searched with a deliberately broad combination of the search terms Jakarta AND ("flood * " OR "coastal hazard * " OR adaptation), limited to roughly the past 20 years (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018)(2019). In total, 339 publications in English (see Appendix A) resulted from the structural searches after excluding nonrelevant research fields (medicine, mathematics, chemistry, biochemistry, nursing, chemical engineering, dentistry, immunology, veterinary medicine), building the basis for the following filtering and analyses. For seven papers, either they could not be accessed or no full-text versions could be found (Appendix B). In a first step, all abstracts of resulting publications were skimmed and topical categories were developed, resulting in an inductively derived thematic scheme (see Table 1 and Appendix C). Based on the number of papers per category as well as their content, we obtained an overview of Jakarta's flood risk and adaptation research landscape and its centers of gravity. This is taken as an indication of whether and how different fields of scientific inquiry contribute to the assessment of flood risk and adaptation solutions. Of course, the number of papers alone does not sufficiently evaluate the importance of a study field for framing the solution space and shaping political decision-making. Nonetheless, we argue that the number of publications can be one proxy -among others -for evaluating the intensity of a certain debate, in this case the engagement with particular flood drivers and adaptation solutions.
What is more, our analysis draws not only on a bibliometric assessment but also on the content analysis of a subset of papers. In line with the research interest of our paper, we selected 93 publications for a detailed content analysis, which deal with otherwise underrepresented segments of the adap-tation solution space, i.e., those focusing on soft and mixed adaptation options and a focus on social vulnerability, policy analysis and integrated management (categories 1, 2, 7 and 8 of Table 1). The analysis was guided by the research questions (Sect. 1). It follows the principles of a qualitative content analysis (Mayring and Früh, 2002) and draws on a combination of deductive and inductive codes (see Appendix C and D).

Results
As depicted in Fig. 3, scientific research on flooding in Jakarta has been rapidly rising since the year 2015. Between the beginning of 2015 and today, almost 3 times as many documents were published compared to the period of 2000 to 2014. In comparison to global flood risk research, which has increased steadily over the years, this represents a remarkable surge, hinting towards the relevance of the topic of flooding in Jakarta as well as an increasing scientific interest.
Research attention was especially high among Indonesian scholars. As visualized in Table 2, almost 50 % of first authors are affiliated with Indonesian institutions. With regard to international attention to Jakarta's flooding issue, most first authors are affiliated with Japanese and Dutch institutions, each making up roughly 11 % of the resulting publications, followed by researchers from Germany, the USA, Australia and Singapore.
According to the subject fields provided by the Scopus citation databank (multiple fields are possible per paper), more than three-quarters of all publications include a natural science or engineering perspective (environmental science, Earth and planetary sciences, engineering, computer science, energy, physics and astronomy, agricultural and biological sciences, materials science, mathematics, psychology, decision sciences) (see Fig. 4). Only 15 % can be attributed fully or partially to social sciences, and very few include an economics perspective (business, management and accounting, economics, econometrics and finance). Only 2 % are labeled to include an arts and humanities angle.
To obtain a clearer and more detailed overview, Fig. 5 shows a classification of the resulting publications in terms of the studies' content and focus, building on an inductively developed categorization scheme (Table 1). Some clear clusters and patterns emerge: first, studies on quantitative flood modeling, land-use (change) impacts on flooding and hard adaptation options together dominate the research field, representing almost 50 % of all publications (grey fields). This corresponds with the high number of papers in subject fields from the natural sciences (see Fig. 4). Second, with around one-third of all publications, studies in the areas of soft and hybrid adaptation analyses, soft factors of adaptation, policy and legal analyses, and watershed management (green fields) together represent another stream of scientific research. The rest of the studies of this analysis are very diverse, spanning  qualitative risk descriptions, the employment of new data types for risk analysis, responses to early warning systems and other decision support systems, and land-use change analysis in the context of flooding. Together, they represent roughly one-quarter of all analyzed publications.

Understanding the drivers of the flood hazard
As shown in Fig. 5, there is a strong focus on flood modeling and mapping, representing almost one-third of all papers published since the year 2000. Flood models (Bahtiar et al., 2018;Farid et al., 2012;Formánek et al., 2013;Jati  2 Policy and legal analysis Papers focusing on, e.g., institutional analysis, national policy, legal frameworks of risk management, vulnerability reduction and adaptation. 3 Hard adaptation Papers that exclusively look at hard physical adaptation measures such as the Great Garuda project, infrastructure for rainwater harvesting, polders, dikes and flood barriers, embankments and river diversions.

Flood models & flood mapping
Papers that present quantitative precipitation models, subsidence models, flood loss estimation models, urban drainage models, sea level rise models, community-based flood risk mapping, shoreline retreat models.

5
Land-use (change) impact on flooding Papers that examine the criticality of watersheds or land-use change and its impacts on flooding with the help of quantitative models or qualitative case studies. 6 New data types Papers that investigate the potential of using new data sources like social media, big data or high-resolution data or new data-generating formats such as crowdsourcing or e-participation for flood risk mapping and analysis. 7 Watershed management and water governance Qualitative analyses of reasons for flooding and water pollution, including drinking water source analysis and/or models.

Soft and hybrid adaptation
Papers that focus on soft and hybrid adaptation strategies including soft measures, local/community-led adaptation, firms and adaptation, resettlement/relocation, alternative energy sources, disaster management, urban adaptation planning.

9
Early warning Papers that present GIS-based early warning systems, risk communication and information needs during disasters.

10
Decision support systems Decision support systems for locations of warehouses, disaster information management systems, socio-economic vulnerability indices, hydrological infrastructure flood vulnerability indices, integrated assessment framework for subsidence.

11
Qualitative risk descriptions Papers presenting information on flood events and impacts or evolution and impacts of land subsidence. In contrast to category 4, no quantitative hazard or risk models are employed.
With the objective to identify spatial patterns of rainfall (Latifah and Setiawan, 2014), subsidence (Abidin, 2005;Abidin et al., 2015; and flooding/inundation Latief et al., 2018;Margatama et al., 2018;Nuswantoro et al., 2016;Tambunan, Figure 3. Number of scientific publications in flood and adaptation research in Jakarta and globally (searches in Scopus were (1) Jakarta AND ("flood * " OR "coastal hazard * " OR adaptation) and (2) "flood * " OR "coastal hazard * " OR adaptation, both for the years 2000-2019 and excluding the following subject areas: medicine, mathematics, chemistry, biochemistry, nursing, chemical engineering, dentistry, immunology and veterinary medicine).   Apart from the latter two exceptions, studies from the categories flood models & flood mapping and hard and physical adaptation tend to focus on climatic, hydrological and physical factors contributing to flooding, thereby providing key information on flood drivers and patterns. Such studies are much higher in number than risk assessments that also include non-hydrological risk drivers (see below).

Hard flood protection measures and their evaluation
Considering the focus on natural or geo-physical drivers of flooding (see Sect. 5.1), it is not surprising that there are many publications that concentrate exclusively on socalled structural, engineered or hard flood protection measures. Such measures had already begun to be implemented during colonial times and remain, until today, a main pillar of Jakarta's approach to mitigate flooding (Colven, 2017;Goh, 2019;Mathewson, 2018;Octavianti and Charles, 2018;Owrangi et al., 2015;Padawangi and Douglass, 2015;Simanjuntak et al., 2012;Ward et al., 2013a). Publications from the category of hard adaptation form a major part of flood risk management research in Jakarta overall (Fig. 5). They focus on specific solutions such as levees, dams, dikes and embankments and analyze or model their protective capacity or suitability for flood protection as well as their vulnerability Mardjono and Setiawan, 2018;Ogie et al., 2016b;Su et al., 2018;Sujono, 2012;Susilo et al., 2019;Suprayogi et al., 2018;Takagi et al., 2016aTakagi et al., , 2017. Others are concerned with water channeling, retention ponds and drainage systems as a means to mitigate flooding Kadri, 2011;Kartolo and Kusumawati, 2017;Mahanani and Chotib, 2018;Mohajit, 2015;Sholichin et al., 2019;Wihaji et al., 2018). Some authors focus exclusively on the Great Garuda project, the central element of the NCICD master plan. Besides outlining the plan and its objectives, a number of studies question its effectiveness regarding flood protection (e.g., Badriana et al., 2017) and its potential impacts on the local environment . Modeled scenarios of flooding with and without the Great Garuda by Yahya Surya et al. (2019) show that the protection wall would slightly increase wave amplitudes, so the authors conclude that the project requires improvements to meet its aim of flood protection. David et al. (2016) point towards the option to complement the sea wall project with ecosystems-based adaptation measures.
Studies outside the category of hard adaptation (Fig. 5) mostly adopt a rather critical perspective on infrastructural solutions in general and on the Great Garuda project in particular. The latter is criticized to be a politically and economically driven, technocratic mega-project that fails to comprehensively address the flood problem (e.g., Colven, 2017;Octavianti and Charles, 2018;Salim et al., 2019). The respective studies do not argue that the project cannot provide any protection from flooding, but they are concerned that it addresses neither land subsidence nor other socio-economic factors contributing to flooding; it therefore does not present a comprehensive and sustainable solution to the flooding problem (e.g., Colven, 2017;Octavianti and Charles, 2019;Salim et al., 2019;Wade, 2019). Infrastructural solutions in general are often portrayed as "technocratic fixes", which do not sufficiently address the hazard's root causes, which are argued to stem from socio-economic and structural-context conditions and vulnerabilities, so the problem persists (Colven, 2017;Padawangi and Douglass, 2015;Wade, 2019).

Soft and hybrid approaches to flood risk management and their evaluation
With a less model-driven and engineering-based perspective, another stream of literature is dedicated to describing, analyzing and/or proposing adaptation strategies and flood governance approaches from a more integrated perspective, i.e., also considering soft measures or hybrid approaches combining soft and hard measures, which, according to the literature, are implemented through both state-led and community-driven initiatives.  (2019a, b) for instance mention the government's involvement in community empowerment and capacity building to facilitate and improve climate change adaptation. Other studies point to the involvement of government at different levels in the dissemination of information about flood risk and adaptation options, which help to raise awareness as well as prepare for and mitigate flooding (Dwirahmadi et al., 2013;Guinness, 2019;Texier, 2008;Ward et al., 2013a). Furthermore, the government's approach of combining hard flood protection infrastructure with the relocation of exposed populations as in the case of, e.g., the Great Garuda could be described as a hybrid approach, which is the subject of discussion in many publications of this stream of literature (e.g., Colven, 2017;Rusdiansyah et al., 2018;Salim et al., 2019;Wade, 2019). Local and community-or NGO-led adaptation initiatives are argued to build on a wealth of context-specific knowledge about and experience with flooding and are often composed of soft as well as hard measures (e.g., Bott et al., 2019;Fitrinitia et al., 2018;Marfai et al., 2015;Padawangi and Douglass, 2015;Purba et al., 2018;Yoga Putra et al., 2019b). The review of community-led adaptation efforts revealed a strong focus on the importance of what can be summarized as social capital for adaptation. A number of studies describe the key role of social networks, which allow for sharing knowledge, experience and best practices Yoga Putra et al., 2019a), which facilitate cooperation and coordination within and among communities, with NGOs, and with universities Goh, 2019;Hellman, 2015;Mathewson, 2018;Padawangi and Douglass, 2015;Yoga Putra et al., 2019a) and which foster mutual support as well as "practices of communing" (Leitner and Sheppard, 2017) such as the pooling of resources (Guinness, 2019;Leitner and Sheppard, 2017;Padawangi and Douglass, 2015;. Social cohesion within networks has also been directly linked to collective action for adaptation . Empirical examples put forward are for instance collective community works -in Indonesia known as gotong royong -for example collective action in river monitoring and the issuance of flood warnings (Bahri and Purwantiasning, 2019;Dwirahmadi et al., 2013;Fitrinitia et al., 2018;Hellman, 2015;Padawangi and Douglass, 2015; ; flood risk mapping (Dwirahmadi et al., 2013); and the establishment of local, community-based institutions to collectively develop and administer saving schemes and funds used for flood response and recovery (Dwirahmadi et al., 2013;Marfai et al., 2015;. Looking at how soft and hybrid measures are being evaluated in the literature, this review finds that many studies exhibit a rather critical perspective on state-led soft adaptation measures. Fitrinitia et al. (2018), Mathewson (2018) and Ward et al. (2011a) elaborate that increased state investments into non-structural soft measures, as described above, are not materializing on the ground. Hellman et al. (2018) refer to this as an "implementation deficit", which they find is facilitated by a lack of reinforcement of laws and regulations. The state's hybrid approach, i.e., the combination of protective infrastructure and relocation of exposed populations, is criticized for not only incomprehensively addressing flood risk (e.g., Colven, 2017;Octavianti and Charles, 2019;Salim et al., 2019;Wade, 2019) but also causing serious negative effects on the environment and local communities (e.g., Dovey et al., 2019;Leitner and Sheppard, 2017;Surtiari et al., 2017;Texier, 2008; ; Ward et al., 2013a) due to the major relocations of informal settlers and the urban poor living in highly exposed areas at banks of the river and coastal areas in the name of flood mitigation (e.g., Goh, 2019; . This state-led approach is described as having accelerated since 2009 (Dovey et al., 2019;Ichwatus Sholihah and Shaojun, 2018), and some authors even claim the evictions to be a bigger threat to those evicted than flooding itself (Dovey et al., 2019;Hellman, 2015;Saridewi and Fauzi, 2019). At the same time, studies highlight the potential of soft adaption measures to create co-benefits with other development objectives. Many studies analyze how the adaptation measures' impacts go far beyond mitigating flood vulnerability as they function as social and financial security, addressing a wide range of vulnerabilities Padawangi and Douglass, 2015; . However, social networks and community cohesion are also described as leading to problematic adaptation effects in some instances. An example is the preservation of exposure to flooding because individuals refuse to leave at-risk areas as they want to stay in the network that gives them a strong sense of belonging and livelihood security (Hellman, 2015;. Besides this, there are authors who criticize that the often positively portrayed community-led adaptation approaches do not lead to optimal and sustainable adaptation, as actions are often implemented in a reactive, ad hoc and rather uncoordinated manner with-out sufficient financial means Ward et al., 2013b).

Discussion
Our review shows that there is a rich, diverse and rapidly growing body of literature analyzing Jakarta's flood problem and identifying as well as evaluating adaptation options. Looking at the discussed drivers of flooding to answer our first research question, we find that one stream of literature, i.e., publications from the fields flood models & flood mapping as well as hard adaptation, predominantly frames flooding as being caused by environmental physical factors and hence as a hazard that can be controlled through engineering solutions and environmental management. Looking at the size of this body of literature measured by the numbers of publications, this can be considered the main stream of flood risk research in Jakarta. A significantly smaller body of literature composed of studies on soft and hybrid adaptation measures, including a focus on social vulnerability reduction and integrated water management, acknowledges the natural and environmental drivers of flooding but also highlights the importance of socio-economic flood drivers, arguing that these drivers have to be considered equally in the consideration and design of adaptation solutions.
Research question two of this study asks which measures for risk management and adaptation are considered in the literature. Our analysis shows that there are largely two separate perspectives on suitable measures for adapting to flooding in Jakarta. One follows a protection approach, identifying predominantly infrastructural measures such as dams, sea walls, water canalization or reservoir constructions as solutions to protect the city from flooding. The second one is not opposed to infrastructural measures but criticizes how they are implemented and demands among other things the inclusion of soft adaptation options to achieve comprehensive flood risk management and hybrid adaptation approaches. This literature argues that without consideration of the root causes of social vulnerability and flood risk, hard infrastructure solutions are bound to be insufficient or even ineffective in the long-run. Likewise, these studies argue that social and environmental effects of hard protection measures need to be considered more stringently and that soft or hybrid approaches are oftentimes better placed to create synergies with other development objectives.
Evaluating these findings from our own perspective we identify the following five points, which we argue might be helpful for advancing the debate and complementing the current perception on the adaptation solution space in Jakarta.
First, the focus on natural drivers of flooding reinforces the perception of flooding being a hazard that can be controlled by technical measures, skewing the perceived solution space towards hard adaptation measures. While it is, of course, very important to assess and understand natural drivers of flooding in Jakarta, other drivers of flood risk need to be considered with the same rigor in order to design effective adaptation options. Hard measures are -and will be -undoubtedly an important part of Jakarta's solution space. Yet, a shortfall in the consideration of the anthropogenic flood drivers yields the risk of designing infrastructural solutions that only address parts of the flood problem, hence risking being less effective.
Second, we find an overall lack of considering future developments in terms of both environmental as well as socioeconomic changes. Surprisingly, the majority of studies that focus on flood modeling does not yet consider future changes in environmental conditions due to climate change. Some studies look at different flood event return periods as a proxy for changing environmental conditions, although with considerable uncertainties remaining (Budiyono et al., 2015;Juliastuti et al., 2018;Kurniyaningrum et al., 2019;Syafalni et al., 2015;Supomo et al., 2017). Only a few of the publications focusing on flood modeling and hard adaptation measures consider future urbanization or sociodemographic changes and their impacts as drivers of flood risk. While some incorporate future changes in land use Fajar Januriyadi et al., 2018;Latief et al., 2018;Mishra et al., 2018;Rafiei Emam et al., 2016;Riyando Moe et al., 2017;Jalilov et al., 2018;Sutrisno, 2011;Takagi et al., 2016b;Vollmer et al., 2015Vollmer et al., , 2016Ward et al., , 2013c, no study in the sample considers future changes in exposure due to, e.g., population growth or urban development. Similarly, the publications on hard flood protection measures mostly neglect future climatic, demographic, socio-economic and land-use changes when assessing the effectiveness of existing or suggested infrastructural measures for flood protection. Some authors use return periods of rainfall events (Mantasa Salve Prastica, 2018) or floods (Ajiwibowo, 2018;Indrawati et al., 2018;, although without referring to potential future changes. An exception  evaluates the effectiveness of planned coastal dikes using flood and subsidence projections until the year 2050. The implications of this shortcoming for the solution space are rather weighty: developed solutions -hard, soft and hybrid -that are lacking the consideration of future developments have an inherent risk of not being sustainable and effective in the long run. While they might address current challenges very well, there is a risk that dynamic changes in environmental and/or socio-economic aspects will impact their effectiveness in the future.
Third, most publications that assess and evaluate the effectiveness of hard measures for flood management in Jakarta do not consider social aspects such as the measures' impacts on social vulnerability or the acceptance of the analyzed measures. This can be an important shortcoming when evaluating the overall adequacy and success of a hard adaptation measure. For instance, resettlements, which are often a precondition for the implementation of hard adaptation measures, significantly influence communities' livelihood opportunities and social structures Ichwatus Sholihah and Shaojun, 2018;Surtiari et al., 2017). Furthermore, authors claim that soft aspects such as risk perception and awareness, risk communication (e.g., , behavioral and cultural factors (e.g., Bott et al., 2019;Yoga Putra et al., 2019b), collective action as well as participatory planning (e.g., , coordination capacities (e.g., Marfai et al., 2015;Padawangi and Douglass, 2015), and law enforcement (e.g., Akmalah and Grigg, 2011) are inherently intertwined with the success of mitigation and adaptation efforts in the long run, hence calling for rigorous consideration. The neglect of aspects such as social acceptance as well as impacts of infrastructural flood protection on local communities influences the solution space in that hard adaptation measures will be considered for flood risk management despite their potential negative impacts on social vulnerabilities. Accordingly, the solution space contains measures that are beneficial for some groups of people while representing a threat to other groups. While this is already alluded to in the assessed literature (e.g., , this seems to have had only marginal influence on flood risk research in Jakarta until today.
Fourth, studies focusing on soft and hybrid adaptation measures converge in their critique of technocratic approaches and provide a broad variety of needs with respect to improving flood risk management, although without providing concrete recommendations for how to achieve them. Depending on the perspective of the author(s), different approaches are advocated for, for instance more integrated and hybrid adaptation approaches (e.g., Akmalah and Grigg, 2011;David et al., 2016;Jalilov et al., 2018) and awareness raising and dissemination of risk and response information to foster behavioral change within the public and authorities and among urban planners (e.g., Akmalah and Grigg, 2011;Amri et al., 2017;Esteban et al., 2017;Goh, 2019;Marfai et al., 2015;Ward et al., 2013b). Others call for regional approaches to land use and urban planning (e.g., integration of upstream and downstream, Jakarta and Jabodetabek 1 ; Asdak et al., 2018;Firman et al., 2011;Goh, 2019;Mathewson, 2018;Noviandi et al., 2017) and more integrated legal and institutional frameworks as well as strong institutional bodies for disaster risk reduction and climate change adaptation Asdak et al., 2018;Firman et al., 2011;Octavianti and Charles, 2018;Ward et al., 2013b). Moreover, studies highlight a need for stronger law enforcement (e.g., Akmalah and Grigg, 2011) and increased community-participation in risk management planning and decision-making (Goh, 2019) as well as in resettlements Texier, 2008). However, the publications provide few concrete recommendations on how to achieve these goals: how can different adaptation measures be adopted for a balanced hybrid adaptation approach? How can the legal and institutional setup be improved, which would most likely include altering current political structures and decision-making processes? How can flood risk policies be integrated with the wider development agenda? How can participatory flood risk management be facilitated? There are only very few publications (e.g., Asdak et al., 2018;Firman et al., 2011; providing slightly more detailed information on the implementation and feasibility of suggested measures. Hence, while studies on soft and hybrid measures add valuable knowledge and potential options to the solution space, their lack of actionable recommendations limits their utility. They are more difficult to consider in actual adaptation planning -especially in comparison to hard adaptation measures which often have clear requirements and quantitative assessments of their use.
Fifth, there is a lack of studies that compare multiple adaptation solutions. Apart from a few exceptions (e.g., Fitrinitia et al., 2018) this lack applies to comparisons between different infrastructural measures as well as between hard and soft measures. This is surprising against the background of the rich diversity of assessments of single measures -be it hard adaptation options such as levees, dams, dikes and embankments Mardjono and Setiawan, 2018;Ogie et al., 2016b;Su et al., 2018;Sujono, 2012;Susilo et al., 2019;Suprayogi et al., 2018;Takagi et al., 2016aTakagi et al., , 2017 and water channeling, retention ponds and drainage systems Kadri, 2011;Kartolo and Kusumawati, 2017;Mahanani and Chotib, 2018;Mohajit, 2015;Sholichin et al., 2019;Wihaji et al., 2018) or soft adaptation measures like mutual support through social networks (Guinness, 2019;Leitner and Sheppard, 2017;Padawangi and Douglass, 2015;, collective action in, e.g., river monitoring and early warning (Bahri and Purwantiasning, 2019;Dwirahmadi et al., 2013;Fitrinitia et al., 2018;Hellman, 2015;Padawangi and Douglass, 2015; , or self-organization of rescue groups for flood response and recovery (Dwirahmadi et al., 2013;Marfai et al., 2015;. For the solution space, this aspect can be considered the most important gap. The absence of comparisons of different measures leaves decision-makers without scientific guidance in understanding the advantages and disadvantages of one adaptation solution over the another. Without comparatively considering the effectiveness, social and environmental impacts, and feasibility of different measures, it is very difficult to identify the "best" adaption options and combine them into sustainable adaptation pathways. While our findings and lessons from Jakarta cannot easily be transferred one-to-one to other risk contexts, recent global assessments (Oppenheimer et al., 2019) suggest that many cities might be facing similar patterns to the ones identified for Jakarta, calling for follow-up research. We therefore hope that the perspectives and questions raised in this paper are useful to inspiring studies on the solution spaces in other high-risk settings.
Appendix A: Results from the structured literature search In total, 339 publications were identified by the keyword search. A total of 14 proceedings and 2 publications for which no full-text versions were available have been excluded. The below list shows the remaining 323 publications. All references that are marked with an asterisk were included in the detailed content analysis and are cited in the paper. Anugrahadi, A., Yuda, H. F., and Triany, N.: The Jakarta Flood Disaster Mapping January 2020, Int. J. Adv. Sci. Technol., 29, 602-605, 2020. Arifin, H. S. and Nakagoshi, N.: Landscape ecology and urban biodiversity in tropical Indonesian cities, Landsc. Ecol. Eng., 7, 33-43, https://doi.org/10.1007/s11355-010-0145-9, 2011.
Author contributions. MG and MW designed the study. MW performed the literature analysis. MW and MG drafted the manuscript.
Competing interests. The authors declare that they have no conflict of interest.
Disclaimer. Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Special issue statement. This article is part of the special issue "Future risk and adaptation in coastal cities". It is not associated with a conference.