The theoretical framework that informs this review is rooted in the linear model of the policy process and the interactive model of policy implementation to explore the landscape of climate services for disaster risk management. The rationale for adopting this framework stems from the assumption that climate services, as part of disaster risk management and climate change adaptation, are broadly enshrined in climate change policy. Hence, theoretically and practically, climate services for disaster risk management are reflected in these policy processes. Grounded in the policy process perspective, the framework provides a lens (Larsen & Adu, 2021; Grant & Osanloo, 2014) to capture policy-oriented priority areas in the field of climate services for disaster risk management. It also provides a perspective from which to identify the actors and stakeholders involved, as well as the dynamics related to the field.

Policies emerge as a result of social needs or problems (Kulaç & Özgür, 2017) . In an attempt to provide a comprehensive definition of policy, Birkland (2020) synthesized a set of attributes of policy such as problem-solving oriented, made in the public interest, interpreted, and implemented by public and private actors; goal-oriented or oriented state, ultimately made by government or via the interplay of government and nongovernmental actors. These elements are considered substantially instrumental to policymaking, which is widely shared by many scholars in the domain.

The linear model, widely known as a staged, sequenced model, represents one of the most traditional theories for understanding policy formulation and implementation (Sutton, 1999) , and remains a useful starting point for understanding policy-making processes (Kulaç & Özgür, 2017) . According to this model, the policy process is allegedly characterized by a well-ordered arrangement of steps, translating a rational trajectory from problem identification to solution implementation, including agenda setting, which refers to the identification and framing of potential issues to be addressed by decision-makers, decision phase defining the formulation of the policy, and implementation phase in which policy is put in action and evaluation in which the impacts and/or performance of the implemented policy is assessed to determine its success or failure (Grindle, 1991; Gültekin, 2014) . This lingering model provides an objective analysis of options and a clear separation between policy and implementation (Thomas & Grindle, 1990; Sutton, 1999) . In practice, a proposed reform gets on the agenda for government action, a decision is made on the proposal, and a new policy or institutional arrangement is implemented, either successfully or unsuccessfully (Grindle, 1991: p. 122) . Very often, there is a strong focus on policy formulation and policy decision-making, while the implementation phase receives less attention. However, since the outcomes of policy initiatives are often determined by a lack of political will, poor management, or shortage of resources (Sutton, 1999) , but not necessarily depending on the policy or decision itself, a proper analysis of a policy process needs to include policy implementation and feedback loops between implementation and policy making.

Although linear policy models have helped to move attention from policy formulation and political decision-making to implementation, they are, according to Sutton (1999) , oversimplifying and overlooking the complexity of the policy processes, which are not necessarily linear, and steps within the process may overlap or occur simultaneously. Furthermore, linear models tend to overlook the influence of power dynamics and the diversity of actors influencing agenda-setting, decision-making, and implementation by focusing on the logic and rationality of steps.

In contrast to the traditional linear model, in the interactive models of policy processes, several instrumental elements influence policy outcomes from agenda-setting to implementation. In a seminal work, Sutton screened the contributions of several disciplines, including political science, sociology, anthropology, international relations, and management studies, to provide an overview of interactive models of policy processes. Against the conventional understanding of policy as a linear process, they argued that policy is not simply about analyzing a problem, appraising possible responses, selecting a response rationally, implementing the chosen action, and evaluating the outcome (Clay & Schaffer, 1984) . Instead, they suggested that the policy process as more complex and chaotic, shaped by a multitude of purposes and accidents. This perspective challenges the traditional ‘practitioner perspective’ which views policies as ‘objective entities’ that are the result of decisions made by some rational authority and which reorganize bureaucratic action to solve particular ‘problems’ and produce a ‘known’ (or desired) outcome (Clay & Schaffer, 1984) . In essence, this perspective emphasizes the complexity and unpredictability of policy-making, highlighting that it is not merely a matter of rational implementation of decisions through selected strategies. Such processes can be understood by using concepts such as policy narratives, policy communities, and change management developed by different scientific disciplines. Policy processes should be both dynamic and processual. Different steps in a given policy process are likely to be reshaped or affected by the interests and reactions of various actors, thus affecting the outcomes (Thomas & Grindle, 1990; Grindle, 1991) . Considering implementation as the most crucial component of the process, the interactive model places it at the center of the policy process analysis.

The science-policy interface in the policy processes:

The process of formulating policies is frequently intricate; actors involved in policy-making navigate through an interplay of a multitude of stakeholders and their interests, which often clash (Burton et al., 2019) . In this context, it becomes crucial for actors, especially researchers and policymakers, to navigate and learn how to integrate their perspectives into the policy process better. This has led to the emergence of the concept of the science-policy interface (SPIs) (Burton et al., 2019) , which are defined as social processes that encompass relations between scientists and other actors in the policy process, and which allow for exchanges, co-evolution, and joint construction of knowledge intending to enrich decision-making (Van den Hove, 2007: p. 1) and represent space where scientific knowledge converges with decision-making processes (Gluckman, Bardsley, & Kaiser, 2021) in the quest to create a common ground to collaborate to resolve practical problems in society (Afrin & Rahman, 2022) . Thus, the emergence of studies on science-policy interface issues and the renewed interest among scholars and decision-makers stems from long-lasting philosophical, sociological, and political questions concerning the roles of science in society (Van den Hove, 2007) , where the cleavage between scientists and policy-makers is pointed out as the triggering factor of failure in science-informed decision-making. In this context, SPI represents a central element of the policy process, which has various implications for different stages and knowledge production. Thus, at the science-policy interface, Gluckman (2018) identified four categories of roles: knowledge generation, knowledge synthesis, and knowledge brokerage. Knowledge production or generation has a lot of interest among scholars and practices out of the need to develop socially relevant knowledge. Based on Gibbons’s (1994) comprehensive work, two main approaches are widely acknowledged. Mode 1 knowledge production is the traditional and well-established model for generating knowledge within academic institutions, which is characterized by a linear, disciplinary-driven, and academically oriented approach to knowledge creation. Mode 2 knowledge production, on the other hand, is a more collaborative, context-driven, and transdisciplinary approach to knowledge generation. Widely considered as a mechanism for transferring research evidence into policy and practice (Ward, House, & Hamer, 2009) , at its simplest level, knowledge brokering simply facilitates the flow of information from one side to another (Shaxson & Gwyn, 2010) , as well as transforming it (Meyer, 2010; Meyer & Brun, 2023) . Thus, the actors playing these roles are mainly scientists, policymakers, boundary organizations, and other stakeholders (e.g., beneficiaries of a project) (Burton et al., 2019) . Thus, in light of the complexity of knowledge production and decision-making, Neßhöver et al. (2013) identified three propositions that are instrumental to the effective science-policy interface, including framing research and policy jointly, promoting inter and transdisciplinary research, and multi-domain working groups that include both scientists and policymakers from various fields and sectors, put in place structures, and incentive schemes that support interactive dialogue in the long-term. Whereas these elements seem to address practical issues arising at the interfaces of science and policy, Van den Hove (2007) , identified a range of points of intersections between science and policy, which induce some theoretical considerations that need to be addressed, which are termed science-policy interfaces defined earlier by the author. In this regard, Van den Hove (2007) identified 15 normative requirements guiding the design, implementation, and evaluation of science-policy interfaces ( Table 1 ).

Theoretical Problems at the Science-Policy Intersection:

Integration of Knowledge: The challenge of integrating scientific knowledge with policy-making, considering the different languages and methodologies used in each domain. Complexity and Uncertainty: Policies often deal with complex systems where scientific knowledge is uncertain, incomplete, or contested. Diverse Values and Interests: The intersection of science and policy involves multiple stakeholders with varying values, interests, and power dynamics.

Normative Requirements for Science-Policy Interfaces:

Credibility: SPIs must ensure the scientific credibility of the information provided to policymakers. Salience: The information must be relevant and timely to the policy issues at hand. Legitimacy: SPIs should be perceived as legitimate by all stakeholders; which involves fairness and ethical considerations.

Van den Hove emphasizes that SPIs are not just about facilitating the flow of information from science to policy, but also about co-evolution and joint knowledge construction. This requires a dynamic ecosystem of processes, actors, and organizational arrangements designed to effectively address complex policy problems. The author’s work suggests that addressing these theoretical challenges and adhering to the normative requirements can enhance the design,

implementation, and evaluation of SPIs, leading to more informed and effective governance and policy processes. Consequently, as these later answer how science-policy interfaces operate, it is arguably suggested that failing to address them would cost the effectiveness of policy processes and governance.

Given the cross-cutting aspect of the subject of inquiry and out of the need to ensure a rich and comprehensive review, the data collection was performed by

relying on multidisciplinary sources as recommended in the Cochrane Handbook (Bramer et al., 2017) , including Research4life, Tylor and Francis online, Google Scholar, ScienceDirect, and JSTOR using a search string ( Figure 5 ) that refined search results to retrieve articles that were most relevant to the review topic. Records were directly saved into Zotero through the web importer extension in Google Chrome.

To collect records not captured via databases, a snowball search strategy was performed in ResearchRabbit to identify records related to the initial one imported by searching for literature citing the initial hit and relevant papers cited in these texts.

Based on the inclusion and exclusion criteria ( Table 2 ), a three-stage screening was performed to identify the final sample from 2168 records. 1) duplicates were removed and titles were scanned against eligibility criteria. 2) abstracts were screened, and 3) the full content was evaluated. Records collected from databases (1460) and records collected through ResearchRabbit (998) resulted in an initial sample of 2168. Within this sample of 2168 records, by processing the first screening, a substantial number of papers were eliminated because of a minor focus on climate services for disaster risk management. Within the sample, some of the papers combined literature, interviews, and questionnaires; these papers were then considered as satisfying inclusion criteria because they did not rely solely on a literature review. In the final step of the screening, 46 papers (32 from the set of five databases and 14 from ResearchRabbit) were removed out of 288, which resulted in a sample record of 242 papers included in the review.

To proceed with data processing ( Figure 6 ), the records were consolidated into a single database using Zotero 6.0.26. Duplicates were removed and initial filtering was performed. The data were then exported to Microsoft Excel for further sorting and organization, facilitating the continued screening and generation of graphs and tables. Data analysis was conducted in two stages: descriptive analysis and thematic/content analysis (Tranfield et al., 2003) . The former provides a comprehensive account of the sample included in the review, whereas the latter offers a qualitative thematic/content analysis (Creswell & Creswell, 2018) , drawing on key conceptual patterns emerging from the reviewed literature. Thematic/content analysis was used to generate conceptual patterns (Braun et al., 2019; Braun & Clarke, 2013; Flick, 2014) , and eligible records were imported into the ATLAS.ti 23 software for analysis. This process facilitated the identification of key themes and patterns in the data. After organizing the data, two coding processes were employed including open coding and axial coding (Adu, 2019; Auerbach & Silverstein, 2003; Saldaña, 2016) . At the first stage of open coding, relevant data were condensed to generate initial units of analysis (codes). The axial coding completed this first stage, which consisted of grouping and connecting codes and placing them in new categories. This allowed to come up with sub-priority areas.

As part of the process of data collection, screening, and extraction, a specific checklist ( Table 3 ) was used. This checklist contains meta-data, based on which relevant information was identified and captured for the analysis. By following this procedure, each article was reviewed according to this meta-data checklist.

The ROBIS (McGuinness & Higgins, 2020) ( https://www.riskofbias.info/ ) tool

*CS: Climate Services, *WCS: Weather and Climate Services, *CIS: Climate Information Services, *MS: Meteorological Services, *DRM: Disaster Risk Management, *DRR: Disaster Risk Reduction.

was applied in this review to assess its quality. This included 1) Relevance Assessment, which evaluated the relevance of the studies with regard to the objective of the review. 2) Identification of concerns with the review process, which identified potential issues or concerns with the process of the review itself, particularly regarding the PRISMA methodology. 3) Bias Risk Judgement, the final step that involved making a judgment about the overall risk of bias and providing an overall assessment of the reliability of the review (Whiting et al., 2016) . Within these three stages, four key domains are examined including (i) Study Eligibility Criteria, which assessed the criteria used to select studies for inclusion ( Table 2 ) and ensured that the selection process is unbiased and consistent. (ii) Identification and Selection of Studies, evaluating the methods used to identify and select studies for inclusion in the review. It ensured that all relevant studies had been considered and that the selection process was transparent and replicable. (iii) Data Collection and Study Appraisal, which examined the methods used for data collection and the appraisal of studies by ensuring that the data collected is reliable and that the quality of the studies included in the review has been appropriately evaluated. (iv) Synthesis and Findings, the final domain that evaluated the way findings have been synthesized to ensure that the conclusions drawn are backed by the data and are presented clearly and understandably. The assessment was conducted in five batches, with the risk of bias for each batch detailed in Appendixes 1-5. Thus, after the assessment, most of the studies included in the review were found to have a low to moderate risk of bias, indicating a high level of reliability of the review.

From an initial sample record of 2458, 2216 documents (including duplicates, and irrelevant documents) were removed through the screening process, resulting in a final sample of 242 based on eligibility criteria ( Figure 7 ). The reviewed record of 242 studies included 201 (83%), journal articles, 33 reports (13 %), and 9 (4%) book sections, all related to climate services and disaster risk management ( Figure 8 ).

Over the ten years (2013-2023) concerned by this review, literature production

has substantially increased. Temporal evolution ( Figure 9 ) portrays that the highest production is made in 2020, followed by 2021, 2019, and 2017, while the lowest scientific literature production is identified in 2013 (taking into account the fact that this work concerned the early publication of 2023).

Among the literature surveyed, Climate Services has the highest number of publications (n = 48) followed by the International Journal of Disaster Risk Reduction (n = 14) and Climate Risk Management (n = 11) summarized in Appendix 6. Concerning reports produced by institutions dealing with climate services and disaster risk management, the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) has the highest record (n = 7), followed by the World Meteorological Organization (WMO) and the World Bank Appendix 7. Appendix 8 summarizes publishers of books sections included in the review where Springer totalizes the highest (n = 6).

Looking at the geographical zones, the highest percentage of publications is addressing Africa at 36.4%, followed by Europe at 24.8%, Asia at 16.1%, and Oceania at 1.2%. Some studies covered more than one geographical zone as highlighted in ( Figure 10 ). The last category concerns studies that have focused on a global scale with 10.7%.

The large array of scientific literature on climate services for disaster risk management and the wide coverage across fields including agriculture, water resources management, and urban development demonstrates the increasing interest in this field. Over the past ten years, scientific production has been abounding with increasing interest from scholars of various backgrounds and multiple framing. By systematically reviewing literature in the timespan of 2013-2023, this work mapped scholarly research priorities in the field of climate services for disaster risk management. This section starts by presenting the climate-induced disasters in the literature and the typology of climate services. The second section delves into the analysis of identified policy-process-related priority areas in the literature.

In the current review, flood appears prominently as one of the most frequent climate-induced disasters. In parallel, drought emerges as another prevalent disaster ( Figure 11 ). These findings concur with confident evidence indicating these hydro-climatic extreme events have increased in magnitude at the global scale due to climate change (IPCC, 2022) . In Africa, over the last twenty years (2002-2021), the Centre for Research on the Epidemiology of Disasters (CRED) indicated the occurrence of 793 floods and 137 droughts (CRED, 2022) which impacted more than 52 million people in 2022 Global Center on Adaptation (GCA, 2022) . Furthermore, food insecurity appears at the third position in the explored literature as a consequential outcome of the interplay of the two first

(CRED, 2022) . In Asia, recent evidence indicates the occurrence of 81 climate-related disasters in 2022, of which over 83% were floods and storms (WMO, 2022) . As a widespread situation witnessed globally, particularly in the south, drought occurrence is worsening, leading to more impacts and damages (Aadhar & Mishra, 2021; Ha, Uereyen, & Kuenzer, 2023) and in the Monsoon regions (Kim et al., 2020) . Other climate-related, events found in the literature include cyclones and storms ( Figure 11 ). In Europe, more frequent and hotter heat weaves are expected, while in some parts drought will increase and flooding in other zones (Guerreiro et al., 2018) . The occurrence and impacts of drought and flood are well documented as appeared in Lehner et al. (2006) in which these events have been projected to increase.

Climate services have rapidly and progressively expanded as a field of interest from various actors depending on their logic and research interests (Vaughan & Dessai, 2014) . The urgent need to mitigate and face climate-induced disasters has sparked attention to the role climate services play in the DRR cycle in managing this societal risk (Hewitt & Stone, 2021) . This has led to the need and development of various typologies of climate services (Bessembinder et al., 2019) for the sake of decision-making for a wide range of end-users. In the quest to meet the needs of both supply and demand sides a wide variety of climate services are developed (Visscher et al., 2020) . Throughout the reviewed literature, the typology of climate services for disaster risk reduction encompasses a range of essential components that aim to enhance preparedness, response, and resilience in the face of climate-induced disasters. These services are broadly categorized into three key areas: climate information, impact assessment, and decision support. Climate services provide reliable and timely data on weather patterns, climate variability, and long-term climate projections. This includes meteorological data, such as temperature and rainfall patterns, as well as climate modelling and forecasting. Impact assessment services analyze the potential consequences of climate-related hazards on various sectors, including water, infrastructure, agriculture, health, and ecosystems. These assessments help in understanding vulnerability and risk exposure, facilitating the development of targeted disaster reduction strategies. Decision support services utilize climate information and impact assessments to inform policy and decision-making processes. They involve the development of user-friendly tools, such as risk mapping, and early warning systems, which enable stakeholders to make informed choices and take proactive measures in disaster risk reduction. By combining these different types of climate services at different scales, policymakers, communities, and individuals can better anticipate, prepare for, and respond to climate-induced disasters, ultimately minimizing their impacts and building resilience in a changing climate. In summary, climate services for disaster risk management identified within the literature include meteorological services, weather information, warnings, forecasts (seasonal climate forecast, sub-seasonal to seasonal forecast…), early warning systems (flood early warning, cyclone early warning), and Impact Based Warning System ( Figure 12 ).

The content analysis of the reviewed literature identified three major and interconnected policy-oriented priority areas including (i) development of climate services, (ii) use-adoption-uptake of climate services, and (iii) evaluation of climate services ( Figure 13 ). These priorities are part of disaster risk management and climate change adaptation, which both fall into the realm of climate change policy as part of encouraging science-based and climate-informed policy development (Larosa & Mysiak, 2020) . Broadly taken within climate change policy, adaptation is considered not a single activity, but rather a continuous process and ongoing policy cycle to understand risk, develop adaptation options, and assess them (Leal Filho & Jaco, 2020) . In the same vein, authors argued for the concept of adaptation services, which complement climate services in the context of risk management. This standpoint on adaptation services concurs with the definition of climate services touted as providing climate information in a way that assists decision-making (WMO, 2014) . Similarly, climate services support all phases of disaster risk reduction (Street et al., 2019) .

Priority 1: Development of Climate Services

The development of climate services offers reliable decision-support in efforts towards climate change adaptation (Leal Filho & Jaco, 2020) . In parallel, as part of their role in decision-making, climate services have been proven to support a better understanding and anticipation of disaster risk. In the explored literature, the Development of Climate services revolves around various aspects namely (1 production: co-production; users-providers engagement), (2 dissemination:

communication; community-based approach; roles of non-state institutions (NGOs and Private sector)), (3 barriers to development: institutional; technical; capacity etc.). To foster effective dissemination, in many projects, a community-based approach is used going beyond a simple top-down approach by involving the community at the heart of the dissemination process. This has brought non-state actors on board since it falls in the realm of their developmental interventions. Non-governmental organizations have made significant efforts in this regard (Kirbyshire & Wilkinson, 2018; Jones, Harvey, & Godfrey-Wood, 2016) in the implementation of climate services programs, bringing a new breath of life to the field of climate services for disaster risk management. By recognizing the widespread impacts of climate change and variability, which became an existential threat to humanity as well as the humanitarian perspective to addressing societal needs to face its impacts, they recognized the need to build a strong and effective climate services system (Coughlan de Perez & Mason, 2014) , there is a renewed interest in funding projects to support countries in their effort to build a sustainable climate services system from NGOs and non-state actors, especially in sub-Saharan Africa

(Harvey et al., 2019) . In this context, they are considered key actors, given the recognition of a “missing middle” of two-way brokering, translation, and validation where NGOs and other non-state actors can support innovation and experimentation, particularly in two-way forms of engagement with local actors and knowledge holders (Knudson & Guido, 2019) .

Priority 2: Use-Adoption-Uptake of Climate Services

The second priority area extensively documented in the explored literature represents the utilization of climate services. This multifaceted aspect encompasses the use, adoption, and uptake of climate services. Thus, beyond methodological considerations, this conceptual variation focuses mainly on understanding two key elements: enablers and barriers (Antwi-Agyei, Dougill, & Abaidoo, 2021) . Interestingly, these elements can sometimes be found within a single factor, which can be both an enabler and a barrier. Climate services are produced to support users’ decision-making processes (Vaughan & Dessai, 2014) , which suggests that tailoring these services to meet users’ needs significantly shapes their adoption (Buckland & Campbell, 2021) . This emphasizes the positive influence of co-design in climate services on their use and uptake (Buckland & Campbell, 2021) . Several factors influence the adoption of climate services. These include gender (Carr & Onzere, 2017) , awareness and perception of climate services (Zongo et al., 2015) , institutional membership (West, Daly, & Yanda, 2018) , institutional design (Vedeld et al., 2021) , and institutional coordination and communication (West, Daly, & Yanda, 2018) . These factors are closely linked to the concepts of salience, legitimacy, and credibility (West, Daly & Yanda, 2018) that influence the usefulness of climate services to end users (Cash et al., 2002; Clifford, Travis, & Nordgren, 2020) . To achieve salience, legitimacy, and credibility, it is crucial to managing boundaries between disciplines, scales, and forms of knowledge (Cash et al., 2002) . Furthermore, several factors are related to the roles of non-state institutions or actors in providing and creating favorable conditions to enhance the usability of climate services. One such aspect is the agency of non-state actors, particularly the emergence of NGOs. These organizations act as brokers in the dissemination of climate services, creating an enabling environment that enhances their utilization (Harvey et al., 2019) . Consequently, as it is discussed in the literature, the enablers and barriers are arguably attributed to the types of approaches adopted in the process of climate services implementation. In most cases, barriers are considered to reflect the supply-driven approach in which users do not have significant inputs in the process. In contrast, the adoption of a bottom-up, participatory, or co-production approach is reflected in the enablers, since the users’ needs are at the heart of the approach in addition to the continuous interactions over the process.

Priority 3: Evaluation of Climate Services

The emphasis on the evaluation of climate services is widely explored in the dedicated literature (Vaughan et al., 2019) . Indeed, the focus of the literature is on the assessment of climate services as an integrated element of the climate services value chain. Scholars identified this priority out of the need to make the processes more effective by evaluating various dimensions of climate services specifically 1 Assessment of use-adoption-uptake (Soares & Dessai, 2015; Amwata, Omondi, & Kituyi, 2018) , 2 Assessment of impact, socioeconomic assessment; willingness to pay for climate services (Ouédraogo et al., 2018; Paparrizos et al., 2021) , 3 technical performance assessment, 4 Assessment of needs (Vincent et al., 2017) , 5 developing assessment methodology (Vogel, Letson, & Herrick, 2017; Tall et al., 2018) , 6 Measuring co-production (Visman et al., 2022) , 7 Business model of climate services for fostering services’ adoption and boosting the market for climate innovation (Larosa & Mysiak, 2020) . Developing assessment methodologies is becoming a critical element, given the need to have robust metrics and indicators for effective evaluation of climate services and ensuring evaluation constitutes a cornerstone of the value chain (Dinku, 2018) . The importance of this aspect is emerging towards the co-evaluation approach as it appeared in priorities as a shift for a more comprehensive approach that involved the entire value chain, the actors involved, and their logic in the field (Lugen, 2020) .

The Emergence of Co-production Approach

Throughout the reviewed literature, co-production appears as an emerging theme in the climate services processes, which translates the response to the call for a shift in the focus of climate services (Daniels et al., 2020) . The traditional approach to the provision of climate services has been dominated by a supply-driven mechanism, which involves essentially a one-way direction in the process (ICPAC, 2021) . Thus, despite numerous calls for useful climate science information to support policy, most endeavors were consistent with this linear model (McNie, 2013) . The supply-driven or top-down approach is mostly characterized by the availability of data technology, and production, rather than by specific user needs (Hewitt, Mason, & Walland, 2012; Taylor, Dessai, & de Bruin, 2015) . However, the approach has been criticized for not adequately considering the needs and capacities of end-users (Sultan et al., 2020) . It can lead to the production of climate services that are not relevant or usable for decision-makers at the local level. This is particularly problematic in developing countries, where local capacities and needs may be very different from those assumed by global models. In contrast, a demand-driven, bottom-up approach starts with the needs of the users and builds climate services to meet those needs. This approach can lead to more relevant and usable climate services, though it requires a deep understanding of user needs and the capacity to meet those needs, which can be challenging to achieve in practice (Lourenço et al., 2016) , which does not slow down its use. Thus, this approach has been continuously improved and over the last decades, the co-production (which focuses on demand-driven) approach has emerged and widely adopted in implementing climate services programs (Jacobs & Street, 2020) as both the antithesis and the solution to the linear model (Tangney, 2022) . Rooted in the theory and practice of Knowledge co-production involves the joint creation of knowledge by experts and non-experts, ensuring that diverse perspectives and experiences contribute to the understanding and resolution of complex issues (Jasanoff, 2004) . The process emphasizes the collaborative and interactive process of producing knowledge between scientists, policymakers, stakeholders, and the public toward addressing complex societal issues (Cash et al., 2003; Jasanoff, 2004) . Applied to the climate services process, it brings together different knowledge sources and experiences to jointly develop new, combined, and relevant knowledge and systems to enable its intended use in specific decision-making contexts (ICPAC, 2021: p. 21) . The process of co-producing knowledge and building the necessary skills and capacity of different user groups, both to guide the production and tailoring of climate information (to meet context-specific needs) and to be able to apply that information (Leal Filho, 2015: p. 15) . Indeed, to be more efficient and effective (Vaughan & Dessai, 2014) , the climate services field needs to address the approach to production (Weichselgartner & Arheimer, 2019) , which involves the shift to a co-production paradigm (Bremer et al., 2019; Katharine Vincent et al., 2018) . Co-production can be highly useful for enabling dialogue and collaborating across the value chain, helping create services based on credible, salient, and legitimate knowledge (Hewitt & Stone, 2021) . This revealed its critical importance as an impetus towards the effective development of climate services for disaster risk management. Furthermore, co-production emerged as a crosscutting aspect over the three main priority areas identified in the current work namely, (Development; Use-adoption-uptake and Evaluation of climate services). Indeed, the process lies in the effective users’ engagement (Daly & Dessai, 2018; Golding et al., 2017) , and that of providers, which may be considered as the two faces of the same coin. Since climate services should be decision-driven, reflecting end-users’ needs, their engagement in the process of production remains critical (Baulenas et al., 2023) . However, Bremer et al. (2019) are against the narrowed acceptance of the process, but rather frame it as multi-faceted. Most importantly, co-production has implications for the evaluation of climate services. Indeed, since the users and providers are interactively engaged in the production, so does the evaluation, with the emerging concept and approach of co-evaluation of climate services (Delpiazzo et al., 2022) , recognized as important for co-development (Tarchiani et al., 2021) framing (Bremer et al., 2021) . An example of a practical and proven application of the co-production approach is described in the Tandem Framework ( Daniels et al., 2019, 2020) . In this approach, to ensure the effectiveness and long-term sustainability of climate services, it cross-pollinates with the main aspects of the value chain, which are also found in the three major priority areas in the present review. Applied throughout the different aspects the approach involves for instance co-design (Santos & Swartling, 2020) , and when it comes to evaluation, considered in terms of co-evaluation (Street et al., 2019; Delpiazzo et al., 2022) . In this regard, co-production brings a new perspective to the (co)-evaluation of climate services providing a methodological approach to assess the impacts of the co-production of climate services (Englund, André, & Swartling, 2022) with a focus on the users’ feedbacks.

The supply-driven and co-production approaches vs. linear and Interactive Policy process? A reflection on the implications of the theoretical framework

The interactive model of policy implementation and the linear model of the policy process are combined in the theoretical framework used in this study. This theoretical lens offers a prism through which the three main areas of specified priority are analyzed. Thus, as revealed in the analysis, the field of climate services has historically been dominated by the supply-driven approach, which is typified by a top-down method of production. This method appears to be consistent with the linear model of the policy process, in which policies or services are produced by experts and then transferred down for execution. However, this approach has drawbacks, notably in terms of climate services. The top-down character of the supply-driven strategy might result in misalignment and a certain level of discrepancies between the services offered and the real needs of end-users. This imbalance presents a high likelihood of leading to underutilization of climate services or slowing down the uptake process, which limits their potential to face and manage disaster risks. As a response to these limitations, the co-production approach has evolved and is gaining momentum in this field. Co-production entails active collaboration between service suppliers and end users throughout the entire process, from development to assessment.

This strategy remains therefore in line with the interactive model of policy implementation, which emphasizes the value of iterative learning and feedback loops. Thus, by customizing climate services to end users’ unique needs and settings, co-production ensures that these services are more relevant and widely adopted. However, co-production is not without challenges. Indeed, it requires a significant commitment of time and resources as well as a change in the balance of power such that end users are more actively involved in the process.

Thus, this emerging approach provides at first glance, a theoretical pathway to understanding the interplay of different actors involved in the process and institutional layering, which have major implications for the development of services and their integration into disaster risk management. From a practical perspective, co-production provides a concrete enabling environment for actors in the process, to better engage and overcome barriers and discrepancies.