The theme for World Meteorological Day 2022 (23 March) was “Early Warning and Early Action – Hydrometeorological and Climate Information for Disaster Risk Reduction”, which emphasises the vital importance of information generation and sharing to minimise the risks from hydrometeorological extremes. Further, the United Nations secretary general announced a major initiative, to be delivered via COP 27 (UN Climate Conference): “everyone on Earth should be protected by early warning systems against extreme weather and climate change within the next five years”. These policy initiatives indicate the growing need for new information and knowledge relating to risks arising directly from hazard but also from the complex interactions with exposure and vulnerability (IPCC defined risk = hazard × exposure × vulnerability / capacity to cope; see details in Cardona et al., 2012). Although our understanding of hydrological extremes, such as floods, has evolved in recent decades as we view them through the lens of hydro-complexity (Kosow et al., 2022). However, floods remain a “wicked” problem and are becoming more destructive with ecological, social, and economic impacts (i.e. source of water pollution, damages to wastewater and irrigation system, excessive erosion damaging riverbank settlements; see details in Kosow et al., 2022; Hannah et al., 2020). In mountainous regions floods are becoming more unpredictable and destructive in response to increasing climatic extremes. This is exacerbated by anthropogenic pressures which have severely modified formerly pristine, high-altitude river catchments. Furthermore, increased encroachment of riverbanks, dumping of solid and sewer waste, and rapid urbanisation has increased the proportion of low-income communities living in flood-prone areas (Mao et al., 2018; Paul et al., 2018). The lack of adequate hydrometeorological monitoring networks or early warning system in these regions causes undue damage to lives and property (Mountain-Evo, 2017; Pandeya et al., 2021). Yet prediction of risks associated with floods is difficult to achieve in such data-scarce mountainous regions.

Indeed, the most recent report of the Intergovernmental Panel on Climate Change (IPCC, 2022) highlighted the urgent need for investment in adaptation and resilience, particularly in developing regions which have been historically underfunded but are already impacted by extreme weather events. A key requirement is to improve early warning alerts of anticipated storms, heatwaves, floods and droughts. To generate such warning information for floods, systematic development of monitoring networks that utilise appropriate technologies is required. These systems should also consider social, cultural and political dimensions to identify context-specific understanding on inequality and its impact on assessing vulnerabilities and exposure, so that the warning system can ensure inclusiveness in responses following appropriate decision-making chains (Mao et al., 2018; Acosta-Coll et al., 2018). Such an integrated and interconnected monitoring system requires science, policy and local community-led approaches that can bring diverse stakeholders (i.e. gender, sex, age, socio-economic status and physical abilities) together and generate knowledge to guide their decision to propose solutions that fit the local context (Buytaert et al., 2018; Kosow et al., 2022; Roque et al., 2022; Zulkafli et al., 2017). Despite this call for an inclusive approach for generating an early warning alert system, the existing flood monitoring practices and designs are strongly technology-driven (i.e. information and communications technology – ICT) and focus less on converging with the local socio-cultural and governance context (Mao et al., 2018; Westerhoff et al., 2021). There are still questions on how, where and at what level science, policy and society may converge and facilitate bottom-up initiatives for decision-making and develop innovative solutions to address challenges posed by floods.

In this commentary, we assess potential approaches for facilitating inclusiveness in the design of a flood early warning system by integrating social, cultural and political aspects and identify preconditions and missing links.

In water and disaster research several approaches are emerging to provide concepts, tools and framings that can be used to support inclusiveness and disciplinary convergence for actionable knowledge production. The concept of knowledge co-production has emerged from science–society interaction under the umbrella of adaptive governance thinking where polycentric models and power relation received attention (see details in Buytaert et al., 2018; Paul et al., 2018; Zulkafli et al., 2017). Scholarly research has identified several potential approaches to achieve knowledge co-production under the broader umbrella of the participatory action research (PAR) including participatory modelling (Sterling et al., 2019), community-based participatory approaches (Wallerstein et al., 2017), participatory scenario analysis (Birthisel et al., 2020; Lakhina et al., 2021; Westerhoff et al., 2021), among others. More recently, citizen science has emerged with an emphasis on “knowledge cocreation and co-generation” (i.e. the interactive processes across science, policy and implementation to collaborate and to generate knowledge for supporting environmental decision-making; see further details in Buytaert et al., 2018) and new technologies, especially ICT, but limited focus on action and development. In addition, citizen science focuses more on participation by volunteers, developing trust and nurturing existing working relationships among involved actors towards knowledge co-production (Buytaert et al., 2018; Zulkafli et al., 2017).

In the contemporary disaster research literature, knowledge co-production is advocated along with participatory actions and transdisciplinary research, which laid the foundation for the participatory convergence concept to translate research into practice (Lakhina et al., 2021; Peek et al., 2020; Roque et al., 2022). Peek et al. (2020) define the participatory convergence research as “an approach to knowledge production and action that involves diverse teams working together in novel ways – transcending disciplinary and organisational boundaries – to address vexing social, economic, environmental, and technical challenges in an effort to reduce disaster losses and promote collective well-being” (p. 2). While this research approach has been identified as one of the best 10 big ideas in funding allocation and research direction by the National Science Foundation of USA (2016), there has been little exploration on the framing (i.e. methods and ethics) to apply this in practice (Westerhoff et al., 2021). Indeed, scholars are focusing on more empirical exploration of convergence research to generate ethics and methods that may deliver successful outcomes, for example, research attempting to address coping with water extremes such as floods and droughts (Lakhina et al., 2021; Roque et al., 2022; Westerhoff et al., 2021). Recently, scholars have proposed ethics that have proven useful. For example, Lakhina et al. (2021) proposed “convergence with CARE: collaboration, accountability, responsiveness and empowerment” which require community engagement and further highlight their perspective, questions and experiences while disregarding traditional hierarchical approaches. However, much hydrological research is focused on improving scientific measurements and developing technological solutions, for example, improving model uncertainty or the instruments and networks used to measure different facets of the hydrosphere (Beven et al., 2020) while being useful for advancing the discipline result in solutions that are often difficult to disseminate to local communities (Birthisel et al., 2020; Roque et al., 2022; Westerhoff et al., 2021). Earlier reviews indicate many empirical investigations on how social context, such as culture, politics and economics, has shaped water knowledge and how and what interventions influence or shape communities' responses differently (Roque et al., 2022). This emphasises a need for future research to understand the underlying principles and ethics that would facilitate bottom-up driven activities or active participation of engaged stakeholders for knowledge co-production to respond to and reshape convergence research methods.

A synthesis of the literature on flood early warning systems was reviewed to develop a schematic representation of an idealised framework for developing an inclusive early warning system (Fig. 1) (for more details see Acosta-Coll et al., 2018; Buytaert et al., 2018; Mashi et al., 2020; Paul et al., 2018; Zulkafli et al., 2017). The foundation of this schematic representation (Fig. 1) is adapted from the concept of knowledge co-generation processes (Buytaert et al., 2018) and co-design framing for environmental decision-making processes in a polycentric system (Zulkafli et al., 2017) and then applied with the key elements (i.e. risk knowledge; technical monitoring and warning service); communication and dissemination of warnings and community response capability (ISDR, 2020) identified by the World Meteorological Organization, International Strategy for Disaster Reduction (ISDR). All these concepts, in general, advocated a participatory and citizen science approach to become inclusive and generate actionable knowledge (Buytaert et al., 2018; ISDR, 2020; Paul et al., 2018; WMO, 2020). The disaster risk equation provided by the IPCC (risk = hazard × exposure × vulnerability / capacity to cope) suggests that reduction in risk is dependent not only on efficient forecasting of hazard, but also on the understanding of associated exposure, vulnerability and capacity to cope by the exposed community. Therefore, in Fig 1, we represent three interdependent steps: (1) mapping the risks through data collection and observation; (2) forecasting hazard risks and establishing an alert system in real time; and (3) communication and dissemination.

We believe that through this commentary we have raised critical questions and identified missing links in the context of disaster resilience and the development of tools to improve preparedness and response. The most important include (i) the absence of diverse contextual risk angle and community reactions; (ii) a lack of community trust in government agencies and technology focused forecasting; (iii) significant data limitations to ensure effective EWS operation and impact-based forecasting; and (iv) a lack of effective communication strategies. All these points need deeper exploration to ensure inclusive EWSs are developed in data-scarce mountainous regions or geographic regions similar in context. We acknowledge that many countries are currently implementing EWSs focusing on active community participation (Red Cross Red Crescent and the UK Met Office, 2022; International Centre for Integrated Mountain Development (ICIMOD), Aranyak and Sustainable Eco Engineering (SEE), 2022, ISDR, 2020; Mountain-Evo, 2017; WMO, 2020); however, solutions to address these missing links are limited, and thus ensuring inclusiveness and impact has remained challenging. We have highlighted the need for multiple lenses to establish and explore the complexity of the risk portfolio and thus understand the architecture of the engaged stakeholders and their behaviour. This is essential to ensure actionable knowledge is generated and bottom-up initiatives are strengthened and the capacity to respond is improved.

Based on the above discussions of key questions, missing links and design needs, we propose the “SMART convergence participatory research” approach to support the EWS development phase and provide a checklist of good practices. The SMART approach highlights crucial activity layers to incorporate into EWS development which can help guide multi-disciplinary teams (e.g. disaster risk manager, hydrologist, engineer and social scientist) (Fig. 2). This will enable the incorporation of diverse disciplinary lenses (i.e. social science and meteorological data) along with risk diversity identified by the community at risk (illegal settlement beside riverbank or slums), which is mentioned earlier as a missing link. This will support exposing vulnerability and risks from different socio-cultural, institutional and scientific contexts. Following a SMART approach will ensure inclusiveness by helping to identify and connect missing components and linkages when designing an EWS.

The first step, S, represents “Shared understanding of the risks”, ensuring all stakeholder engagements are diverse and representative (irrespective of their gender, sex, age, socio-economic status and physical abilities), and a wide range of data forms and collection methods are utilised, as stated in EWS step 1 (Fig. 1). This knowledge generated from the community will help the expert group to better understand context-specific risks with a more focused exposure and vulnerability analysis. This further helps to identify common goals and anticipate damage from the natural hazards and thus ensures impact though appropriate forecasting.

Secondly, M – representing “Monitoring of the risks” – aligned closely with establishing alert system and forecasting hazard information as stated in step 2 (Fig. 1). This includes an intersection of generated knowledge that will lead towards practising collaborative activities, such as building trust (which is key to inclusive and impact-based forecasting), exchanging critical risk information to enrich data sets, feedback and forming small groups for maintaining forecasting system.

Thirdly, A, building “Awareness” (i.e. training and capacity development activities to embed understanding of real time weather and alert information) is critical for this approach and is a continuous process throughout the development and utilisation of early warning systems, with a particular focus on EWS step 3 to support effective communication and dissemination and will further also support legacy and sustainability of the warning system in a local context.

Finally, RT – indicating pre-planning “Response actions on Time” (i.e. comprehensive disaster management plan, evacuation plan) based on the alert produced by the EWS – could be used to inform on the effectiveness of the overall EWS to minimise risks from the anticipated hazard. This will inform further on the level of knowledge produced through collaboration and how this can facilitate effective action by the community and responsible agencies.

We advocate the use of this SMART approach to facilitate bottom-up initiatives for developing an inclusive and purposeful early warning system and to benefit the community at risk by engaging them every step of the way along with including other stakeholders at multiple scales of operations (i.e. scientific and policy actors). We advocate that the SMART convergence approach along with the dominant largely top-down initiatives will contribute to developing capacity and redefining adaptation and resilience in the face of more extreme water extremes (floods, droughts) and increased uncertainty under global change.