United Nations International Strategy for Disaster Reduction (UNISDR) defines early warning systems (EWS) as ‘an interrelated set of hazard warning, risk assessment, communication and preparedness activities that enable individuals, communities, businesses and others to take timely action to reduce their risks’ (UNISDR 2015). Responding to the Hyogo Framework for Action priority area 2 on knowing the risk and taking action, the government of Malawi instituted the development of EWS through its National Disaster Risk Management Policy of 2015 (GoM 2015b). Priority areas 2 and 3 called for the establishment of a comprehensive system for disaster risk identification, assessment and monitoring, and development and strengthening of a people centred early warning system, respectively (GoM 2015b) and finally the Malawi National Flood Early Warning System (F-EWS) was established.

Unfortunately, the Malawi F-EWS is not discussed much in literature. Figure 2 summarises how the F-EWS is set up. It is divided into three levels of operation: national level, district level and community level. It is reported that F-EWS is primarily run and managed by Department of Climate Change and Metrological Services (DCCMS), Department of Water Resources (DWRs) and Department of Disaster Management Affairs (DoDMA) (GoM 2015b). Department of Climate Change and Metrological Services and DWR have a number of gauging stations on major rivers that are used to measure water levels and discharge. Despite its strength of having a number of gauging station in some major rivers plus a strong institutional link between involved departments, F-EWS has a number of challenges too. Since there are not enough gauging stations and a lot of bureaucracy in communications, there is no guarantee that communities get the early warning messages in time, also no models are used to predict floods, just to mention a few (GoM 2015b).

Within local communities, EWS are sometimes referred to as CBEWS (Smith, Brown & Dugar 2017) or people-centered early warning systems (PCEWS) (UNISDR 2006). Therefore, CBEWS for floods are there to help local communities in using available resources to efficiently prepare and respond to floods event (Smith et al. 2017). For an effective and efficient CBEWS, UNISDR (2006) developed a checklist comprising four crucial elements (Table 2) in instituting EWS. Almost all the EWS are developed around these elements.

The four elements are the cornerstones used to benchmark the effectiveness and efficiency of CBEWS. Risk knowledge is measured by the community’s ability of knowing and understanding the kind of vulnerability that they are prone to. Technical monitoring and warning services stipulate that the vulnerable community must have the ability and capability of monitoring the hazard and producing warning messages. Dissemination and communication of warning promotes that generated warning messages have to be clear enough and should be able to reach out the community at risk. Molinari et al. (2015) also observed that during emergencies, there is need to increase the level of communication by improving the target warning messages. Response capability is the capability of the community to take heed to the warning messages, the community’s ability to react to the warnings and take appropriate action. Khan et al. (2018) used the same benchmark in studying the effectiveness of CBEWS in the State of Azad Jammu and Kashmir. Dugar et al. (2017) and Brown and Dugar (2018) also used the four elements as benchmark to establish the effectiveness of CBEWS in Nepal by looking at the lead time.

In addition to these elements, there are other important issues to be considered in developing CBEWS. These include issues on effective governance and institutional arrangements, involvement of local communities, consideration of gender perspectives and cultural diversity and a multi-hazard approach on EWS (UNISDR 2006). In view of these difficulties a conceptual framework (Figure 3) was developed as a benchmark for assessing CBEWS in Malawi.

The study was conducted in two districts (Figure 4) in the southern region of Malawi: Nsanje and Chikwawa. Even though these are the main areas concerned, some other districts within the country were visited for a comprehensive and richer understanding of the status of EWS within the country. The other districts are Salima and Dedza (in Central Region), which are along the lakeshore of Lake Malawi and Zomba and Phalombe (in Southern Region). These two districts lie at the base of Zomba and Mulanje mountains, respectively.

Nsanje and Chikwawa districts were selected for the study because geographically they are located in the low lying areas and the major river Shire river runs through these districts. According to the Malawi disaster profile on floods (GFDRR 2019; GoM 2019) was based on the intensity and frequency of occurrence of floods in those districts.

In order to have a holistic view and clear understanding of the workability and performance of the CBEWS in Malawi, the study employed a mixed (qualitative and quantitative) approach type of research. Qualitative method is mostly used in behavioural sciences where the researcher attempts to draw meanings out of natural phenomenon. On the other hand, quantitative method is mostly used in ‘hard science’. Quantitative methods assume a fixed and measurable reality, which is concerned with discovering facts about a phenomenon (Jonker & Pennink 2010). Therefore, data collection for this research was done through questionnaires, interviews and group discussions and document analysis.

The term document extends beyond just written materials and includes any symbolic representation that can be recorded and retrieved for analysis (Oates 2006). In this regard documents include textual documents (e.g. reports, memos and academic publications), audiovisual data (e.g. photographs, diagrams, animations, video and sounds) and electronic data (e.g. screenshots, computer games and archives) (Oates 2006).

During the study, a wide range of documents related to disaster management in Malawi were collected and reviewed. Such documents amongst others included policy documents – to gain insight into what the policy states on managing a disaster; what has been reported on the EWS; in Malawi; assessment forms that indicate what information is collected after the disaster and research publications to understand what has been done already relating to the study and identify the gaps within the study areas. Some of the documents that were reviewed included: Disaster Risk Management Operational Guidelines for the Republic of Malawi; A Disaster Rapid Assessment Form; Disaster Impact and Needs Reporting Form and The National Disaster Risk Management Policy.

For a better and clear understanding of what really happens when a disaster happens, group discussions were conducted within the selected communities and non-probability purposive sampling was used. This sampling technique ‘does not need underlying theories or a set number of key informants’ (Tongco 2007). The researcher decides on whom to interview on the basis of their experience and knowledge in the particular field (Creswell & Creswell 2017; Tongco 2007). The key informants (KIs) in this research were those members who know much of the phenomenon and are proper representation of the entire population. Participants or informants included members of District Civil Protection Committee (DCPC), Village Civil Protection Committee (VCPC) and Areal Civil Protection Committee (ACPC). Members of the committees are from government agencies, departments and NGOs. A total of nine Group Village Heads (GVHs) were visited from six districts of Salima, Zomba, Phalombe, Nsanje, Chikwawa and Dedza, respectively. In each of the GVHs, two VCPCs and two ACPCs formed the grouping of KIs with whom interviews and discussions were carried out. On average the discussions and interviews took about 60 min. Predefined questions were used. Depending on the answers provided, further questions were asked seeking for further clarification. The sessions were recorded so as to go through the material at a later stage for any missing information.

As a result of constraint in time and busy schedules of prospective respondents, questionnaires were also used to collect data from three key staff from DoDMA and a total of 37 members of DCPC: 16 members from Nsanje and 21 members from Chikwawa Districts. Of the 37 members, 81% were male and 19% were female. A total of 27% were between ages 25 and 30 years, 16% between 31 and 35 years, 33% between 36 and 45 years and 24% were above 45 years of age. This grouping consists of officers from different government departments and NGOs. There is a deliberate move that within the members there must be at least a female and a youth.

Some of the questions that were asked during the course of interviews and on the questionnaire include the following: (1) Are there any CBEWS installed within their villages? (2) How vibrant are their Civil Protection Committees (CPCs) in relation to warnings because of imminent flooding? (3) What challenges are being faced with the operations of the CBEWS? (4) Are the CBEWS being effective within their areas?

This research assumed a mixed approach type of research. With this type of approach, analysis occurs both within the qualitative (description and thematic text or image analysis) approach and the quantitative (descriptive and inferential numeric analysis) approach (Creswell & Creswell 2017). Quantitatively, International Business Machines Corporation Statistical Package for the Social Sciences (IBM SPSS) Statistics was used in analysing responses from the questionnaire administered to DCPCs. Descriptive analysis and frequencies in Results and Discussion sections of certain responses were derived from this. Whilst MS Excel was used in tabulating graphs, providing comparative visualisation of response obtained from SPSS. This provided a clear understanding behind the meaning of the outcomes.

This article followed all ethical standards of research without direct contact with human or animal subjects.

Administratively, two distinct EWS have been fused together in an attempt to form a national F-EWS. At national level – this is run and managed by governments’ departments: DCCMS, DWR and DoDMA. At local levels, which are within communities prone to flood, they are managed by CPC. Members in some of the committees are volunteers or elected by members within the communities. Figure 2 summarises how the CBEWS are supposedly integrated with the F-EWS.

Technically, the system run by government departments, have weather monitoring stations whose information and data are sent to the offices for further processing. Communications of warnings are carried out after some bureaucratic procedure has taken place. Whilst at community level, it is the designated persons within the committee who are responsible for reading the water levels off the river gauges and disseminating information accordingly.

The respondents to the research were asked what are the strengths and weaknesses of the CBEWS, the opportunities the current CBEWS is presenting and threats the CBEWS is exposed to. Figure 5 presents the outcome of the Strength Weaknesses Opportunities Threats (SWOT) analysis based on the responses from the respondents. From the SWOT analysis it is obvious that 25% account to threats to the current system, 33% present opportunities posed by the systems whilst strengths amount to 38%. From all these findings, the current CBEWS presents 1% opportunity that it can be easily replicated. Figure 6 shows how the CBEWS are viewed by different member of CPCs.

Under the element of risk knowledge, the KIs highlighted that everyone within the affected areas within the district are aware of the prevailing hazards and vulnerabilities. In most of these areas floods re-occur on a yearly basis, the communities are very conversant with the dynamics of the hazard. Evacuation plans and routes are known to all courtesy of the periodic meetings community leaders call for to update and sensitise communities to the weather patterns. Hazard maps (Figure 7) showing flood prone areas are made available by DoDMA to DC and TA offices.

From the interviews with members of the DCPCs, it was revealed that the villagers do not have the technical expertise and know-how to monitor hazards. The local communities only monitor the rising and falling of water levels in rivers using the river gauges (Figure 1). Apart from river gauges, weather monitoring stations in specific areas of the districts have been installed by the Ministry of Agriculture, the DWR and the DCCMS. There are stations to monitor weather patterns and help in predicting whether the weather will be favourable or not. Unfortunately, there is no direct linkage between these tools and CBEWS found in the districts (Hagenlocher et al. 2012).

The essence of EWS is to save lives and loss of property. From the given findings, in all the districts it was reported that there are different means of communicating and disseminating warning message. These include community radios, megaphones, drums and word of mouth. It was reported that the warning messages are conveyed and communicated by the members of the VCPC and ACPC although these are mostly communicated when the danger is imminent (when water is at red mark on the river gauge). Megaphones, whistles and drums are used in warning people of imminent flooding so that they move to higher grounds or safer places. The bureaucratic setting the system (GoM 2015b; Hagenlocher et al. 2012; Mijoni & Izadkhah n.d.), is ideal for warnings whose time span is longer because the information trickles down to vulnerable communities late. ‘The other problem with this setting is that warning messages are generalised and not area specific, hence at times people don’t take them seriously’ – commented one respondent in Salima. The same sentiments were shared in Dedza, Zomba and Nsanje. Warning massages have to be specific to an area.

Within the Disaster Risk Management Operational Guidelines (GoM 2015a), local authorities have been mandated to ensure that their respective communities are capable of responding to disasters. Multi-hazard contingency plans, evacuation plans and recovery plans have to be developed by local authorities in line with national plans.

In all the places visited, KIs under the VCPC and DCPC stated that periodic meetings are held in respective villages, which are exposed to flood hazards reminding each other of what is expected of them during the disaster. Luckily enough each area has its own identifiable safe place when floods hit. School grounds, places of worship area, nearby hills have all been identified as safe havens when floods arise. Sometime drills are carried out simulating flood disaster and to test the response rate of the community (COOPI 2018).

If there is a weakness or failure in any element of a system, the whole system is bound to fail (Macherera & Chimbari 2016; Molinari et al. 2015). Although the four key elements were considered in CBEWS implementation in Malawi, the system is facing some challenges. ‘Lack of resources or funding’ supporting the operations of CBEWS (PARTICIP 2017) came up as one of the main challenges the systems is facing. Civil Protection Committees members not having mobile phones to use or even credit in the phones to alert others by calling or messaging also presents problems (Hagenlocher et al. 2012). Furthermore, most of the river gauges have been installed under projects that have life span (Hagenlocher et al. 2012). These are not sustained when the project concludes its operations such as the ones in Salima the project concluded in 2017. In addition, ‘lack of proper coordination’ amongst different stakeholders has greatly affected the operations of the CBEWS. ‘Absence of inter-river or inter-districts alliances’ in most districts has rendered information not being passed from upper-stream to down-stream communities.

Continual change of CPCs that has resulted in ‘deficiency in CPCs capacity’ to interpret and carry out duties properly is another factor in contributing to the collapse of the system in place. As the case in Nsanje where it was reported that most outgoing CPCs do not provide handover observed for the incoming ones. Chiusiwa lamented that there is need to build capacity of the communities to enable them to ably interpret and use early warning information effectively (Chiusiwa 2015). One of the major deterrents to the success of the CBEWS is that only major rivers are monitored whilst small rivers and other tributaries that equally contribute to flooding are not monitored (Chiusiwa 2015).

With reference to Figures 5 and 6 on how the CBEWS are faring, it can be observed that the systems are not performing as anticipated. A total of 46% (sum of very good and good) respondents perceived that the CBEWS in place are helping and performing above average whilst 54% think otherwise. As stated by Van Belle, Eccles and Nash (2003) that within a system, there is a special relationship between components that are interrelated and are supposed to work together for a specific objective or purpose. Unfortunately, this is not the case with CBEWS in Malawi. Community-based early warning system are prone to vandalism, heavily relying on donor funding, managing them on voluntary basis and the fact that warnings come late are some of the aspects rendering the CBEWS ineffective. Macherera and Chimbari (2016) and Khan et al. (2018) concurred with the UNISDR that CBEWS are bound to fail if there is a failure or weakness in any element of that system.