Invited Review Paper Climate change and hydrological risk in the Pacific: a Humanitarian Engineering perspective

Pacific Island communities have adapted to floods, droughts and cyclones over many generations. Small and low-lying islands are particularly exposed to natural disasters, and many countries have limited access to water resources. Anthropogenic climate change is expected to further increase these environmental pressures. Any associated engineering response needs to consider the cultural, societal and historical context, and prioritise the agency of local communities to determine their preferred outcomes. It follows that Humanitarian Engineering, a discipline centred around strengths-based and context-appropriate solutions, has an important role to play in climate change adaptation. In this review, the interplay between hydroclimatology, geography and water security in the Pacific Islands is described and projected climate shifts summarised to highlight future adaptation challenges. A key source of uncertainty relates to the dynamics of two convergence zones that largely drive weather patterns. A broad overview of societal factors that present challenges and opportunities for Humanitarian Engineers is given. Finally, actions are recommended to inform climate change adaptation given the scientific uncertainty around hydrologic risks, and outline lessons for best practice Humanitarian Engineering in the Pacific. Enhancing data sharing, building resilience to climate variability and integrating traditional knowledge with convention engineering methods should be key areas of focus.


INTRODUCTION
A key part of sustainability for engineering solutions is that there is local capacity for operating and maintaining systems and infrastructure into the future, which to date has been a key limitation in the Pacific (Baker & Week ). The final step of the Humanitarian Engineering cycle is to evaluate the project and solutions, and potentially iterate to ensure that all the community's goals have been met. There are two main differences between Humanitarian Engineering and a more standard project approach to engineering or other approaches such as rational decision making. The first is an inherent and overarching focus on protecting and promoting human rights. Secondly, Humanitarian Engineering projects should explicitly aim to 'do no harm' by ensuring that projects do not have unintended consequences (CDA Collective ) and understanding that any intervention or action in a community will lead to changes in that community, whether planned or not.
As argued by Rubow  groundwater. Their small size and geographic isolation make them particularly vulnerable to natural disasters such as storm surges. Volcanic islands often have very steep slopes with limited land available for housing and agriculture, and much of the infrastructure is located close to the coast (Kumar & Taylor ). This geography can also cause rain shadows, leading to heterogeneous access to water and risks of drought (Pearce et al. ). The land area of many small islands is less than 10 km 2 ; atoll islands are frequently less than 1 km 2 in area (Dijon ). Therefore, the ocean has a very strong influence on the climate    Importation has also been used to supplement water supply during droughts, particularly for the outer islands of Fiji, PNG and Tonga (Falkland ), although water transfers between islands hundreds of kilometres apart are generally impractical and uneconomic (Duncan ).
As highlighted in Figure 2, only half the population in the  There is a perception of increased drought in many Paci-    There are clear distinctions in remote communities between water sources used for consumptive and non-consumptive uses (Elliott et al. ). Thus, the discussion in Table 3 is mainly focused on drinking water supply. These infrastructure choices are discussed further in the following sections with respect to culture and social perspectives in the Pacific. As discussed in the following section, adaptation options that do not consider the cultural and societal contexts of communities may not be successful. Even worse, they may lead to maladaptation, whereby unintended consequences of the project lead to more perverse outcomes for  Table 3 Rainwater tanks provide buffers against seasonal (or longer) droughts ENSO and drought risk Refer to Table 3 Preparing for extreme weather events by using particular planting a  In such cases, designs need to be resilient to, for example, increased rainfall extremes as well as changed seasonal or interannual patterns of rainfall.

Gravity-fed river systems
Not discussed Catchments in the Pacific tend to be relatively small and rainfall seasonal, so river-based systems are already not appropriate in some locations. Many islands do not have permanent surface water resources. However, in some locations, e.g. Solomon Islands, up to 90% of households use river water, although relatively rarely for consumptive use (6-11%) (Elliott et al. ).

Treatment processes
Processes are resilient but climate change may increase treatment requirements Remote and distributed geography increases cost and potential for maintenance delays, particularly following extreme events/disasters. Maintenance can be affected by lack of resources in water authorities. Back-up rainwater tanks are suggested as one alternative for mitigating such risks (MacDonald et al. ).
Piped water Damage to systems can impact large populations. Highly complex systems, with options for more robust design and operationdepends on management and the financial position of the water authority.
May be strongly groundwater dependent in some parts of the Pacific so sustainability of use and impacts of climate change on recharge need to be considered (Carrard et al. ).
Alternative sources Not discussed Population densities unlikely to be high enough for desalination to be cost efficient. Access to parts and maintenance needs to be considered (cost and availability  Methods fall into either consensus approaches or science integration approaches in which TK is 'validated' to produce forecasting models. Both approaches have advantages and disadvantages, and differ in their requirements for human resources, historical data and community involvement, as well as in their cultural sensitivity (Plotz et al. ).

Religion and spirituality
Understanding the way that risk is perceived by individuals and communities is vital in considering and communicating climate change adaptation options. Community responses to extreme weather and climate change are shaped by myth and religion (Fair ), and risk perception is also influenced by spiritual beliefs. Following colonialism in the Pacific, Christianity has become a strong system in PICs. living is now considered to be normal (Nunn & Campbell ). More recently, as noted in the 'Pacific nations and peoplesgeography, culture, urbanisation' section, urbanisation is increasing in many parts of the Pacific. One of the issues around urbanisation in the Pacific is that peri-urban areas and informal settlements are generally not separately measured and are underserved in terms of water infrastructure (Anthonj et al. ). This makes it particularly difficult to identify the needs of these communities and design infrastructure appropriately. Future research efforts need to be focused on rectifying this situation.
The driver for most urbanisation is education and economic opportunities, and therefore, it has tended to be young adults leaving outer islands. This has implications for both the rapidly urbanising communities as well as the communities in rural areas and outlying islands. This leads to a higher than average percentage of elders in outer communities. A strength of these elders is that they are likely to hold TK that can be used  Table 3, different water sources have different levels of resilience to climate change, both in terms of quantity of supply and also the impacts of climate variability on water quality (Guo et al. ). Therefore, the need for future investments in water supply will be spatially heterogeneous even within one country. Another consideration is that the sustainability of groundwater resources may not be taken into account, and in parts of the Pacific, urban areas have higher reliance on groundwater for drinking than rural areas (Carrard et al. ).
Informal settlements present particular challenges.

Regulations around land tenure and informal settlements
can be a barrier to improving access to water supplies (Sinharoy et al. ), and these issues need to be fully con-  • there are substantial uncertainties with ENSO projections which is a key driver of Pacific hydrological risk; • discrepancies in the spatial resolution of GCMs compared with the size of PICs; • country-level projections of changes in annual average rainfall are not available for many PICs; • interactions between changing rainfall extremes and Pacific Island geography are not well understood; and • drought projections tend to focus on SPI which does not holistically consider all changes in future drought (e.g. both evapotranspiration and precipitation).
Due to the known problems in GCM simulations, along with the importance of topographic effects and the fact that many islands are much smaller than GCM grid cells, GCM outputs require effective bias correction and downscaling before they can be used as inputs to hydrologic models and analyses (Fowler et al. ). produce a detrimental impact. A key advantage is that a range of plausible future cases can be systematically covered, which is often more informative than a discrete set of results from an RCM that may not encompass all possibilities. However, scenario neutral assessments require several decades of observed data on key variables (generally rainfall and temperature) to characterise the baseline climate and form the basis for the perturbed future climate series. In many parts of the Pacific, limited data availability will be a barrier to these studies.
While data and resource constraints are a barrier to quantitative future climate assessments in parts of the Pacific, evidence-based adaptation is still available. The traditional 'anticipatory' adaptation approach, whereby climate changes are modelled and substantial investments are made to manage the impacts, could lead to wasted resources and lost opportunities where future climate uncertainty is high (Barnett ). A better strategy is flexible and resilience-focused, allowing initially local projects to gradually scale-up over time and resources to be reallocated in the face of emerging impacts or projections. In general, systems that are adapted to high climate variability are inherently resilient to climate change (Nathan et al. ).
Even if the precise nature of future climate changes is unknown, communities can prepare by building their resilience to hydrological risks using knowledge gained from the Pacific's inherent high climate variability (Tables 2 and 3).
Such a strategy should place TK at its centre since island communities have been adapting to a highly variable climate for generations.

Best practice Humanitarian Engineering
Data and modelling can only partially contribute to climate adaptation in the Pacific. Many of the major opportunities and challenges to advance adaptation are centred on Pacific communities and their institutions. Best practice Humanitarian Engineering recognises that projects are rarely successful unless the infrastructure and systems that provide the context for a problem are also considered. To this end, identified knowledge gaps evident from this review include: • research methods to integrate TK with contemporary climate and weather forecasts, • Pacific-centric research on the relationship between natural hazards and risk perception, • Pacific-centric research on the value of open data sharing, • policies and planning that accounts for the needs of informal communities given high urbanisation rates in the Pacific, • strengthened institutional frameworks for water policy and legislation, • improved institutional processes around data sharing and data availability, and Pacific Islands are overwhelmingly represented as extremely vulnerable in the face of climate change. While this discourse may have been helpful in leveraging international support, it also denies the agency, resilience and ability of Pacific Islanders to adapt to climate impacts (Campbell & Barnett ). Alternative framing, which considers climate risk, but focuses on adaptive capacity is more likely to result in constructive outcomes (Campbell & Barnett ). • on-site community workshops to develop research topics that are useful for communities and policymakers and to disseminate findings, • community review of data prior to publication, • locally produced and disseminated findings, which prioritise Indigenous language, are developed through participatory approaches, and use relevant multimedia tools (e.g. videos, photographs, podcasts and maps), and • compensation for research participants.