Search

Direct policy search (DPS) is emerging as one of the most effective and widely applied reinforcement learning (RL) methods to design optimal control policies for multiobjective Markov decision processes (MOMDPs). Traditionally, DPS defines the control policy within a preselected functional class and searches its optimal parameterization with respect to a given set of objectives. The functional class should be tailored to the problem at hand and its selection is crucial, as it determines the search space within which solutions can be found. In MOMDPs problems, a different objective tradeoff determines a different fitness landscape, requiring a tradeoff-dynamic functional class selection. Yet, in state-of-the-art applications, the policy class is generally selected a priori and kept constant across the multidimensional objective space. In this work, we present a novel policy search routine called neuro-evolutionary multiobjective DPS (NEMODPS), which extends the DPS problem formulation to conjunctively search the policy functional class and its parameterization in a hyperspace containing policy architectures and coefficients. NEMODPS begins with a population of minimally structured approximating networks and progressively builds more sophisticated architectures by topological and parametrical mutation and crossover, and selection of the fittest individuals concerning multiple objectives. We tested NEMODPS for the problem of designing the control policy of a multipurpose water system. Numerical results show that the tradeoff-dynamic structural and parametrical policy search of NEMODPS is consistent across multiple runs, and outperforms the solutions designed via traditional DPS with predefined policy topologies.

Agriculture is one of the most vulnerable sectors to changes in climate patterns. Moreover, climate change is a complex and interdisciplinary problem, where natural processes are closely intertwined with socio-economic aspects, especially in areas where human activities are widespread. Thus, when planning for natural resources management, it becomes essential to consider how storylines, attitudes and behaviours can influence the decision-making towards adaptation measures. In particular, when considering an agricultural system, farmers are key agents that cannot be neglected, as the decision to adapt or to change their agricultural practice is ultimately in the hands of the individuals. Understanding the reasoning behind farmer adaptation can help create a sounder framework to recognize farmers’ awareness and experiences regarding climate change, while reinforcing their resilience to face climate change scenarios. Consequently, finding patterns in attitudes and similarities between farmers is essential both to better share an overall picture of climate change effects and perspectives in a specific study area and to summarize the complex set of factors influencing farmers’ behaviours to facilitate their modelling. Different tools and methods have been provided by the literature in the last two decades to delve into farmers’ attitudes and perspectives regarding climate change. One of the most used tools are structured surveys, mainly due to their strongly case-specific nature and the capacity to synthesize climate change scenarios in a standardized way. Here, we provide an overview of the results obtained through a survey of 460 farmers from northern Italy about climate change risk awareness, perceived impacts, and adaptive capacity. In addition to a descriptive statistical analysis to delve into farmers’ profiles and farming characteristics, this triple-loop approach was analysed through Multiple Correspondence Analysis (MCA), an interesting data analysis technique that allows to highlight underlying structures in categorical data used to recap farmers’ behaviours and define the association between dependent and independent variables. The resulting factor map allows for the identification of those variables that most explain the variance in the dataset and expected similarities and differences between respondents. The obtained results show how certain variables describing the agricultural practice of the respondents, such as farm extension or the preferred irrigation method, are key driving factors in differentiating and grouping individuals’ behaviour. In general, farmers with the same modus operandi share similar behaviour with respect to other aspects of their activity (e.g., water source). Interestingly, and contradicting similar experiences from the literature, this pattern differs among farmers with comparable demographic background, requiring more attention to farmers’ heuristics. These results can be useful in multiple ways: from creating an informative picture of farmers’ attitudes and concerns regarding climate change in a certain area, to its application in profiling farmers to identify common demands and shared worries; from helping with the creation of customized policies at the regional scale from a bottom-up approach, to the implementation of farmers’ profiles into Agent-Based Models to reinforce the human dimension in decision-making processes.

Today the management of water systems requires robust policies capable of withstanding deviations from the conditions for which they were originally designed due to the large degree of uncertainty about future inflows and water demands. Under evolving and deeply uncertain hydroclimatic inputs, the performance of these systems may degrade to a point where they become unable to meet the primary objectives for which they were built, potentially causing declines in water resource system performance or even complete system failure.Here we present a Multi-Objective, Robust Decision Making analysis applied to the case study of Jucar river basin in Spain, where the balance between water demand and available resources is already precarious. As in most Mediterranean basins, climate change is expected to further reduce water availability, increasing the intensity of drought episodes, and challenging the long-term sustainability of water use. Using a hydroeconomic model of the basin, we assessed the performance of the current system’s operations in terms of agricultural and hydropower benefits, along with ecosystem services under different CMIP6 scenarios over the time horizon 1980-2100. These climate projections are then synthetically perturbed to generate a larger ensemble of future scenarios, which is used to complement the robustness analysis and identify via scenario discovery the most critical drivers under which the system is expected to fail.Preliminary results using the nominal IPCC scenarios indicate substantial system vulnerabilities emerging over the next decades, especially under the pessimistic SSP5-8.5 projection. These findings suggest the need of identifying candidate adaptation options to be triggered according to the future evolution of the system in order to ensure a timely adaptation of water management strategies to future changes.

Development pathways for Sub-Saharan Africa project a substantial increase in population and living standards. To accommodate the future energy demand, the power and the energy system community have been developing least-cost optimization models to support long-term planning. Given the rise of climate change impacts and the necessity to mitigate them in the future, investments in new energy infrastructure should contemplate carbon neutral alternatives wherever economically viable.However, integrated assessment and long-term energy planning models usually focus on annual or seasonal energy balances neglecting higher time resolution dynamics that can actually lead to short but high impact events. Indeed, the variability of renewable generation and power demand can lead to significant risks of high electricity prices, transmission lines overload, and power generation deficits.To quantify these impacts, we inform a power system simulation model, PowNet, with energy development pathways from the long-term energy system planning model, OSeMOSYS-TEMBA. While the latter models the development of all the countries in continental Africa with a seasonal resolution from 2015 to 2070, the first has an annual horizon with hourly resolution and focuses on countries included in the Southern African Power Pool. In particular, PowNet is used to optimize the dispatch of power from each source as well as the usage of transmission lines, and it is constrained to the power capacity available according to the long-term energy planning provided by the OSeMOSYS-TEMBA model. We assess these impacts in 2025 and 2030 under three climate policy scenarios: no climate policy, and constrained to 2.0°C and 1.5°C warming constraining emissions to a consistent pathway. We study the difference in generation mix, the impact on transmission lines overloading, power generation deficits, and electricity prices.Preliminary results show an increase of the generation during the years, in particular of the renewable resources, that varies depending on the selected scenario. Moreover, power generation deficits and transmission lines overloading are observed in many countries, especially during the night. These impacts are to be associated with insufficient total power system capacity to meet power demand due to the low time and spatial resolution of the energy system model. Indeed, the increased dependency on variable renewable resources, and a higher resolution demand profile prove the need to further expand total capacity, the importance of flexible generation adopting a diverse energy portfolio, and the potential benefits of increasing transmission lines’ capacity. Finally, the lack ofpower storage technologies in the energy system model might also significantly affect capacity expansion plans and consequent impacts. These results show the importance of the assumptions embedded in energy system model and motivate methodological improvements to design coupled energy and power system pathways that remain reliable at high spatial and time resolution.

Urban water demands vary across spatio-temporal scales, driven by multiple socio-demographic, climatic, and urban form factors. Identifying influential drivers, along with their individual and compound effects on urban water consumption, is essential to forecasting future water demand, addressing urban water security, and informing water governance. Model-free and model-based Input Variable Selection (IVS) has been extensively applied to investigate important predictors of urban water demands. However, most IVS methods identify correlations and mutual information between variables, which do not imply causation. More recently, causal discovery has developed as an active area of research in many research fields, including, e.g., neuroscience, finance, and climate science. Causal discovery improves IVS by identifying causally meaningful relationships between variables, distinguishing indirect from direct dependencies, and recognising relevant drivers among multiple variables.In this work, we investigate predictive and causal factors of urban water use across the Contiguous United States (CONUS). We rely on open data of monthly municipal water consumption from 126 cities in the US for the period 2010-2017 and data on candidate socio-demographic, climatic, and built environment predictors from multiple sources, including the U.S. Census Bureau, the American Community Survey, and the PRISM climate data set. We first test the state-of-the-art W-QEISS wrapper method to identify equally-informative subsets of predictive factors for urban water demands. These subsets are the solutions of a four-objectives optimisation problem that maximises the predictive accuracy of a data-driven model and feature relevance while minimising the number of selected predictors and their redundancy. Results show that historical water consumption is the most relevant factor to predict future demands, followed by some socio-demographics, climatic factors, and building characteristics, including the median number of rooms in housing units, unemployment rate, Palmer Drought Severity Index (PDSI), and building construction years. Preliminary results for individual climate regions also highlight local effects, with PDSI becoming more relevant for arid regions than the continental-scale results. Second, we extend our analysis to causal discovery by applying a neural Granger model to interpret non-linear Granger causality and temporal structures within time series. Granger causality describes whether past values of a time series xt could predict future values of another series yt, assuming causal effects are ordered in time (i.e., cause before effect). This allows for finding the specific causes of urban water demands in our case (in a Granger’s sense). We finally compare the causality results with the results of IVS to illustrate the different interpretations of urban water demand drivers.

Tropical Cyclones (TCs) have the potential to cause extreme rainfall and storm surge, which in turn can lead to riverine and coastal flooding with huge damage to property and loss of lives.The use of precipitation forecasts in the context of decision-making and anticipatory action is currently hampered by the limited skill of numerical weather prediction models in forecasting the characteristics of such extreme rainfall events (especially their severity and location) with a sufficiently long lead time.In this study, we present a post-processing scheme for precipitation forecasts based on a popular deep-learning algorithm (U-Net). We design our Machine Learning (ML) model to reduce the local biases of precipitation forecasts from TCs and adjust the spatial distribution of extreme rainfall. For this, we use a composite loss function to train the model, based on the combination of the Mean Absolute Error (MAE) and the Fractions Skill Score (FSS). We first demonstrate the potential of our ML-based approach working on ERA5 reanalysis data and subsequently apply it to the ensemble mean of ECMWF sub-seasonal forecasts with a lead time up to 10-days. As for the ensemble spread, we investigate possible post-processing adjustments based on the improvement of the spread-error relationship and of action-relevant scores of interest for humanitarian agencies, namely False Alarm Ratios (FAR) and Hit Rates (HR). We train and validate the model on a historical dataset of global TC precipitation events, using ECMWF re-forecasts over 20 years and a multi-source observational dataset (MSWEP) as reference. The results are evaluated with a multi-criteria approach including MAE, FSS, FAR, and HR, to assess the capacity of improving the predicted severity and spatial patterns of TC precipitation, as well as their potential for triggering anticipatory actions. Finally, we discuss how the outputs of our model can be used and further improved to support humanitarian actions aimed at saving lives in vulnerable communities in Mozambique.

A drought is a slowly developing natural phenomenon that can occur in all climatic zones, and propagates through the entire hydrological cycle with long-term economic and environmental impacts. Climate change has made drought one of the greatest natural hazards in Europe, affecting large areas and populations. Different definitions of drought exist, i.e. meteorological, hydrological, and agricultural droughts, which vary according to the time horizon considered and differ in the variable used to define them. Just as there is no single definition of drought, there is no single index that accounts for all the types of droughts. As a consequence, capturing the evolution of drought dynamics and associated impacts across different temporal and spatial scales remains a critical challenge.In this work, we analyze existing standardized drought indexes in terms of their ability in detecting drought events at the pan-European scale using data from HydroGFD2.0 reanalysis and E-HYPE hydrological model simulations over the time period 1993-2018. We firstly compare the frequency and mean duration of drought events detected by different indexes to identify the river basins mostly affected by droughts and to assess similarities and differences in the information provided by different indexes. We then compare them with the drought impacts recorded in the Geocoded Disasters (GDIS) dataset to examine agreements and discrepancies between index-detected droughts and impact data.Preliminary results show that different indexes generally agree in pointing out that Southern England, Northern France, and Northern Italy are the regions that experienced the highest number of drought events, whereas other regions, such as Southern Spain, experienced intense droughts events, which are not consistently indicated by all indexes. In terms of drought duration, the areas affected by the longest droughts are instead the Baltic Sea region and Normandy. Clustering the 35408 European basins according to dominant hydrologic processes reveals that the variables mainly controlling the drought process vary across clusters and depends on the characteristics of each cluster. While substantial agreement exists between observed impact and detected drought, several areas without GDIS records show critical index values. Such asymmetry can be explained by incomplete reporting in GDIS but also due to some non-physical hydrometeorological factors influencing drought dynamics, such as controlled water infrastructure, that are not adequately captured by standardized indexes. These findings suggest the need of adjusting the formulation of drought indexes to the specific characteristics of different river basins in order to improve drought detection and management.

Extra-Tropical Cyclones are major systems ruling and influencing the atmospheric structure at mid-latitudes. They are characterised by strong winds and heavy precipitation, and can cause considerable storm surges potentially devastating for coastal regions. The availability of historical observations of the extreme events caused by intense ETCs are rather limited, hampering risk evaluation. Increasing the amount of significant data available would substantially help several fields of analysis influenced by these events, such as coastal management, agricultural production, energy distribution, air and maritime transportation, and risk assessment and management.Here, we address the possibility of generating synthetic ETC atmospheric fields of mean sea level pressure, wind speed, and precipitation in the North Atlantic by training a Generative Adversarial Network (GAN). The purpose of GANs is to learn the distribution of a training set based on a game theoretic scenario where two networks compete against each other, the generator and the discriminator. The former is trained to generate synthetic examples that are plausible and resemble the real ones. The input of the generator is a vector of random Gaussian values, whose domain is known as the “latent space”. The discriminator learns to distinguish whether an example comes from the dataset distribution. The competition set by the game-theoretic approach improves the network until the counterfeits are indistinguishable from the originals.To train the GAN, we use atmospheric fields extracted from the ERA5 reanalysis dataset in the geographic domain with boundaries 0°- 90°N, 70°W - 20°E and for the period 1st January 1979 - 1st January 2020. We analyse the generated samples’ histograms, the samples’ average fields, the Wasserstein distance and the Kullback-Leibler divergence between the generated samples and the test set distributions. Results show that the generative model has learned the distribution of the values of the atmospheric fields and the general spatial trends of the atmosphere in the domain. To evaluate better the atmospheric structure learned by the network, we perform linear and spherical interpolations in the latent space. Specifically, we consider four cyclones and compare the frames of their tracks to those of the synthetic tracks generated by interpolation. The interpolated tracks show interesting features consistent with the original tracks. These findings suggest that GANs can learn meaningful representations of the ETCs’ fields, encouraging further investigations to model the tracks’ temporal evolution.

The integrated management of water reuse technologies and their coordination with the operations of the other water system components are fundamental to fully exploit the reuse potential. Yet, these technologies are usually designed considering their individual parameters (e.g., efficiency, durability, maintenance costs, energy consumption), more than the integration with traditional water management practices, and the impacts on the final users at the system scale.Here, we adopt a portable framework based on optimal control methods and machine learning to evaluate the cross-sector impacts of water loops. The framework is developed for the Apulia Region, Southern Italy, a drought-prone area characterized by the presence of a complex water distribution network and multiple conflicting users across agricultural districts, industry, and drinking water supply.The robustness of each adaptation strategy is comprehensively investigated through a scenario-based approach, including the analysis of climatic, socio-economic (drinking, irrigation, and industrial water demand pattern), legal (environmental flow constraints), and technological (water reuse implementation) aspects.Results show that the combined effect of climate and socio-economic changes will dramatically affect the Apulia water system, leading to unsustainable pressure on freshwater resources. In addition, the implementation of the environmental flow constraints will further reduce the operation space. Future water deficit is thus expected to increase at half-century (2050-2059) as well as in the long-term (2090-2099), especially under the more extreme climate projection (RCP 8.5).Results also show that water reuse actions remarkably improve the situation, but the effect is only partial and far from entirely closing the gap with the current situation. This means that the specific adaptation actions here adopted are not sufficient and that it is necessary to further promote the spread of the reuse technologies and increase their efficiency.The proposed framework is a decision support system that aims at assisting policy-makers in the transition to a circular water economy by integrating water management and treatment-reuse technologies.

Agricultural systems are adversely influenced by climate change through increased temperatures, change in run-off patterns and seasonality fluctuation. Farmers are, hence, a valuable source of first-hand observations of climate change as they may provide a deeper understanding of their manifestation and relevance. Farmers are aware of climate change impacts and promote adaptation measures such as changing crop varieties, adjusting planting dates, introducing agroforestry practices, and promoting soil and water conservation practices. However, some adaptation barriers persist such as the limited knowledge of potential adaptation strategies, high adaptation cost, or poor institutional support. Understanding why and how farmers aim to adapt to climate change is imperative to provide informed decisions to policy-makers and the first step to minimizing misconceptions or maladaptation practices. Consequently, drivers and influencing factors of farmers’ behavior toward climate change have received increasing attention in the last two decades. Social and behavioral sciences have investigated the influence of farmers’ experiences in increasing climate change adaptation capability and improving decision-making processes at the system level, concluding how local perceptions provide sufficient baseline information for understanding individual and collective exposure to climate risks and risk aversion patterns, an essential element for effective policy formulation and implementation. Traditional management approaches based on simple, linear growth optimization strategies, overseen by command-and-control policies, have proven inadequate for effective adaptation to climate change. Conversely, accurate bottom-up approaches focused on social learning can complement the system transformation by building collaborative problem solving. In this line, associative processing methods, such as interviews and surveys, have been discussed for their ability to delve into knowledge-based data and monitor the nature, significance, and influence of personal experience on climate change adaptation. Agent-Based Models (ABM) can include feedback between social and physical environments, define individuals’ narratives, and map the social network with agents’ interactions. This proposal aims at presenting a transdisciplinary approach that integrates survey data, with behavior and agrohydrological modelling in order to support policy-makers and managers to understand and re-think water management and climate change policies at the regional scale, which is essential to address climate change risks. From a system dynamics perspective, we characterize farmers’ risk aversion patterns and examine how ABMs can most effectively integrate these insights to increase robustness in decision-making processes while attending to farmers’ adaptive capacity. In the application to the case study of a large irrigated area in northern Italy, we surveyed 460 farmers to deepen a triple loop analysis on climate change awareness, perceived impacts, and adaptation measures and barriers. Statistics and computer-assisted data analysis were applied to gain insights from farmers’ profiles and risk perception. We included the profiles in an ABM coupled with a distributed agrohydrological model that covers the whole irrigated area. We expect farmers’ profiles influence agents’ risk perception and their ultimate decision on the adopted crop types and irrigation methods. Provisional results indicate that the approach can provide new insights across complexity in modelling farmers’ behavior and human adaptation to climate change and enrich the discussion about the gaps and benefits of including qualitative data in ABM.

Runoff from Arctic rivers has a direct influence on the sea ice dynamics in the Arctic Ocean, producing significant effects from local to global scales. Despite their key role, the knowledge of the processes that influence the Arctic rivers streamflow is still limited, and their behavior is not fully understood.In the literature, these analyses are usually performed adopting classical statistical methods and simple linear models, which are probably unable to fully capture underlying nonlinearities and redundancy of candidate drivers.In this study, we use automatic feature selection techniques to detect the main drivers of the five major Arctic Rivers’ runoff (Ob, Yenisei, and Lena in Asia, Mackenzie and Yukon in North America). Daily time series of temperature and precipitation recorded by several stations spread across the Arctic region, and the average snow cover of each basin are used as candidate input variables.The feature selection analysis is carried out with two algorithms: Wrapper for Quasi Equally Informative Subset Selection (W-QEISS) and Iterative Input Section (IIS). W-QEISS adopts neural predictive models to select alternative sets of drivers providing similar in terms of accuracy, but with different relevance, redundancy, and cardinality. Conversely, IIS directly produces a ranking of the input variables relying on tree-based models and combining computational efficiency and scalability to high input dimensionality.The two algorithms achieve noticeably consistent results, with minor differences that can be explained by numerical factors typical of machine learning. Results also show that autoregressive terms have a crucial role in all the hydrological basins, while the importance of the other drivers is different for each river.This preliminary research opens the floor for further analysis to broaden the knowledge of Arctic hydro-meteorological dynamics.

How to create an impact on policies, operations, and society across the interdisciplinary sectors in which we - as researchers - are involved? Managing the Water-Energy-Food-Ecosystem (WEFE) nexus and pursuing climate resilience is the core task of several European (EU) projects and is in the highest interests of our society. The European Commission’s research funding programs attempt to address a large range of topics and offer unique opportunities for scientists to create a tangible impact on the environment and society. We are currently involved in different EU projects, including AWESOME (PRIMA), which aims at managing the WEFE nexus across sectors and scales in the South Mediterranean exploring innovative technologies such as soilless agriculture in the Nile Delta; CLINT (H2020), which is developing Machine Learning (ML) techniques to improve climate science in the detection, causation, and attribution of extreme events to advance climate services; IMPETUS (H2020), whose efforts are dedicated on the elaboration of climate data space enhanced with ML algorithms to support the elaboration of climate policies; REACT4MED (PRIMA), which focuses on combating land degradation and desertification by improving sustainable land and water management through the identification of local good restoration practices and their potential upscaling; Gaza H2.0: Innovation and water efficiency (EuropeAid), which aims at promoting efficient and sustainable water supply and demand as well as knowledge transfer to enhance resilience against water scarcity in Gaza; GoNEXUS (H2020), which is developing an evaluation framework to design and assess innovative solutions for an efficient and sustainable coordinated governance of the WEFE nexus; NexusNet (COST), which creates the network and the community of WEF nexus advocates for a low-carbon economy in Europe and beyond; NEXOGENESIS (H2020), which focuses on streamlining water-related policies with artificial intelligence and reinforcement learning; MAGO (PRIMA), which builds web applications for water and agriculture in the Mediterranean; BIONEXT (HEU), which is interlinked with the Intergovernmental Panel on Biodiversity and Ecosystem Services and aims at creating transformative change through nexus analysis. Despite the efforts of the scientific community, there is still a gap between research and practice. Researchers face difficulties in engaging stakeholders and decision-makers to jointly explore and shape the developed solutions, as well as to truly adopt them. The large-scale implementation of suitable technological solutions might require time and financial resources beyond the project’s lifetime and capacity. The lack of follow-ups and collaboration among projects with similar aims can be some of the reasons lying behind. Also, the complexity of finding open data in data-scarce regions makes results less trustable in the eyes of international agencies, while the pressure of publishing often turns research tasks into pure academic exercises. To what extent does the European strategy work? Is it only gaining scientific advances or also leading to local policy changes? Engaging important local actors (e.g., ministries), small-medium enterprises and societal members in the project consortia, empowering scientists by ensuring feedback loops with local governmental agencies, including the human dimension into modelling, and running effective capacity-building campaigns can be some food for thoughts to shape new strategies.

Tropical cyclones (TCs) are one of the costliest and deadliest natural disasters due to the combination of their strong winds and induced storm surges and heavy precipitation, which can cause devastating floods. Unfortunately, due to its high spatio-temporal variability, complex underlying physical process, and lack of high-quality observations, precipitation is still one of the most challenging aspects of a TC to model. However, as precipitation is a key forcing variable for hydrological processes acting across multiple space-time scales, accurate precipitation input is crucial for reliable hydrological simulations and forecasts.A popular source of precipitation data is the ERA5 reanalysis dataset, frequently used as input to hydrological models when studying floods. However, ERA5 systematically underestimates TC-induced precipitation compared to MSWEP, a multi-source observational dataset fusing gauge, satellite, and reanalysis-based data, currently one of the most accurate precipitation datasets. Moreover, the spatial distribution of TC-rainfall in ERA5 has large room for improvement.Here, we present a precipitation correction scheme based on U-Net, a popular deep-learning architecture. Rather than only adjusting the per-pixel precipitation values at each timestep of a given TC, we explicitly design our model to also adjust the spatial distribution of the precipitation; to the best of our knowledge, we are the first to do so. The key novelty of our model is a custom-made loss function, based on the combination of the fractions skill score (FSS) and mean absolute error (MAE) metrics. We train and validate the model on 100k time steps (with an 80:20 train:test split) from global TC precipitation events. We show how a U-Net trained with our loss function can reduce the per-pixel MAE of ERA5 precipitation by nearly as much as other state-of-the-art methods, while surpassing them significantly in terms of improved spatial patterns of precipitation. Finally, we discuss how the outputs of our model can be used for future research.

As many developed countries prepare their transition to net zero emissions energy systems, Africa must plan a substantial expansion of its energy production sources, to meet the growing demand driven by increasing population and energy access. Among the existing capacity expansion potential, hydropower plays an important role in providing clean and cheap electricity. Yet, large hydropower schemes bring many negative social, environmental, and geopolitical externalities.Least-cost optimization models constrained to satisfying predefined energy demands are used for large-scale energy system planning. Multi-objective optimization models can also incorporate environmental impacts in energy system planning, for instance by constraining the optimal solutions on GHG emissions or geomorphologic connectivity losses. However, these traditional techno-economic approaches overlook governance considerations, which are relevant to energy security, especially in unstable and conflictual political contexts. In fact, concerns about political instability are ranked among the main investment risks for foreign investors in developing countries. The subject becomes even more significant in transboundary river basins, where institutional stability and the absence of conflicts are crucial for effectively building and operating large hydropower projects.To assess the political risks associated with the hydropower sector, we examine six pathways of energy generation for the African continent, from 2020 to 2050, developed using the OSeMOSYS-TEMBA energy system model. The model considers more than 600 existing and future hydropower projects in all countries of continental Africa, including available information for each individual power plant. Moreover, it incorporates ISIMIP2b scenarios to integrate coherently final energy demands, land-use change, and climate impacts on water availability.For each scenario considered, the political risk deriving from the associated electricity generation and exchange patterns is characterized at the country-level using six energy-related dimensions. The more vulnerable transboundary river basins are then selected by intersecting the countries with high energy-related political risk and regions with high hydro-political conflict based on existing literature. We use a worst-case perspective for these basins and assume that electricity generated from planned or existing hydropower projects would not be exchanged between co-riparian countries due to the lack of cooperation. Finally, the impacts on the energy system are re-evaluated for the resulting cost-optimal energy system reconfiguration, and the difference with the fully connected solution is assessed.Our results show that integrating political stability in energy system planning can produce precise spatial information about potential risks. Indeed, the lack of cooperation in transboundary river basins affected by high political instability can emphasize pre-existing vulnerabilities. Since this issue severely influences decisions related to energy planning on a continental scale, energy analysts can improve energy security using these results to design capacity expansion robust to political shocks.

A resurgence of dam planning and construction is under way in several river basins where untapped hydropower potential could meet growing energy demands. In Africa, more than 300 new hydropower projects are under consideration. Yet, hydropower expansion is a contentious issue given the uncertainty in water and energy demand as well as the negative impacts of these infrastructures on other sectors. Despite calls for a more comprehensive evaluation of hydropower projects, most dams continue to be planned with traditional methods that neglect interdependencies between planning and management and the cumulative impacts of multiple new dams. Here, we use the transboundary Zambezi Watercourse in southern Africa to present a novel dam planning approach that integrates sequencing of planned reservoirs with adaptive, multipurpose operations to address increasing and competing demands for water, energy, and food in the region.Results show how seeking compromise through operations while constructing dams early improves environmental and irrigation objectives by 50% and 80%, with an 8% loss in hydropower compared to an operation and sequencing strategy that singularly maximizes hydropower. Alternatively, seeking compromise only through delayed dam construction yields modest environmental and irrigation improvements of 6% and 9%, respectively, with a 22% loss in hydropower. Our findings indicate that while additional hydropower capacity reduces structural energy deficits, operating policies emerge as the main driver of human-environmental tradeoffs. Consequently, traditional single-objective operating policy selection may lead to erroneous perceptions of tradeoffs across infrastructure options. The robustness of this result is tested under an ensemble of stochastic hydrologic projections where environmental flow and irrigation deficits are found more sensitive to operations than shifts in water availability. The predominance of operating policies is relevant for improving multi-objective dam planning in other river basins already fragmented by dams built in the 20th century.

The Lancang-Mekong River basin (LMRB) is facing blooming water resources infrastructure development, long-term transboundary conflicts and trade-offs between economic goals and ecosystem services provision. Yet, most studies optimizing the pathway towards sustainable infrastructure operation have lacked multi-sectoral and cross-country perspectives. Here, we quantify how and to what extent transboundary cooperation generate economic and environmental co-benefits by jointly using a coupled simulation-optimization approach and cooperative game theoretical analysis. We find that full cooperation outweighs non- or partial cooperation modes to promote economic benefits by 5 to 27%, and to minimize the losses in fishery and sediment transport from 14% and 33% to 8% and 10%, respectively. Full cooperation becomes more beneficial and stable alongside infrastructure expansion, climate change, and the degree of satisfying hydrological needs for river ecosystems. These findings underscore the importance of full cooperation for sustaining socio-environmental systems and highlight the needs of benefit reallocation mechanism and designed flow management for stabilizing basin-level full cooperation in the LMRB.

Shared water resource systems spark a number of conflicts related to their multi sectorial, regional, and intergenerational use. They are also vulnerable to a myriad of uncertainties stemming from changes in the hydrology, population demands, and climate change. Planning and management under these conditions are extremely challenging. Fortunately, our capability to approach these problems has evolved dramatically over the last few decades. Increased computational power enables the testing of multiple hypotheses and expedites the results across a range of planning alternatives. Advances in flexible multi-objective optimization tools facilitate the analyses of many competing interests. Further, major shifts in the way uncertainties are treated allow analysts to characterize candidate planning alternatives by their ability to fail or succeed instead of relying on fallible predictions. Embracing the fact that there are indeterminate uncertainties whose probabilistic descriptions are unknown, and acknowledging relationships whose actions and outcomes are not well-characterized in planning problems, have improved our ability to perform diligent analysis. Multi-objective robust planning of water systems emerged in response to the need to support planning and management decisions that are better prepared for unforeseen future conditions and that can be adapted to changes in assumptions. A suite of robustness frameworks has emerged to address planning and management problems in conditions of deep uncertainty. That is, events not readily identified or that we know so little about that their likelihood of occurrence cannot be described. Lingering differences remain within existing frameworks. These differences are manifested in the way in which alternative plans are specified, the views about how the future will unfold, and how the fitness of candidate planning strategies is assessed. Differences in the experimental design can yield diverging conclusions about the robustness and vulnerabilities of a system. Nonetheless, the means to ask a suite of questions and perform a more ambitious analysis is available in the early 21st century. Future challenges will entail untangling different conceptions about uncertainty, defining what aspects of the system are important and to whom, and how these values and assumptions will change over time.

Participatory decision-making is a well-established approach to address the increasing pressure on water systems induced by growing multi-sectoral demands and increased competition among different water users. However, most existing approaches search for system-wise efficient solutions and do not quantify their distributional effect among the stakeholders. In this work, we investigate how to operationalize equity principles to design improved water systems operations that better balance efficiency and justice. More specifically, we explore the extent to which the inclusion of equity principles reshapes the space of efficient solutions. Numerical experiments are conducted on the Lake Como system, Italy, operated primarily for flood control and irrigation water supply while also providing recreation and river ecosystem services. Our results show how incorporating equity considerations into the design of water system operations enriches the solution space by generating more compromise solutions than those obtained using a traditional multi-objective optimization. Moreover, we find that including equity in the operating policy design can indirectly improve the performance of marginalized sectors, such as recreation and ecosystem, which are not explicitly considered by the current lake operation. Lastly, we illustrate how the aggregation of multi-sectoral interests into an equity index strongly shapes our results. Thus, eliciting the preference structure of stakeholders and policymakers becomes paramount for the identification of a fair balance across competing interests. This work bridges the gap between multi-objective optimization approaches and equity-informed decision-making for real-world water resources planning and management, providing an effective tool to promote efficient and equitable policies.

Climate and socio-economic changes bring multiple challenges to river basin development worldwide. The large uncertainty characterizing future conditions requires robust and adaptive planning and management solutions capable of handling uncertain future changes. This is particularly true in monsoonal Southeast Asian catchments, where large multipurpose reservoir systems play a crucial role in flood protection and providing water, energy, and food to a rapidly changing society. In such river basins, high intra-annual and inter-annual hydroclimatic variability, as well as increasing frequency of extreme events, further challenge the management of multi-sector water demands across multiple time scales.In this context, we develop a robust decision-analytic framework for supporting the strategic planning of river basins in monsoonal areas with respect to future changes in water availability and demands. The framework integrates future climate scenarios, including a catalogue of extreme climate events, future water demand scenarios, a high-resolution infrastructure-accounting hydrological model, Topkapi-ETH, and a strategic, operational model to design multiobjective optimal water management policies. We first build climate change driven projections of water availability; second, we apply the optimization engine to select a subset of operation policies optimized based on key selected indicators; and third, we use the spatially distributed hydrological model to evaluate the impact of the chosen policies on a broader set of indicators capturing the spatially distributed impact of dam operations.We focus here on the Red River Basin, a large transboundary river basin in China and Vietnam. In the basin, conflicts among different water uses, such as flood control, hydropower production, agriculture and aquaculture, are expected to increase under the combined pressure of increasing water and energy demands and climate change. A specific focus is given to extreme rainfall events, expected to increase their frequency and magnitude. The framework proposed will allow us to assess the vulnerability of the basin under future scenarios as well as the sustainability and robustness of future river basin development plans in the context of the water-energy-food-environment nexus.