Manantali is a dam located on the Senegal River and is mainly used for hydropower production. Before the dam's construction, the annual river flood alimented the flood recession agriculture, a practice based on natural irrigation and fertilization of the flood plain, used traditionally by the local populations downstream. Analysis of the actual reservoir operation shows that annual floods have been largely reduced for the benefit of hydropower production. Moreover, the Senegal River Basin authority is evaluating the construction of different new dams, which could reduce even further the water available for flood support, given that the current operational focus is on satisfying hydropower demand. This study investigates the effects of an optimal reservoir operation strategy that maximizes hydropower production only, analyzing the results of this strategy in terms of effects on the two main objectives, i.e., hydropower production and flood support. The problem of finding optimal reservoir operation strategy is solved by applying the stochastic dual dynamic programming method. Results show the existence of a release strategy in which both objectives improve (+9% for hydropower and +7% for flood production) with respect to the historically observed operation. This solution, however, may require the electric system to compensate for the variability in energy supply along the year.

@en

The basin of the Seine River is an extremely important economic region for France and Europe. Four reservoirs are operated to reduce the natural variability of the Seine River, reducing both flood and drought risk. Presently, reservoir operation is not centrally coordinated, and release rules are based on empirical rule curves. This study presents the setting of an optimal and centralized solution to the problem of reservoir operation on the Upper Seine-Aube river system, found by applying the stochastic dual dynamic programming (SDDP) procedure. The novelty of this study lies on the combination of reservoir and hydraulic models in SDDP for flood and drought protection. Including the hydraulic process in SDDP is required for estimating flood and drought at different locations along the river, and for representing the delay between the release from the reservoirs and their effects downstream. The study case covers the Seine basin until the confluence with the Aube River: this system includes two reservoirs, the city of Troyes, France, and, at the confluence of the two rivers, the nuclear power plant at Nogent-Sur-Seine. Results shows that the SDDP solution can be effectively used to optimize the operation of a water system made of multiple reservoirs and multiple hydraulic transfer components, solving a relatively large stochastic dynamic programming problem in an acceptable time. The management obtained from SDDP rules exploits the centralized operation and, compared to the current operational rules, results in more frequent but shorter, less intense, and less severe flood and drought events at Nogent-Sur-Seine.

@en

Climate change raises serious concerns for policymakers that want to ensure the success of long-term policies. To guarantee satisfactory decisions in the face of deep uncertainties, adaptive policy pathways might be used. Adaptive policy pathways are designed to take actions according to how the future will actually unfold. In adaptive pathways, a monitoring system collects the evidence required for activating the next adaptive action. This monitoring system is made of signposts and triggers. Signposts are indicators that track the performance of the pathway. When signposts reach pre-specified trigger values, the next action on the pathway is implemented. The effectiveness of the monitoring system is pivotal to the success of adaptive policy pathways, therefore the decision-makers would like to have sufficient confidence about the future capacity to adapt on time. "On time" means activating the next action on a pathway neither so early that it incurs unnecessary costs, nor so late that it incurs avoidable damages. In this paper, we show how mapping the relations between triggers and the probability of misclassification errors inform the level of confidence that a monitoring system for adaptive policy pathways can provide. Specifically, we present the "trigger-probability" mapping and the "trigger-consequences" mappings. The former mapping displays the interplay between trigger values for a given signpost and the level of confidence regarding whether change occurs and adaptation is needed. The latter mapping displays the interplay between trigger values for a given signpost and the consequences of misclassification errors for both adapting the policy or not. In a case study, we illustrate how these mappings can be used to test the effectiveness of a monitoring system, and how they can be integrated into the process of designing an adaptive policy.

@en

Adaptive policies have emerged as a valuable strategy for dealing with uncertainties by recognising the capacity of systems to adapt over time to new circumstances and surprises. The efficacy of adaptive policies hinges on detecting on-going change and ensuring that actions are indeed taken if and when necessary. This is operationalised by including a monitoring system composed of signposts and triggers in the design of the plan. A well-designed monitoring system is indispensable for the effective implementation of adaptive policies. Despite the importance of monitoring for adaptive policies, the present literature has not considered criteria enabling the a-priori evaluation of the efficacy of signposts. In this paper, we introduce criteria for the evaluation of individual signposts and the monitoring system as a whole. These criteria are relevance, observability, completeness, and parsimony. These criteria are intended to enhance the capacity to detect the need for adaptation in the presence of noisy and ambiguous observations of the real system. The criteria are identified from an analysis of the information chain, from system observations to policy success, focusing on how data becomes information. We illustrate how models, in particular, the combined use of stochastic and exploratory modelling can be used to assess individual signposts, and the whole monitoring system according to these criteria. This analysis provides significant insight into critical factors that may hinder learning from data. The proposed criteria are demonstrated using a hypothetical case, in which a monitoring system for a flood protection policy in the Niger River is designed and tested.

@en

Uncertainty reduces reliability and performance of water system operations. A decision-maker can take action, accepting the present uncertainty and facing its risks, or reduce uncertainty by first obtaining additional information. Information, however, comes at a cost. The decision-maker must therefore efficiently use his/her resources, selecting the information that is most valuable. This can be done by solving the Optimal Design (OD) problem. The OD problem balances the cost of obtaining new information against its benefits, estimated as the added value of better informed actions. Despite its usefulness, the OD problem is not widely applied. Solving the OD problem requires a stochastic hydro-economic model, whereas most of hydro-economic models presently used are deterministic. In this paper we introduce an innovative methodology that makes deterministic hydro-economic models usable in the OD problem. In deterministic models, Least Squares Estimation (LSE) is often used to identify unknown parameters from data, making them, and hence the whole model, de facto stochastic. We advocate the explicit recognition of this uncertainty and propose to use it in risk-based decisions and formulation of the OD problem. The proposed methodology is illustrated on a flood warning scheme in the White Cart River, in Scotland, where a rating curve uncertainty can be reduced by extra gaugings. Gauging at higher flows is more informative but also more costly, and the LSE - OD formulation results in the economically optimal balance. We show the clear advantage of monitoring based on OD over monitoring uniquely based on uncertainty reduction.@en

Dams can produce electricity and ensure water security, but at the same time they radically alter the hydrological regime of rivers with significant consequences for the economic and environmental welfare of the region in which they are located. Cost-benefit analysis (CBA) is currently the most frequently used framework for the economic evaluations of dams. Changes at different time scales influence the economic appraisal of dams. However, change and adaptation at both the operational and the structural level are often not included in the CBA evaluation. Not including change and adaptation limits the realistic estimation of cost and benefits, and the appreciation of resilient solutions that offer satisfactory responses for a large set of future scenarios. In this paper we consider the specific features of large dams in an African context, and identify methods for an economic evaluation that takes into account for change and adaptation at both the operational and the structural scales, as well as their interplay. These methods are then applied to the ex-ante evaluation of a system of existing dams on the Senegal River Valley. Results indicate the economic potential of the dams under changing conditions, for both adaptive and non-adaptive reservoir operation strategies.

@en

Despite all research efforts, climate change remains unpredictable in the long term. Climate change unpredictability  raises serious concerns for policymakers that prepare long-term policies. To guarantee that climate adaptation  policies perform satisfycingly despite the unavoidable uncertainty, adaptive policies might be used. Adaptive  policies are designed to be adapted over time in response to how the future is actually unfolding. Signposts are  used to track exogenous developments that critically affect the performance of a plan, and when pre-specified  triggers are reached the policy is adapted. This so-called monitoring system is used to gather evidence of change,  on the basis of which a policy is adapted. The ability of the monitoring system to effectively detect change is  pivotal to the success of the whole adaptive policy.  A decision-maker would like to have sufficient confidence about the future capacity to adapt the policy on  time. On time means adapting the policy not too early for that incurs unnecessary costs, nor too late which  incurs avoidable damages. The capacity to adapt the policy on time is to be tested before the adaptive policy is  implemented. Despite the substantial and growing attention by researchers for supporting the making of climate  adaptation decisions under uncertainty, the ex-ante evaluation of the degree to which the monitoring system  enables timely adaptation has been largely overlooked.  We present innovative graphical instruments to assess ex-ante the level of confidence that a monitoring system for  adaptive policies will offer at the moment at which adaptation is required: the Trigger-Power-Significance (T-PS)  plot and the Trigger-Consequences (T-C) plot. These instruments explore the interplay between the possible  trigger values of a given signpost and the level of confidence about the need for adapting the policy. We illustrate  how these instruments can be used to test the effectiveness of a monitoring system, and how they can be integrated  into the process of designing an adaptive policy.  The use of the proposed instruments and approach are demonstrated using a case study of designing an  adaptive policy for coastal flood protection in the Netherlands. This application shows the effectiveness of these  instruments in evaluating ex-ante the capacity of a monitoring system to provide the information required to adapt  on time with sufficient confidence, and redesign it if this is not the case.  @en

The Upper Senegal River, West Africa, is a poorly gauged basin. Nevertheless, discharge predictions are required in this river for the optimal operation of the downstream Manantali reservoir, flood forecasting, development plans for the entire basin and studies for adaptation to climate change. Despite the need for reliable discharge predictions, currently available rainfall-runoff models for this basin provide only poor performances, particularly during extreme regimes, both low-flow and high-flow. In this research we develop a rainfall-runoff model that combines remote-sensing input data and a-priori knowledge on catchment physical characteristics. This semi-distributed model, is based on conceptual numerical descriptions of hydrological processes at the catchment scale. Because of the lack of reliable input data from ground observations, we use the Tropical Rainfall Measuring Mission (TRMM) remote-sensing data for precipitation and the Global Land Evaporation Amsterdam Model (GLEAM) for the terrestrial potential evaporation. The model parameters are selected by a combination of calibration, by match of observed output and considering a large set of hydrological signatures, as well as a-priori knowledge on the catchment. The Generalized Likelihood Uncertainty Estimation (GLUE) method was used to choose the most likely range in which the parameter sets belong. Analysis of different experiments enhances our understanding on the added value of distributed remote-sensing data and a-priori information in rainfall-runoff modelling. Results of this research will be used for decision making at different scales, contributing to a rational use of water resources in this river.@en

This paper presents a new methodology to generate a tree from an ensemble. The reason to generate a tree is to use the ensemble in multistage stochastic programming. A correct tree structure is of critical importance because it strongly affects the performance of the optimization. A tree, in contrast to an ensemble, specifies when its trajectories diverge from each other. A tree can be generated from the ensemble data by aggregating trajectories over time until the difference between them becomes such that they can no longer be assumed to be similar, at such a point, the tree branches. The proposed method models the information flow: it takes into account which observations will become available, at which moment, and their level of uncertainty, i.e. their probability distributions (pdf). No conditions are imposed on those distributions. The method is well suited to trajectories that are close to each other at the beginning of the forecasting horizon and spread out going on in time, as ensemble forecasts typically are.

@en