Possibilities for PhD in Hydroinformatics

The Hydroinformatics Chair Group of the UNESCO Institute for Water Education (IHE Delft) is presenting four main themes that potential PhD students may want to consider as the basis for a PhD study. IHE Delft, Delft, the Netherlands, is a world-leading education and research institute organising Master of Science and PhD Programmes.

Theme 1 

Application areas include rivers, river basins, urban applications, floods in rural and urban areas, coastal and groundwater resources with the overall objective of better management practices. Special attention will be given to taking into account explicit considerations of risk and uncertainty, especially important in climate change studies. We would like to explore novel approaches and developing new integrative modelling frameworks including multi-model and ensemble modelling, data assimilation, integrating remote sensing and other data sources, meteorological and climate models, hydrological and hydraulic models, GIS-based hazard mapping tools and decision support systems. Set of the modelling tools with which we have experience includes Mike11, Mike 21, Sobek, Delft 3D, InfoWorks, ISIS, MOUSE, SWMM, EPANET, HEC-HMS, HEC-RAS, HBV, SWAT, MikeSHE, MODLFOW, various GIS platforms, etc.

Theme 2

Research into and development of appropriate data mining (machine learning, soft computing, data-driven modelling) methods and tools, and their application in water-related problems. The following technologies gain special attention: artificial neural networks, fuzzy rule-based systems, support vector machines, committee machines (ensemble modelling), uncertainty analysis, non-linear dynamics (chaos theory) etc. Special attention will be given to the integration of data-driven and simulation (process) models (leading to hybrid models).

Theme 3

Further development and application of risk-based robust multi-objective optimisation techniques and model-based real-time control methods in solving practical water-related problems. Applications may include reservoir optimisation and control, river basin management, groundwater remediation, urban water networks optimization, control and rehabilitation, model calibration and uncertainty analysis. I optimization methods main interest is to increasing efficiency of randomized adaptive search and genetic and evolutionary methods. Use of data-driven surrogate models replicating complex computationally intensive simulation models in optimization loops, and use of cloud computing for parallelization of complex tasks is also foreseen.

Theme 4 

Development of collaborative decision support frameworks for communities of practice and networks of stakeholders focussed on particular aspects of water management, such as modelling, forecasting, warning, monitoring, optimization, planning, etc. These frameworks would be model-based, use Internet, mobile telephony and other ICT technologies, and support the best management practices.

The Hydroinformatics Chair Group has experience in all of these areas, so that the experiences and techniques can be applied complementary to each other. Staff members have also been involved in a number of practical civil engineering projects where such tools were applied in combination with other computer-based technologies, such as information and knowledge systems, interactive Web portals and GIS.

Some possible PhD topics:

Topic 1: Data fusion

The use of modelling and forecasting in the management of natural resources is ubiquitous. The scarcity of data (e.g. rainfall data) often limits the calibration, validation and subsequent usage of hydrological models. Satellite based rainfall estimates (such as from Tropical Rainfall Measuring Mission) and rainfall forecasts from numerical weather prediction models (such as from ECMWF) serve as alternative data sources, and are increasing being used in hydrological models. Recently, data from social/ human sensors, i.e. people who collect information (e.g. rainfall) with smartphone, also are being available as another potential source of information. However, data coming from diverse sources are collected at different space and time scales, and their accuracy may vary if these scales are changed. Merging data from diverse sources requires the development of a data fusion methodology (based on e.g. statistical approaches) that allows combining diverse data at different space and time scales with an estimated error level. Such merged data can be subsequently used in hydrological simulation for improved forecasting capacity. Merged data may also increase the uncertainty of model prediction and therefore, an assessment of the uncertainty of model predictions should be carried out.

Topic 2: Assessment of flood risks and adaptation measures under global changes

The risk from flooding is computed by combining hazard, exposure and vulnerability. The risk can be increased/ decreased by altering any of these three components. Climate change, population growth and economic developments, particularly in the developing countries, increases flood risk. This requires adopting/ updating adaptation measures so that risks from flooding can be managed. The study will require developing hydrological models, carry out frequency studies, develop hydraulic models to generate inundation patterns and compute risks. Changes in population and economic developments need to be predicted and used together with Assessment Report of IPCC to forecast future changes in risks. Various adaptation measures (e.g. spatial planning) and their effect on reducing the risks need to be assessed.

Topic 3: Model based averaging

Rainfall runoff models are frequently used to simulate runoff from known, synthetic or design rainfall data. Such models can be deterministic or stochastic, or, physically based or lumped conceptual models. Due to the complexity of the process involved and the simplistic approach adopted in some of the models often different models provide very different runoff of a catchment. If measured data is available then an analysis of accuracy of the different models can be investigated but still over a long period of time the preference for a model over the others cannot be easily concluded. Moreover, at ungauged locations we are often unsure of which model result to use. A way out is seen in integrating outputs from several models. The objective of the research is to compare several methods of model combination (including the Bayesian model averaging), and to develop a methodology to combine multiple predictions of runoffs by multiple models. An additional objective will be to carry out an uncertainty analysis of the runoff prediction. 

Topic 4: Coastal flood inundation prediction

A number of coasts around the world, such as of Bangladesh, suffers inundation from coastal flooding. The objective of this study to develop a coastal flood inundation prediction system. The study will require the development of a coastal storm surge model to forecast storm water level. A wave model will be used to forecast storm surge level on the top of astronomical tide. Two complementary components may be developed as well. First, a data-assimilation component, based on a non-linear filtering method (particle filtering or Kalman filtering) to improve the predicted storm surge level by updating the system states with measurements. Second, a database of model pre-runs will be built on the basis of realistic hydrometeorological scenarios and past events, and in operation nearest neighbour or a similar approach will be developed. It will aggregate the data from ensembles of previous storm events and corresponding model forecasts to generate fast forecasts of the storm water levels and inundations in real-time. The forecast storm surge level will be used as a boundary of a flood inundation model, which will simulate the onrush of storm surge water inland. The model will provide forecast flood depths at designated locations along with a spatial flood inundation map.

Funding 

There are various possibilities for funding PhD studies along these themes. We specially invite PhD candidates from developing countries and countries in transition with the opportunities to fund there studies from their governments or universities. We also regularly acquire research projects where a PhD student can be placed, however such opportunities are limited. There is also a possibility to apply for funding from the Dutch Fellowship programme

Requirements for doing PhD in Hydroinformatics

Please be aware that for admission for a PhD programme in our Chair group we request a a candidate to have:
- strong CV, and the motivation letter, with the demonstrable ability to do research
- strong PhD proposal matching the research programme of the Chair group - to be reviewed and approved by the promoter and the mentor
- excellent oral and writing English skills
- reasonably good knowledge of (some aspects of) modelling, data analysis, computer programming and ICT. 
- an interview with the promoter and the mentor (including the presentation of the candidate, and the discussion of the proposal - could be by teleconference or Skype) - this is decisive for recommending admission
- clear understanding where funding will be coming from (approx. 130,000 Euros). IHE Delft does not have readily available funds to support PhD studies - we need to find a sponsor. 

If you think all the requirements above can be met, please check our web site for the details of the formal application (in the application form, please mention name of D.P. Solomatine as the possible promoter, and the Hydroinformatics Chair group as the placement). 

You can find more information here: https://www.un-ihe.org/phd-programme