Q3 uncertainties in hydrology and how can they be mitigated?

Question 1

What are the principle uncertainties in hydrology and how can they be mitigated?

Answer 1

Intro
- There are many sources of uncertainty in hydrological systems, even if we think only about water and its flow pathways.

Uncertainties in Data

• The most important, are uncertainties in hydrological data, since these are basic inputs to any hydrological analysis or modelling application. Basic to and hydrological analysis is the water balance equation:

Inputs = Discharge + Actual evaporation + change in storage + other potential fluxes

• each term uncertain in itself (measurement)
• rainfall, evapotranspiration and change in storage require interpolation from point measurement to catchment scale
• snow accumulation and melt hard to predict
• The other potential fluxes term, such as losses into deeper ground water systems, are often neglected because they cannot be measured at all
• For example, it would be very useful to have a measurement technique that would allow an assessment of the incremental discharges over successive (short) stream reaches. Where this has been done in small catchments using tracer dilution methods, it has revealed some very interesting results
• Tracer not viable for large scales
• Ultransonic only accurate to 5%
• Remain important for the foreseeable future

Uncertainties in Models

• uncertainties feed into models, both in the calibration of the model parameters and the way uncertainties in input data affect the model prediction
• In the past it has been common to optimise model parameters in calibration and use only the ‘best’ model found in prediction. This is even more common practice today, although increasingly some attempt is made to evaluate the prediction uncertainties around the outputs from the ‘best’ model
• We should, in fact, expect the predictions to be uncertain because of the difficulty of representing the complexity of hydrological processes and the input uncertainties
• Agencies, including in the UK, are moving towards risk based management strategies and decision making that seek to assess the probabilities and consequences of what can sometimes be a wide range of possible alternative scenarios generated using hydrological (and other) models

Uncertainties is decision making

• If uncertainties in data and model predictions are significant, then they may affect decision-making within a risk based management context
• There are a number of different ways of making a decisions of ranking alternative strategies within a risk- based framework where the uncertainties can be qualified (again see bevan),
• but it may be the case that in some applications the uncertainties may be so large that another type of decision assessment might be appropriate, e.g. by applying the precautionary principle
• Flood damages in all countries in the developed world, for example, would be must less if land planners had been more precautionary about development in flood plains. In the UK alone it has been estimated that property worth some £800 billion, including 10% of the housing stock, is at risk of flooding. The Uk summer floods of 2007 led to insurance claims of some £4 billion (Environment agency, 2010).
• An important principle in making decisions in the face of uncertainty is the concept of robust decision. This is a decision that allows a reassessment of strategy over time but which does not preclude a change to another option as a result of that reassessment. Robustness in this sense is an important part of adaptive management, leaving possible option open in case the characteristics of the system change in unexpected ways.

Engineering applications

1) Flood risk management
2) Water resource management
3) Urban Hydrology
4) Hydrology, Climate and Catchment Change

Conclusion

• increasing importance due to knock on effect
• data errors look to remain important for the near future
• statistical and risk models must be used to control scope of error and its affect