Uncertainty in models

Uncertainty can be defined as the randomness that is due to the unknown or missing information from the data, parameters, or the process itself. Uncertainty causes underestimation or overestimation of variables leading to results that are way different from the actual (Bosshard et al., 2013; J. Chen et al., 2011; Giorgi & Francisco, 2000; Trzaska & Schnarr, 2014). The process of determining the impacts of climate change on hydrology, based on the GCM outputs includes two main steps: (i) The outputs from the GCM are downscaled to the local scale at which hydrologic impact is measured; and (ii) Projections of climate change are applied to the hydrological model to find out the future hydrological projections. Based on the results of the predicted conditions, a decision maker can make decisions and policies (J. Chen et al., 2011). Understanding the uncertainty and its quantification with the results will help decision-makers to interpret the reliability of the results and to use the results in a probabilistic manner (Deser et al., 2012). It is thus important to understand the uncertainty in each and every process of the impact assessment. The uncertainty can be divided into the following (J. Chen et al., 2011):

  1. Greenhouse gas emission scenario uncertainty
  2. Uncertainty arising from GCM structures and parameters
  3. GCM initial conditions uncertainty
  4. Uncertainty caused by the downscaling methods
  5. Uncertainty due to hydrological model structures
  6. Uncertainty in hydrological model parameters
  1. Greenhouse gas emission scenario uncertainty:
    Uncertainty in GCM simulations can be due to uncertainty in projected greenhouses gas emission scenarios and the response of GCMs to changes in atmospheric forcing, which is associated with the model structure, parameterization, and spatial resolution. Especially at regional scales, the differences between GCMs often result in different climate outputs and the uncertainty from GCMs is generally larger than that from their sources (H. Chen et al., 2012; Graham et al., 2007; Kay et al., 2009; Prudhomme & Davies, 2009). The different emission pathways cannot be given probabilities as we do not yet have a standard method for the same. This is due to the uncertainty in the main causes of emission, such as future world economy (Wilby & Harris, 2006).

  2. Uncertainty arising from GCM structures and parameters:
    Since the beginning of climate modeling, humungous amount of models have been developed world wide by scientists which usually project climate changes according to the scenario provided. But each model is different and they have different sensitivities and they show different patterns of change when projected. Parameterization is another approach used by GCMs when they are unclear about processes happening at a scale lesser than GCM scale. This also contribute to the uncertainty in GCMs (Kay et al., 2009).

  3. GCM initial conditions uncertainty:
    When the same GCM is run using different initial conditions, it is called as ensemble runs.J. Chen et al., 2011 performedensemble runs on MRI model and it was found that with the increase in number of initial conditions, there were significant difference between the run performed.

  4. Uncertainty caused by the downscaling methods:
    Downscaling transforms the coarser-resolution GCM outputs to a finer scale which eventually helps in the assessment of the impact of climate change on water resources at catchment level/watershed level. The systematic biases associated with the different downscaling methods also add uncertainty to climate projections. Broadly, the downscaling methods have been classified into three categories: i) Statistical downscaling ii) Dynamic downscaling iii) Delta method or change factor method (Ekström et al., 2015).Uncertainty in downscaling is mainly due to the scale issues between the GCM that operates at the large scales and the local climate processes that operates at the mesoscale or small scale (Mearns, 1991). Khan et al., 2006 used three different statistical downscaling methods for precipitation downscaling, J. Chen et al., 2011 used 28 climate projections from a combination of seven GCMs and GGEs, six downscaling methods to predict the stream flow in Quebec province. Both studies found that under future climate scenarios, the differences in results produced by different methods are significant but less significant than using different GCM projections. In the future scenarios, the uncertainty of their results would be mostly governed by the uncertainty of the GCM output. It is found that SDSM and change factor methods are best for downscaling, as uncertainties from these methods are in the natural variability limits (Prudhomme & Davies, 2009).

  5. Uncertainty due to hydrological model structures
    From the analysis by J. Chen et al., 2011, it is clear that there are no consistent differences between the two lumped models and the distributed physically based model. The uncertainty is found to be dependent on the choice of model parameters is very low compared to model structure uncertainty (Prudhomme & Davies, 2009).

  6. Uncertainty in hydrological model parameters:
    Sources of the uncertainty in hydrological modeling are resulted from the model itself and parameterization. However, the relative contribution of the uncertainty to the results  are lesser than that to the difference in predictions. Recent studies comparing the sources of uncertainty suggest that the hydrological models have a relatively minor impact on the results of hydrological simulations driven with climate projections (Kay et al., 2009; Wilby & Harris, 2006), but can vary significantly between catchments (Prudhomme & Davies, 2009). It is evident that the hydrological modeling only reflects changes to runoff to changes in climate inputs without considering potential changes in the rainfall-temperature- runoff relationship in a warmer and higher CO2 environment.

Uncertainty Quantification:

The analysis of uncertainty has been done on the basis of the statistics and statistical methods majorly. The appropriate model chosen by ‘validation’ study, in which the goodness of fit of the model is assessed by comparing the results of the model with observed data. The importance of structural uncertainty analysis is in knowing when and where a model is to be applied to produce reasonable results, more importantly, where the model is bound to fail (Samadi et al., 2013). Uncertainties in climate model projections have emerged from a combination of various socioeconomic scenarios and climate models. Decision-makers and scientists alike have expressed a desire to assign probabilities to each scenario so that there is a better sense of whether certain scenarios are more likely than others (Quiggin, 2008). Further, probabilities of various scenario occurrences will likely even change as soon as a prediction is made because society begins to react and therefore changes the outcome in ways that the prediction did not incorporate (Sarewitz et al., 2003). Therefore, there is a strong need for studying aspects of the uncertainty of climate projections in order for scientists and practitioners to know how to characterize and treat uncertainties.

References

Bosshard, T., Carambia, M., Goergen, K., Kotlarski, S., Krahe, P., Zappa, M., & Schär, C. (2013). Quantifying uncertainty sources in an ensemble of hydrological climate-impact projections. Water Resources Research, 49(3), 1523–1536. https://doi.org/10.1029/2011WR011533

Chen, H., Xu, C. Y., & Guo, S. (2012). Comparison and evaluation of multiple GCMs, statistical downscaling and hydrological models in the study of climate change impacts on runoff. Journal of Hydrology, 434–435, 36–45. https://doi.org/10.1016/j.jhydrol.2012.02.040

Chen, J., Brissette, F. P., Poulin, A., & Leconte, R. (2011). Overall uncertainty study of the hydrological impacts of climate change for a Canadian watershed. Water Resources Research, 47(12), 1–16. https://doi.org/10.1029/2011WR010602

Deser, C., Phillips, A., Bourdette, V., & Teng, H. (2012). Uncertainty in climate change projections: The role of internal variability. Climate Dynamics, 38(3–4), 527–546. https://doi.org/10.1007/s00382-010-0977-x

Ekström, M., Grose, M. R., & Whetton, P. H. (2015). An appraisal of downscaling methods used in climate change research. Wiley Interdisciplinary Reviews: Climate Change, 6(3), 301–319. https://doi.org/10.1002/wcc.339

Giorgi, F., & Francisco, R. (2000). Evaluating uncertainties in the prediction of regional climate change. Geophysical Research Letters, 27(9), 1295–1298. https://doi.org/10.1029/1999GL011016

Graham, L. P., Andreáasson, J., & Carlsson, B. (2007). Assessing climate change impacts on hydrology from an ensemble of regional climate models, model scales and linking methods – A case study on the Lule River basin. Climatic Change, 81(SUPPL. 1), 293–307. https://doi.org/10.1007/s10584-006-9215-2

Kay, A. L., Davies, H. N., Bell, V. A., & Jones, R. G. (2009). Comparison of uncertainty sources for climate change impacts: Flood frequency in England. Climatic Change, 92(1–2), 41–63. https://doi.org/10.1007/s10584-008-9471-4

Khan, M. S., Coulibaly, P., & Dibike, Y. (2006). Uncertainty analysis of statistical downscaling methods. Journal of Hydrology, 319(1–4), 357–382. https://doi.org/10.1016/j.jhydrol.2005.06.035

Mearns, O. (1991). Approaches Regional To the Simulation Change : of Climate a Review. Reviews of Geophysics, 90, 191–216.

Prudhomme, C., & Davies, H. (2009). Assessing uncertainties in climate change impact analyses on the river flow regimes in the UK. Part 1: Baseline climate. Climatic Change, 93(1–2), 177–195. https://doi.org/10.1007/s10584-008-9464-3

Quiggin, J. (2008). Uncertainty and Climate Change Policy. Economic Analysis and Policy, 38(2), 203–210. https://doi.org/10.1016/S0313-5926(08)50017-8

Samadi, S., Wilson, C. A. M. E., & Moradkhani, H. (2013). Uncertainty analysis of statistical downscaling models using Hadley Centre Coupled Model. Theoretical and Applied Climatology, 114(3–4), 673–690. https://doi.org/10.1007/s00704-013-0844-x

Sarewitz, D., Pielke, R., & Keykhah, M. (2003). Vulnerability and risk: Some thoughts from a political and policy perspective. Risk Analysis, 23(4), 805–810. https://doi.org/10.1111/1539-6924.00357

Trzaska, S., & Schnarr, E. (2014). Methods for climate change impact assessment. September.

Wilby, R. L., & Harris, I. (2006). A framework for assessing uncertainties in climate change impacts: Low-flow scenarios for the River Thames, UK. Water Resources Research, 42(2), 1–10. https://doi.org/10.1029/2005WR004065

Scroll Up