Transferability Intercomparison: An Opportunity for New Insight on the Global Water Cycle and Energy Budget

Abstract

Transferability intercomparisons provide a new approach for advancing the science of modeling the water cycle and energy budget on regional to global scales by using multiple limited-area models applied to multiple domains. The water and associated energy cycles introduce exponential, episodic, and other nonlinear processes that create difficulties for observing, simulating, and predicting climate variations. The water cycle both creates and responds to spatial heterogeneities that feed back strongly on the energy budget and circulation system. These feedback processes represent some of the largest uncertainties in our ability to simulate future scenarios of Earths climate, especially scenarios that suggest warming beyond the temperature bounds of recent interglacial conditions and hence for which we have no previous observations for comparison. Water cycle processes also occur on a wide range of spatial and temporal scales, many being far too small to either be globally observed and or simulated by global climate and weather forecast models.Transferability intercomparions represent a new approach for understanding the water cycle and energy budget on regional to global scales. This new class of intercomparisons applies multiple regional climate models to a prescribed collection of domains where enhanced observations are conducted and results are archived in a coordinated manner. The primary goals of the transferability intercomparisons are to understand the complex interactions forming the water cycle and evaluate our ability to simulate these processes. The transferability framework goes

The Court of Miracles of Hydrology: can failure stories contribute to hydrological science?

“…In the context of hydrological modelling, avoiding the one-case-study application should become the rule rather than the exception, as also argued by Andréassian et al (2006). This is one convincing way to make our results more generalisable and our models more reliable (Takle et al, 2007).…”

A preliminary assessment of glacier melt-model parameter sensitivity and transferability in a dry subarctic environment

“…A model is considered transferable if it is able to produce realistic results outside of the domain for which it was developed and tested (Takle et al, 2007). Model transferability is closely related to the concept of model sensitivity, which is a measure of how variation in model output can be attributed to variation in model input (Saltelli et al, 2004).…”

A conceptual framework for multi-regional climate change assessments for international market systems with long-term investments

“…A climate change assessment for international market systems also allows for what Takle et al (2007) refer to as "transferability intercomparisons." Models developed for one region may not be transferable to another region, as the parameters of the functions and models may have been specifically "tuned" to phenomena unique to a region.…”

“…While poor performance when models are applied out-of-sample is a challenge, an industry-wide assessment provides an opportunity for comparing multiple models in multiple regions, which, in contrast to typical model intercomparisons that are limited to either applying multiple models to a single region or a single model to multiple regions, is more likely to expose the limits of current models and model parameterizations and contribute to global model improvement (Takle et al 2007). Furthermore, more confidence can be placed in the transferability of models to future time slices if, based on comparisons for current and historical periods, the models perform well in multiple spatial domains that experience a wide range of conditions, some of which might be reflective of the future expectations for a specific region.…”