Weather risk management has traditionally been addressed by creating design standards based on the recurrence rate of extreme events. For example, infrastructure may be designed to withstand a one hundred year storm or flood. The implicit assumption is that larger events are unlikely enough that they can safely be ignored.
Climate change complicates the picture, since what was a one hundred year event historically may be a fifty year event forty years in the future. And, to add further complexity, it may also be a forty year event or a sixty year event as a result of climate uncertainty. Climate uncertainty is due to three factors: the lack of knowledge about what future Greenhouse gas emissions will; the fact that climate models are not perfect representations of the real world; and the natural variability of climate and weather due to their chaotic nature (which simply means that very small differences in initial conditions can result in very large differences in the ultimate outcome and is sometimes referred to as the “butterfly effect”). The uncertainty in future emissions is dealt with by examining several different emission scenarios, while comparing multiple climate projections from multiple models (a multimodel ensemble) is a method frequently used to explore the range of possible climate conditions due to natural variability and model differences.
From a practical point of view, this means that you can’t base your plans on a precise prediction of what the future climate will be. Instead you need to consider a range of possibilities. If the historical one hundred year flood level was 15 feet, for example, flood levels of 19 and 24 feet may both be possible thirty years from now as illustrated by the blue line in the hypothetical graph below.