Step 8: Analyze Projections
Once the relevant climate change projections are constructed, they need to be understood and interpreted.
In many cases, the purpose of this step is to increase understanding of the possible reasons for the projected changes and to inform the confidence in a particular set of projections. For example, if the rainfall is projected to decrease you may ask whether this projection is consistent, for instance, with the observed trend in rainfall in the region, or with results from previous projections work, or with projected changes in atmospheric circulations influencing rainfall.
There are many ways to accomplish this.
Evaluate the magnitude of the change values relative to the natural variability. This can be done by comparing the projected changes against the natural internal model variability for the historical period (i.e. without changes in the concentration of atmospheric greenhouse gases and aerosols as prescribed by RCP scenarios).
Compute the significance of the changes. Even though there might appear to be a change, if the results are not significant in the statistical sense, they may not need to be considered by the information providers and/or the users.
Evaluate the large-scale drivers of the change. While just looking at projected changes to a variable at a location may be relevant for the application, a greater understanding of the changes can be gained by considering the key drivers of the change (which might explain the causes of the projected changes). For example, changes in rainfall in the South East Asia region are largely dominated by the monsoon, so investigating the projected changes in the monsoon may help explain the changes in the rainfall. If your resources are limited, this could be conducted through literature review.
Evaluate the rationality of the projected changes. Due to lack of precision in models, the location of some change features may not be precise, since the inexact locations can possibly cause different change signals. Caution is needed, especially in regions where projected changes vary within the region. Investigating and evaluating the projected changes over the region surrounding the actual location of interest can help build confidence.
Consider all variables comprehensively. If multiple variables are used in the application, the changes in these variables should not be assessed independently of each other. For example, assuming there is a negative correlation between annual rainfall and temperature in your area, you may theorize that the models that project an increase in annual rainfall will likely project lower warming, while models which project decreases in annual rainfall will likely project larger increases in temperature for your area. If your results do not follow this simple logic, then you need to check whether this is what the climate model is telling you or if it is the result of an error in your calculation.
Be aware that one cannot create a scenario for an impact assessment by combining the largest temperature increases from one model projection with the greatest increases in rainfall from another. When supplying model projections for an application or impact assessment, all variables from one climate model need to be used together. As emphasized in step 6, one should not combine variables from different models.
Compare the projected trends against observed trends. Greater confidence in the projected changes is gained if both the observed trends and the projected trends are similar. If the model projections are not consistent with the observed trends, one could understand the underlying drivers of the local trends so that one may have more confidence in the model results. One needs to carefully assess the impact of model internal variability on the trends before drawing conclusions.
Generate additional information if needed. In some cases, climate projection information needs to undergo additional analysis prior to use by impact assessment models. For example, projected changes in mean rainfall can be caused by changes in the frequency of days with rain, changes in low-intensity rain days, and/or changes in extreme rain events. Determining which factor applies will help you provide the end user (e.g. impacts modeller) with more detailed and useful information.
CASE STUDY EXAMPLES
