I suggest impact analysis or segmented regression.

Hello Jon,

Can you elaborate on your point.

My question was actually about, whether, an extreme events analysis of the simulations from a climate model a good idea or do we always need an observation system based analysis for analyzing and identifying the extreme events.

My experiences are restricted to using observational data - I have never done simulations. The methods I suggested can be use with observational data.

I suggest geoestatistics simulation model.

Hello Jon, 

Yes, you are right. For observational systems, impact analysis is a good approach for studying the effect of extreme events. 

thanks 

Hello Aniket.  I think the questions to answer first are:

• what sort of extreme events are you interested in? (and how extreme?)
• does the climate model simulate these events well?

For example, if you are interested in local wind speeds around a mountain range, then a global climate model at a grid resolution of (say) 50km would not be very realistic or helpful.  But for other variables you may find that the climate model output is a better representation of real-world extreme statistics.

Even in the latter case, however, I would advise the need for an observation-based analysis in the first instance, to be sure that the model is appropriate.

Then, in addition to the model output, you should consider the underlying physical reasons for the changes you observe or predict.

Hello Erica,

Thanks...

I am trying to study the drought events using climate model simulations as there is a lack of long term observation system based continental data sets.

I do understand that, observation system based analysis would be the best way forward in studying drought events, but lack of observations over India, means I have to use model simulations.

I just wanted to know from literature and from discussion if climate modeling systems are able to capture the drought events or not.

So, if you have any literature regarding the limitations of using Climate models for drought predictions, please do share the link.

Pure obsetvstion-based analyseselskapet have obvious limitations when the climate changes. Regional Flow systems may change both wrt position and frequency of occurrence, henne you cannot extrapolate statistical properties in the way implied by calibrating some empirical model by observational data.

Henne, numerical modelling is needed.

But they are far from perfekt, and need to be heavily evaluation over periods with available observations.

Good luck.

One thing that would be sensible to do is to use multiple climate models to get an idea of the range of projections of drought /precipitation changes.

For example, you could look at the same drought index in 

This would give you an idea of how good the climate models are, and also the spread of projections over the next century.  I don't know whether this has already been done though; you would need to search the literature.

Yes I think so because the reanalysis data are averaged for time and space. Numerical model is the only way that we could reproduce the extreme events. However,  the high'resolution model results do not have comparable observed results.

thanks ... based on the above discussion, I think the best way forward is, first, perform an error analysis of the model simulation with respect to observations then use the simulations for detecting extreme events. 

And in the case of lack of in-situ measurements, validated satellite observations should be used.