For example I have last 50/100 years of rainfall data for a large country like Brazil or India or Australia. Can I project rainfall for the next 100 years from that data on yearly, monthly, daily and even hourly basis in terms of intensity? What are the methods and models can I use for that?
There is absolutely no reliable method for doing that - not even on an annual basis. Natural variations in rainfall are hopelessly complicated - subjected to a multitude of different factors. Quite possibly, there are cycles in annual rainfall totals [or in their five or ten-year running means] - but which are not at all easy to identify and pin down. And there could well be a long-term trend too - either with rainfall amounts decreasing, or increasing.
I agree with David. Your question seems to suggest that you want to know how much rain will fall on a particular hour, day or even year up to 100 years in the future. This is impossible, so you need to revise your expectations.
The rainfall in Australia is driven by a series of drivers, the most important of which is ENSO, the Indian Ocean dipole and the southern annular mode. Forecasting even monthly rainfall means forecasting these cycles. BoM does produce seasonal forecasts, but these are only over the coming few months. Disaggregating rainfall to a higher temporal resolution is not possible in the sense of estimating the actual rainfall for particular days or hours (see previous discussions on this topic).
All you can look at is a possible 100 year long sequence of rainfall under a future climate, but that is not associated with particular years. Rather this is 100 samples of a possible rainfall under a future climate scenario (e.g. particular CO2 emission between now and 2050). In regard to rainfall intensity, all you really can do is produce a scenario of the probability distribution of rainfall intensity (e.g. changes in the rainfall intensity-duration curves under climate change).
You could use scaling to adjust historical rainfall to give possible future rainfall, though this is problematic as you need to consider how the distribution of rainfall changes across all scales (e.g. relative change in summer versus winter rainfall, changes in the frequency and intensity of rainfall events). The better approach is to use a weather generator to produce a rainfall sequence given estimated statistical properties of the future climate. For example http://www.bom.gov.au/climate/ahead/about/model/access.shtml
Using CMIP5 GCM models and different RCP scenarios you can project futrue rainfall for next 100 years. But for this you need 30 years historical daily weather data like minimum temperature, maximum temperature, rainfall and solar radiation. For more details check out https://agmip.org/. But i dont have any idea of your second question of disintegrating monthly data into daily, or daily rainfall to hourly.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India