Hello Majid, there is a mathematical method by which the return period and the probability of occurrence are estimated using a mathematical law. In addition, there are statistical methods by which rain can be estimated in the future.

Dear Mr. Aljubouri,

Thank you for your reply. Could you please name the mathematical methods and statistical methods commonly used for this?

Hi, Mr. Majid I think you can use Cliodynamics mathematical modelling for future predictions from the past...

Dear Majid

I think that it is important to note that tend analysis based on historical data is often not the most appropriate way to assess future values of rainfall intensity, or indeed other climate variables for future climate change scenarios. Trend analysis is normally done by fitting a line, (e.g. via a regression analysis), to historical data and extrapolating this into the future. The use of trend analysis has a number of disadvantages in the consideration of long-term climate change estimates. In the context of climate change, the use of Global Climate Models (GCMs) is considered a more appropriate means to estimate the values of variables affected by climate change, such as rainfall, especially for many years in the future (e.g. up to the year 2100), than the use of trend analysis. Trend analysis may provide a useful source of additional information to investigate recent and near term future changes ( 20 years) or long term estimates of climate variables such as rainfall

The figure attached, from Intergovernmental Panel on Climate Change (2007) Fourth assessment report, shows the global mean temperatures from the year 1850 to about 2005. This graph shows the dangers of fitting trend lines, especially linear ones, to historical data in order to assess the future impacts of climate change. In The figure attached shows linear trend fits to the last 25 (yellow), 50 (orange), 100 (purple) and 150 years (red) are shown, and correspond to the following periods: 1981 to 2005, 1956 to 2005, 1906 to 2005, and 1856 to 2005, respectively. It is important to note that for trends fitted to shorter recent periods, the slope is greater, indicating accelerated warming.

To be meaningful, a statistical method needs to represent adequately the underlying physical processes simple trend analysis as some people have carried out for rainfall gauges around the world will not do this. GCMs take account of physical processes. GCMs and historical data indicate strongly that future long term changes in climate are very unlikely to be a smooth line and that the climate is a complex system with non-linear behaviour which will affect future rainfall intensities.

I hope that this is of help.

Darren

Dear Majid,

You can look at using machine learning models such as the LSTM or GRU which are suitable for long-term forecasting. There are numerous papers that have applied these machine learning methods which have very good results.

There is an easy to follow online tutorial on how to build these models in the link below:

https://machinelearningmastery.com/how-to-develop-lstm-models-for-time-series-forecasting/

I hope this can help you.

Collins

Majid,

I would use the CLIGEN software that accompanies WEPP to do this. You can build a statics file from your observed data, and then use CLIGEN to generate a long term weather sequence using CLIGEN.

Hi Majid, You can look at our latest work on the Uttrakhand region. We have used a high-resolution rainfall dataset to assess the century-long trend in the Upper Ganga Basin. I would be happy to provide the full text of this manuscript if it is not accessible to you. Thank you.

Article Spatio-temporal rainfall trends in the Ganga River basin ove...

Dear Majid

By using the tend analysis based on historical data but if the data are changing dramatically then use multiple trends for different periods or decades in order to see variables for future climate change scenarios.

Regards

Darren you are very right and I am in agreement that trend lines may give some first order near time forecast. But all this depends wgst us the science question Majid you want to address. If you are forcasting for flood management etc, croping season then such forecast will be oquite misleading.

Darren you are very right and I am in agreement that trend lines may give some first order near time forecast. But all this depends what is the aim. If you are forcasting for flood management, croping season then such forecast will be quite misleading. If you are looking st aquifer recharge may be you can use such rough estimates.

Darren you are very right and I am in agreement that trend lines may give some first order near time forecast. But all this depends what is the aim. If you are forcasting for flood management, croping season then such forecast will be quite misleading. If you are looking st aquifer recharge may be you can use such rough estimates.

Darren you are very right and I am in agreement that trend lines may give some first order near time forecast. But all this depends what is the aim. If you are forcasting for flood management, croping season then such forecast will be quite misleading. If you are looking st aquifer recharge may be you can use such rough estimates.

Hi,

There are straightforward statistical relationships between rainfall (as a dependent variable) and some meteorological parameters such as temperature, relative humidity, and air pressure (as independent variables). In my opinion, these relationships can be modeled by using statistical techniques such as multivariate regression, data mining algorithms based on association rules, and machine learning approaches. In this way, using the data related to the independent variables in the constructed model, the rainfall can be estimated.

Regards.

Hi Majid

Search about ARIMA key word

Also you can use Regression Learner toolbox in MATLAB that provides you many methods such as SVM for estimating rainfall as an dependent variable and some independent variable which have high correlation with rainfall.

you can go though the Stochastic Analysis Modeling and Simulation (SAMS), and use the application of stationary Auto-regressive Moving Average (ARMA) models which have been widely used, and generate multi- ensembles of the Future data, you should be careful when use it to generate rainfall data, that you should put a boundary (for the maximum and minimum values ) especially if you use it in arid area.

This article provides relevant information to understand the steps to perform an impact study. Supplementary material includes a guide to perform some key tasks leading to the production of hydroclimatic projections.

Chapter Hydrological Modeling of Climate Change Impacts

you can use several Machine Learning (ML) models