AgriMetSoft
Ask Question

Difference between Rainfall Anomaly Index (RAI) in MDM/DMAP/RDIT and other tools
It is a conversation between Dr. Musa Esit and Support of AgriMetSoft

First Email from DR. Musa Esit

Hi Dear; My work is over drought analysis as indices, copulas, return period, frequency analysis, and trend analysis. I realized that your MDM software gives to error (incorrect values) data according to R statistical software. Just only I checked to Rainfall Anomaly Index (RAI) by your tool and the R Package, when I realized that so different values, I also tested by hand in excel to find which one is correct. As a result, your RAI index (I looked only this index) give wrong data as I send you my document for you to check both in R and MDM results. I also see that in MDM, for same precipitation data, RAÄ° doesn't give same index value that you will see in Excel, and even your software program calculates one more year data; I don't know why. I am sending my precipitation data, R output in excel, your software output in also excel and a screenshot for checking. Thank you for your polite, if I give any disturb, I am sorry if am I wrong, please provide me with information about this point.

First answer from AgriMetsoft

Hi Thank you for checking our tool. Which formula do you use? Do you use van Rooy (1965)? Where is your manually check? Which R package do you use? Previously a researcher compared SPI in MDM by DrainC, but we check it by other tools. Find it on this page. We reviewed the MDM manually again, and we found that it is true. I calculate RAI by your data, and I cannot find an extra year. Maybe you have a problem with loading data. See the bellow fig

Finally, the output of your R package is not correct, and for example, you have 373 months for 1964-1994, but you should have 372 months. Which month is extra?

Second Email from DR. Musa Esit

Maybe because of selecting a little problem station, I chose a new one station to calculate manually and R (Documentation for package 'precintcon' version 2.3.0) and again with your MDM software and I compared other two indices also.

When I solved manually with you referenced to Van Rooy Equation (1965), I realized that the output of R precintcon and my manually calculate again are entirely similar than your MDM software. I am sending a new station that you will see row data, manually calculate, R output and two other index output.

But I see that when I correlated four indexes by Pearson Kendall and Spearman correlation, the best suitable again with your data to SPI and z index rather than R output, I am also sending correlation analysis. I can't see the code of R precintcon, but when l calculate by manual, my calculation and R precintcon is looking same, Maybe when I calculate, I am missing something from eyes like who create in R program also.

Second answer from AgriMetsoft

We are working on our tool in RAI section around 6 hours (in today). We are sure our tool is entirely accurate. I don't know, why you get a correlation between the outputs of tools. The tools should return similar data, and if it doesn't return similar, it means one of the tools is wrong, and we are sure the R packages is wrong. I checked the excel file, and I found that your mistake is in the calculation of the Ten Highest and Ten Lowest. If you calculate RAI for Jan so you should get an average of Ten Highest/Lowest in all Jans in your periods but you calculated in all months and it is not true. We will add the conversation in FAQ if you don't want to write the name of you, please inform me.

Third Email from DR. Musa Esit

Thank you very so much for giving and attractive to my question, now I understood my mistake because of taking the highest/lowest value not only just all month, must be a specific month, some researcher has published an article using precincton package from another country, I will say to them this wrong about their wrong. Sorry about my detailing researchers because I want to prove which is good or not. Because of my manual calculating in your referenced paper doesn't say how I estimate monthly, when I look a YouTube I found to calculate by excel that he is also doing wrong.

But when I checked indices, I saw that your calculating indices compared to the other is so high. That is showed that you are calculating correctly and I was wondering to my calculating.

Thank you so much again. Nowadays I will publish an article that I will appreciate your company and I will also show your program the other platform even the other people who work at drought, flood, etc.

I took your time, because of this situation, and your interest, thank you a lot.

Third answer from AgriMetsoft

You are welcome. You know, during our projects we used different available tools and packages, we finally find out there are different obscure and uncertain values between the results, so we started to deep focus by the equations in the original papers. Therefore, the developed software tools in Agrimetsoft are yielded from 4-years of hardworking and investigations on various articles. Now, we can state that the published tools in Agrimetsoft are complete and perfect. Furthermore, we welcome to the researchers and scholars that they have innovations, and they want to customize the available tools of Agrimetsoft. We would be satisfied with such collaborations. Appreciate for this conversation, and hope you do the best in your research. Kindest regards, Agrimetsoft

Name: DR. Musa Esit