Dear Subash, 

Weather observatories in India record daily rainfall at 08:30 am.

Total rainfall for the preceding 24 hours ending at 08.30 am of the recording date of the measurement is recorded as daily rainfall. It should be taken as 24-hour rainfall recorded on dd/mm/yyyy at 08:30 AM.

For example: Rainfall received from 08: 30 AM 10/07/2016 to 08:30 AM 11/07/2016 is recorded on 11/07/2016. 

IMD never reports the 24 hour rainfall as today's or yesterday's rainfall. It is just the rainfall recorded at 08:30 AM on some random day.

Though you need not to but you can visit your nearby IMD observatory sometime to verify this.

--------------

Therefore to your doubt "The cumulative rainfall from yesterday 7-July-2016 8:30 AM to today 8-July-2016 8:30 AM if 50mm is it for yesterday or today?"

It is neither today's rainfall nor yesterday's rainfall, It is 50 mm rainfall received from 7-July-2016 8:30 AM to 8-July-2016 8:30 AM. 

I am not sure about WMO guidelines but this is how IMD weather observatories record rainfall. 

To add more, places like Cherapunji where it rains cats and dogs these measurements are made much frequently; sometime at 3-hour interval instead of 24 hour to avoid overspill from raingauges.

Edit: Reference Line 178-183

Preprint Does the GPM mission improve the systematic error component ...

The common standards that I sm aware of for a day, is midnight to midnight, standard time for your time zone.  So if taking last hourly reading in a day, it is midnight, and first reading of new day is 0100 hours, or 1:00 am if you prefer.  However if your country has decided to use international time with readings from 0830 yesterday to 0830 today, then I would report it as yesterday, as most of the time was yesterday.  But this system is confusing, and if you are using International time, then I would report that and use it.  

I am not familiar with the specific document mentioned, but it is likely the three hour data was developed from more detailed readings, so I would first contact the authors to see if available.  If not available, the option is simply take the three hour readings and divide by 3 for a uniform intensity.  

I understand the need to compare data with uniform units, and suggest as presented, you combine your data into 3 hour reading to compare with the 3 hour data in TRMM-3B42 data.  

The last part of question suggests that the TRMM data uses yet another time period.  I dont know reasoning, but if you goal is comparing with that data, you might want to do likewise and adjust your data days to be consistent with the TRMM days -3 hour readings unless you can acquire the TRMM data at hourly intervals to match your data. Whatever you decide to do, make it clear what you did so those who look at your findings or follow you will know what and how you adjusted data in making the comparison.  

Using Aggregation method.

I agree with Osama by using Aggregation method.

Rainfall 830Am yesterday to today 830am is cast back as yesterdays rain. Use Aggregation.

You may use CDO (Climate Data Operator) to compute "daysum" from sub-daily files. It would be the easiest to compute daily rainfall from hourly or sub-daily data. You may need to install CDO if you don't have it already.

@Ammar, @John, @vimal. Thanks all for the suggestions to arrive at daily rainfall from hourly records. While the main question is not about how to do it. It is more on how the reporting of daily rainfall is done by IMD. Do they report the 24 hr cumulative rainfall to previous day(yesterday) or the current date(today). 

@ William rightly pointing out that as most of the was of yesterday, it has to be reported as yesterday's rainfall, however i came to know from some of colleagues that they report it as today's rainfall. I did not find any relevant references so posted as a question in RG.

Yes, IMD records 24 h cumulative rainfall. For start and end time, you may refer to the reports from IMD.

Dear Subash, 

Weather observatories in India record daily rainfall at 08:30 am.

Total rainfall for the preceding 24 hours ending at 08.30 am of the recording date of the measurement is recorded as daily rainfall. It should be taken as 24-hour rainfall recorded on dd/mm/yyyy at 08:30 AM.

For example: Rainfall received from 08: 30 AM 10/07/2016 to 08:30 AM 11/07/2016 is recorded on 11/07/2016. 

IMD never reports the 24 hour rainfall as today's or yesterday's rainfall. It is just the rainfall recorded at 08:30 AM on some random day.

Though you need not to but you can visit your nearby IMD observatory sometime to verify this.

--------------

Therefore to your doubt "The cumulative rainfall from yesterday 7-July-2016 8:30 AM to today 8-July-2016 8:30 AM if 50mm is it for yesterday or today?"

It is neither today's rainfall nor yesterday's rainfall, It is 50 mm rainfall received from 7-July-2016 8:30 AM to 8-July-2016 8:30 AM. 

I am not sure about WMO guidelines but this is how IMD weather observatories record rainfall. 

To add more, places like Cherapunji where it rains cats and dogs these measurements are made much frequently; sometime at 3-hour interval instead of 24 hour to avoid overspill from raingauges.

Edit: Reference Line 178-183

Preprint Does the GPM mission improve the systematic error component ...

Although the timing of daily reporting is unusual to me, I understand your approach now and would stick with your normal methods.  If you decide to compare with a procedure from another area, you may have to provide some detail associated with the timing differences or alter one data set or the other.  These records can become precious, so keep the data detail for posterity.  

We often record at 15 minute intervals, but for some high level study, we used 5 minute intervals when we could.  Preserving rainfall amounts and intensity detail should always be done, and may be the responsibility of the person or entity collecting the data.  

Sorry if I provided some confusion with my timing suggestion, as the 830 am recording time as todays reading was not what we use and indicated to stick with the methods common for your area.  

If you use Excel for rain data or something similar with columns and lines, it takes a few minutes to alter hourly data to daily records, or into three hour records, with just a short script.

Mr. Bisht has already answered. Thanks

Dear Subhas

Can you please send me link to download the TRMM daily rainfall for Tehri Garhwal districts of Uttarakhand.

Actually I have rainfall of my study area for last two years and I want to do some LULC study for that I am in need of past rainfall.

According to WMO standards rainfall observation is supposed to be done at 0600z (GMT) every where, case of Kenya 0600Z is 0900 local time and the rainfall between recorded between yesterday 29th July 0900 local time and today 30th July 0900 local time will be yesterdays rain i.e for 29th July. This is known as cast back or throw back and is well understood by meteorologist who use the data operationary

I do agree with John but in case of India its 08:30AM and 5:30PM.