One can develop a Unit Hydrograph for ungauged catchments using information on rainfall and physiographic characteristics of the catchment. The most elementary approach to do this is by using the Synder's Synthetic UH. A more advanced approach is to use the Geomorphological Instantaneous UH (GIUH).

The Synder's approach is well documented and available in standard texts of Hydrology. The Central Water Commission of India has developed a standard procedure for deriving unit hydrographs for design flood estimation in various river basins. You can download the CWC manual from the attached link:

Good Luck!

http://www.cwc.nic.in/main/webpages/MANUAL-GUIDELINES.html

Thanks Lakshman,

I have read it.

What I mean to say here is that ' considering one day rainfall and one day runoff as pair of observed entities , is it of any practical utility or significance to fabricate or derive an unit hydrograph  which could be termed or designated as observed unit hydrograph  or say  Real representative unit hydrograph derived by using observed data of rainfall and runoff , but at time step of one day i.e. 24 hours.  I am not for deriving a synthetic kind of unit hydrograph like that of snyder , GIUH or others.

changing unit duration from 24 hour to one hour ??

In a situation where we have hourly runoff data for a catchment but we do not have hourly rainfall data. the available rainfall data is only of daily type.

How you react ..it will be of good use atleast for me 

Your questions seem to be on  the rainfall-runoff relationship at varying time scales. For ungauged watersheds, one challenge is  the determination of the runoff coefficient for your watershed. As this depends on soil drainage and land cover, one approach is to start with standard values of the coefficients and iteratively find appropriate aggregate runoff values. You could use CN method to establish the rainfall-runoff relationship under a particular soil moisture state. downscaling from 24 hours to an hour could be done with the CN model by inputting rainfall to hourly scales. 

Yes Randhir

thanks

I agree for down scaling with CN model,, but once again my basic question remain ..

converting daily time series data i mean daily rainfall data ( a single digit) into probable hourly time stepped hyetograph for that particular catchment ...how to do it..can it be done ? i am eagre to know ....interest or goal is not to compute daily rainfall and runoff depths.. but intension is to have a representative unit hydrograph of one hour unit duration using observed 24 hour rainfall data..without going for empitrical or synthetic approaches... as I have hourly runoff data only problem is that the rainfall data available is of daily ..not hourly ..and me eagre to derive an hourly hyetograph by any means... is it possble n feasible ...that i need ur important view

Gaur,

OK. I understand your problem now. You have measured runoff at hourly time step, but only 24-hr rainfall depth measurements. You want to derive a 1 hr UH. 

Yes, it is possible to do so using the CWC procedure. Check out Part III in the attached file on how this can be done for catchments in the Betwa Sub-Zone 1(c). You will have to refer to the appropriate CWC document for other sub-zones in India. 

Hope this is useful.

http://www.cwc.gov.in/main/downloads/Betwa%20subzone%201(c).Pdf

Thanks a lot  Lakshman

Its just excellent help from you. Thanks again

Only one update i need , as it looks to be location specific, where the conversion coefficients etc are synthesized for that region. For other regions like of where I need to apply , will it be OK to go ahead with same configurations of equations of conversions ..24 hours to one hour rain...

Gaur,

No. You will have to use the relevant coefficients for your region. You may get the manual relevant to your region either from CWC or by searching the internet.

Glad you found my answer useful.

Good Luck.

The above conversation is of some concern. Generally speaking, the runoff-response relationship for a particular catchment is non-linear and non-stationary, and the implied mathematical problem falls into a class called "ill-posed" problems. It is true that over a range of catchment "wetness", the concept of a linear response function (the IUH in this field) has often proved useful, and sometimes the derived time-constants and coefficients (perhaps describing the major pathways for water flow - e.g. fast-, intermediate- and slow- pathways) have more-or-less regional validitity, but in general this is not the case, as it is impossible to know a priori if these unobserved pathways are hydrologically similar between catchments.

From memory, the Australian Tom McMahon showed that under certain circumstances it is possible to infer rainfall intensity as a function of time from hydrographs of sufficient length, but I do not believe that this is generally of practical use. To try to use this approach to quantitatively interpret real results in terms of an approximate model to solve an inverse problem like this is optimistic at best; however, with sufficient experience with "similar" catchments and climates where hourly data is aggregated to daily, it may be possible to infer the dominant sub-daily time constant etc. from the data you have. However, such a model is of little use without also determining the uncertainties in its predictions, which is a more difficult (maybe not impossible) job.

Chris Barnes

I am gratified by your response. It is altogether a different and innovative approach i guess. Wonderful idea to think of inverse or reverse research. i do not have words to appreciate this approach narrated by you. Yes we can see, lets evaluate the patterns of  hydro graphs and draw inference..may be some thing good can be churned out.  I totally agree that trying any correlation blindly without looking into real physics and logic is just a black box magic ..put some thing in and get some thing out  and if out put is ok  its your model model and model very excellent model..if output is inferior then ..data data data..data is poor data is poor .....but what is real logic or physics inside the four black colored dark walls ..no one knows..and interestingly no one like to know as they are in hurry to get some thing good as output ....also they afraid to report negative or bad outputs....this is what making us handicapped..

Lets invite views of others if they feel it of any worth...for we hydrologists and watershed managers in particular..

thanks n regards

gaur

Mr. Gaur,

I hope the following suggestions can solve your problem.

Your problem is that you have hourly runoff data and 24 hr rainfall data. How to simulate the runoff hydro-graph?

This can be solved in two ways. First by using a non-linear rainfall-runoff model and other by way of convolution of hourly rainfall over a unit hydro-graph.  In both the cases you need the hourly rainfall data. Hourly rainfall can be obtained by disaggregation of 24 hr rainfall into time step of one hour. In my opinion, now-a-days every region has got real time or SRRG stations that record the hourly rainfall. By using the those information the 24 hour rainfall can be converted to hourly rainfall following the procedure given in " Kottegoda, N. T., Natale, L., Raiteri, E. (2014). Monte Carlo Simulation of rainfall hyetographs for analysis and design. J. Hydrol., 519, 1-11." Then by using the most simplest non-linear rainfall-runoff model suggested by Bhunya et al. 2015 (Paper attached) the entire direct hydrograph can be generated. But care must be taken while filtering the flood event and the corresponding rainfall event, since it is very much subjective regarding the beginning and end of both the processes.

In my view, we can't  follow adopt Snyder UH or CWC UH as these are region specific and CWC UH is prepared by using data of few bridge locations of a region. The limitation of CWC UH has been highlighted in 'Flood Hydrograph with Synthetic Unit Hydrograph Routing. Water Resour Manage (2015) 29:5765–5782

DOI 10.1007/s11269-015-1145-1'. Moreover you can determine the average UH by using few selective flood events of your own region.

Second, if the purpose is to determine the design flood at an ungauged location, 24 hr rainfall can be statistically disaggregated using a suitable probability distribution function. Please note that the pdf has to be double bound only otherwise the volume will spill beyond the coarser time step i.e. 24 hr in your case. I have made a similar attempt and have become successful and my work is at present unpublished. 

Even the concept of runoff coefficient is very much misleading. For that please go through' Viglione, M., A. R. , Blöschl , G. (2009). On the role of the runoff coefficient in the mapping of rainfall to flood return periods.  Hydrol. Earth Syst. Syst. Sci. 13, 577-593.'

Now your comment please.

All the best.

Thanks a lot Mahendra

wonderful and most relevant information you gave

Really it very nice work you are going along. all the best.

I have read ur good paper. Let me get me much educated on this aspect in particular the statistical part. Only my curiosity is to see ,

1.  One day rainfall ( a single value in mm) needs to be converted into hourly discrete values that means we need to change the shape of hyetograph. In one day form it is almost like a rectangle , which could be seen in a shape like bell or atleast a triangle , if we make it hourly. How simply it could be done is still a food for thought as per me 

2. is there any simple way like intereactive kind of spread sheet (Excel) where such conversion could be atleast attempted first in a hypotheical input data and later real by blending or giving connectivity (hyperlinking) to very nice approaches, statistical functions, probability density functions or n number of approaches what our excellent friends (who answered or reacted on my tiny question on this wall) have suggested.

3. The work which is unpublished at your end, seems to be extremly important and unique. my best wishes . good luck

in last  happy deewali....n new year greetings ...

thanks again for ur nice help

rgds

Gaur