Dear Tanveer,

Flood estimation based on frequency analysis is of two types i.e rainfall  or runoff based. If it is to be calculated from analysis of rainfall then we require information from a number if raingauge stations those are located inside the catchment and very few outside and the number depends on its size and type whether it is in planes or hilly etc. A single raingauge that to located outside  the catchment does not qualify it using its information for flood estimation. It is not only the distance but also other factors like location, altitude, leeward side or windward side and  geomorphological homogeneity etc will govern its candidature for flood frequency analysis. 

Further while using the rainfall for analysis of flood at a gauging station, it is generally assumed that the T-year quantile of rainfall corresponds to T-year runoff which is not true. The antecedent soil moisture, rate of evaporation and other climatic factors control the catchment response. It is already established that a 10-year rainfall generates upto 100-year flood. So in my opinion a single station information is not so helpful in arriving at hydrological decisions and hence I discourage. 

But yes, if it is for planning purpose that to in its pre-feasibility stage then the information may be used.

I have never tried to estimate flood peaks from just rainfall.  If you have any stream and associated rainfall gauging stations in your physiographic and climactic area of interest, it would be good to review their data and evaluate relative to your catchment of interest.  If you are able to read bankfull flow indicators for your stream (often suggested where stream expands onto floodplain or near top of point bars), Dr. Dave Rosgen has suggested that if you project 2 times bankfull depth at thalweg across the channel, that may give you an approximation of flood stage for 50 year event.  There may also be flood level indicators of damage.  With the cost of transducers to measure water levels somewhat reasonable, you might even want to consider obtaining some water level and collect some stream flow data through flume or measuring flow to have a  better idea of how your catchment is responding.  I will be also interested if someone responds who knows how to estimate flood peaks from just rainfall.  There are some models such as the SCS CN method that apply the catchment soil factors of response that may help, but these should be used with care unless validated to your area.  I think tr55 publication may be helpful, especially if you can validate this with some existing gauged stream data..  

#Mahendra Kumar & William F. Hansen Thank you for your answers. But right now i have only one rain gauging station 25 km away. What would you be suggesting?

Максимальные расходы дождевых паводков малых рек (с площадью водосбора менее 200 км2) определяются по формуле предельной интенсивности (тип III)

                                             QP% = q'1% φH1%δ λрF,                                            (4.6)

где  q'1% – максимальный модуль стока обеспеченностью 1 %, выраженный в долях от произведения  φH1%;

φ – сборный коэффициент стока;

H1% – максимальные суточные осадки обеспеченностью 1 %, (по Илирнею –  38 мм);

δ – коэффициент, учитывающий влияние водохранилищ, прудов(здесь δ = 1);

λр – переходной коэффициент от обеспеченности 1 % к другой обеспеченности;

F –  площадь водосбора, км2.

Для выполнения работы были выбраны гидрологические посты с площадью водосбора менее 200 км2. Из-за низких коэффициентов корреляции пропуски в наблюдениях за максимальными расходами дождевых паводков восстановить удалось только для одного поста 01513. По рядам наблюдений за стоком были рассчитаны максимальные расходы дождевых паводков обеспеченностью 1 и 3 %, а также переходные коэффициенты от расхода обеспеченностью 1 % к расходу 3 % (таблица 4.23).

Далее в соответствии были рассчитаны значения сборного коэффициента стока для условного водосбора φ0 (таблица 4.24).

Схема расчета φ0:

- по изученным рекам определяем модуль  q1% = Q1% F,  м3/с·км2;

- считаем гидроморфометрическую характеристику русла Фр при этом в первом приближении принимаем φ =0,80

                               Фр = (1000L)/[mpI A0,25(φH1%)0,25],                               (4.7)

где L – длина реки, км;

mp и m – параметры определяемые по [11];

Ip – средневзвешенный уклон реки, ‰.

- склоновое время  добегания τск=60 мин;

- относительный модуль максимального расхода q'1%=q1%/(φH1%) определяем по [11] (аргументы Фр и τск);

- считаем  новое значение сборного коэффициента стока

φ = q1%/( q'1%H1%);

- снова считаем гидроморфометрическую характеристику русла Фр при этом в принимаем новое значение φ (см. табл. 14 прилож. 2 в [11])

- уточняем относительный модуль максимального расхода q'1%=q1%/(φH1%) по табл. 9 прилож. 2 [23] (аргументы Фр и τск);

- считаем φ0

                                       φ0 = φ(F + 1)n3/(c2(Iск/50)n2),                                         (4.8)

где с2=1,2; n3=0,07; n2=0,65 – по табл. 11 прилож. 2 в [11].

Из полученных значений φ0 посчитано среднее для района (см. таблицу 4.24).

While it is recommended to have rainfall estimate from stations within a watershed, using data from outside the watershed is not unusual.distance of 25 km is too far to reasonably predict flood peaks in your watershed. I recommend looking for more stations in all directions of the watershed and perform an interpolation of rainfall stations using Hyetograph method. This way you could get a reasonable approximation of rainfall pattern in your watershed. I also recommend you to check such interpolation estimates with addition rainfall measurement within the watershed using some rain gage installations. 

When you say that you are trying to estimate flood flows, I assume you mean the exceedence probability and not actually the flow for a particular flood? The answers above focus on estimating flood peaks for particular events, and I agree with the statements there. If you are after the exceedence curve, then you might be able to make an estimate of this based on a single rainfall gauge. To do this, you will need to take into consideration:

• the long-term spatial distribution of rainfall (e.g. annual) - maybe using satellite data to convert the point rainfall data into an estimate of the areal rainfall data. You will also need to consider the spatial extent of the rainfall producing a flood event, which may lead to a decrease in the estimated areal rainfall exceedence curve (depending on the size of your catchment, and the type of event likely to produce a flood: tropical cyclone, frontal system, convective storm).
• One you have an estimate of the areal rainfall exceedence curve, you will need to consider the impulse response of the catchment, which includes the unit hydrograph as well as the conversion of rainfall to effective rainfall.
  • For extreme flood events, you could take the areal rainfall as an upper limit on effective rainfall. The result would be an upper limit on the corresponding flood peak (remember this is not for a particular event, but for a particular probability of occurrence). How well this captures what the actual effective rainfall depth would be will depend on the characteristics of your catchment and the climatic conditions there (more likely to be valid in wet areas, very problematic in drier areas - there you will need to have some means of estimating the likely runoff coefficient).
  • You will also need to estimate the form of the unit hydrograph in order to convert the effective rainfall depth into a peak discharge. There are ways of doing this, from using gauged data through to using a kinematic wave model. All depends on what other information you have available.

This is all based on a lot of assumptions, and you will need to consider the impact of those assumptions on the confidence in the result - i.e. estimate the uncertainty in the predictions and see whether this is small enough to make the result useful. To discuss further we would need a lot more information about your catchment: area, topography, climate, types of weather systems, ...

You can use a rainfall station 25 km away if there is no reason to suspect great variation of local rainfall in this area (e.g., no mountains, etc.). In practice, similar assumption are routinely made by hydrologists.