In field studies Lysimeter and Pan Evaporimeter can be used to quantify the percentage of losses.
You can also try the available methods for quantifying the evaporation using Penaman-Montieth, Hargreaves etc. to quantify the evaporation provided the adequate meteorology data is available.
Adham: Amount of rainfall will transform into Evapo-transpiration depends on so many factors (e.g. air temperature, types of plants, relative humidity, vapor pressure, soil moisture, geographical pattern, etc.). Similarly, amount of rainfall will transform into infiltration depends on many factors such as soil type, land use/ land cover, ground slope, etc.).
Therefore, Percental Prediction of Evapo-transpiration and Infiltration based on Rainfall Quantity or total water storage will certainly not give you accurate prediction for a particular watershed. If you have Field Measurement Data for Evapo-transpiration and Infiltration of a very similar characteristics Watershed (e.g. size, shape, slope, climatic condition, etc.) then only you may use Percental Prediction with Caution.
Comes down to a question of what data you have. If you have daily (or better) record of water level in the rainfall harvesting structures, then you can use this to look at the overall losses in dry conditions (i.e. where it hasn't rained for at least several days). This will give you an estimate of the overall losses (ET + infiltration + extraction). With information on 2 of these, you can then estimate the remaining term.
As Deepak says, the best method would be direct measurement of PE and infiltration. However, if you need ET rather than PE, you will need to convert to ET, which will need information on the vegetation in the area.
If you don't have an evaporation pan, then there are a lot of methods that have been developed to estimate PE, each with their limitations. Typically, you will need some additional climate data - the more data you have the better. The best method is Penman's equation, which needs temperature, relative humidity, solar radiation and wind data.
Infiltration can be difficult to estimate, so if you have information on the water use, that might be the best method to estimate the infiltration rate.
For that you can use Water Balance equation for small-scale studies. And it would be ,some how, an estimation NOT calculation, since many factors effect on it.