It sounds like a water budget using daily or weekly timesteps should be prepared. To estimate inputs, you will need to estimate surface runoff. Why do you think TR-20 is not a good tool? What is the land use in the tributary basin?
It also sounds like you are essentially looking at rainwater harvesting to meet water demand. My experience is that if you are trying to provide a reliable water supply by collecting storm runoff, then your design should NOT be based on average precip values, unless you apply a generous safety-factor. If you can analyze/model/simulate multiple years of real rainfall data you will have a much more reliable design. Remember, about half the time rainfall will be below the "average", so you might run out of water if you don't account for drier-than-avg conditions climatic conditions.
If it was me, I would do a long-term simulation with a continous hydrologic model, if possible. This approach let's you draw conclusions such as, "the pond would have been dry twice in the last 30 years". If you can not look at the system performance based on measured rainfalls, then base the design on a drier-than-average year.