What you are trying to do is ...

What you are trying to do is to quantify the water balance in your field area. The parameter of interest is the percentage of applied irrigation water consumed by the crop. Assuming no rainfall, you have 3 elements to the equation: Applied water = Evapotranspiration + Runoff + Infiltration. Your available monitoring approaches are to quantify at least 2 of the unknowns.  As Feki Mouna has commented, evapotranspiration can be measured directly using Eddy Covariance (basically sophisticated weather stations) or can be estimated from remote sensing data. Field runoff to drains can be measured directly with weirs. Infiltration can be measured by examining groundwater levels or careful vertical characterisation of soil moisture profiles. In practice projects will often use field measures for runoff/infiltration data, as these are fairly easy and cheap to measure, and calculate evapotranspiration, or use remote sensing products for evapotranspiration and estimate infiltration.  The direct measurement of evapotranspiration is complex and expensive and usually only done by academic projects to validate the remote sensing products.

An alternative approach is to model based on meteorological measurements, crop types and irrigation technology - but of course how well models match the real world depends on both the model and the quality of its inputs.

I think its valuable to consider issues of irrigation efficiency holistically. For an individual field high irrigation efficiency may mean more 'crop per drop' but water that runs off one field or infiltrates may support ecosystems or be the water resource for other farmers., so at a command area scale pursuing efficiency can be subject to a law of diminishing returns.