Some classical hydrological questions may be considered as single points in a common 3D "knowledge space" where first axe X1 would represent information content on forcing meteorological variables (e.g. Rainfall), second axe X2 information content on prognostics hydrological variables (e.g. discharges) and third axe X3 information content on hydrological processes (e.g. model parameters values). Indeed model calibration problem use information on rainfall and discharges values and try to identify model parameters: for a given amount of rainfall and discharges data, model calibration may then be seen as a point in the (X1,X2) plan. Model simulation problem use, for example, a given amount of rainfall data and the parameters values of a given model and try to estimate discharges: it is then a point in the (X1,X3) plan. It is suggested in this communication that some others hydrological questions may also be seen as other points in the same 3D knowledge space. For example, knowing the only rainfall events volume and the discharges time-series is it possible to identify both rainfall intensities time-series and model parameters values? Knowing rainfall intensities and the only river heights time series, is it possible to both identify model parameters values and the rating curve? Placing these resolved questions in this common knowledge space helps to identify step by step a minimal information content surface. Problems above this surface contains redundant information: can they help to reduce uncertainty? Problems below this surface are bad-constrained problems: which supplementary data would be help in reducing identifiability problems?