This doctoral thesis provides the first comprehensive study on the physical processes controlling hydrodynamics and transport in Lake Garda. The investigation is carried out in parallel on three different levels: data collection and analysis, three-dimensional numerical modeling and theoretical study. On the first level, data are collected by building up a network of research institutes and local administrations in the lake area. New data are acquired through traditional field campaigns (CTD, thermistor chains, satellite imagery), while a citizen-science approach, based on local knowledge harvesting, is successfully tested to gather qualitative data on surface circulation. On the second level, a three-dimensional modeling chain is set up, by coupling one-way a mesoscale atmospheric model to a hydrodynamic model. Both models are validated on multiple temporal and spatial scales, allowing to identify the main interactions between the weather forcing and the hydrodynamic response of the lake. Circulations in Lake Garda are found to be very sensitive to the thermal stratification, to the spatial distribution of the wind forcing and to the Earth’s rotation. Surface cyclonic gyre patterns develop in the lake as a residual outcome of alternating wind forcing of local breezes and differential acceleration induced by Earth’s rotation, whereas unidirectional currents flow under a nearly uniform and constant wind. Both model and observations evidences show that, under weak thermal stratification, Ekman transport activates a secondary circulations in the northern part of the lake, driving surface water to the deep layers and possibly preconditioning the lake for subsequent buoyancy-driven deep mixing events. On the third level, the relevance of the Coriolis term in the equations of motion for relatively narrow closed basins is analytically addressed. The classical Ekman problem is solved by including the presence of lateral boundaries and a new analytical solution is formulated. The validity of the new solution is proved by numerical tests of idealized domains of different size, geographical location and turbulent regime, and on Lake Garda as a real test case. The meaningful length scales are discussed, and the significance of Rossby radious as a reference horizontal scale is disproved for steady-state circulations driven by wind and planetary rotation.