The first observation of ephemeral hydrocarbon lakes on the surface of Titan was made on May 27, 2009. SAR imagery of the south polar region (63°-58°S, 144°E-153°E) was obtained in 12/2007 (T39) and 05/2009 (T55). A collection of features, whose morphologies match that of previously identified partially-filled lakes, show more than an order of magnitude increase in backscatter cross-section in T55, where the boundaries are no longer visible. This change is inconsistent with common scattering models at the relevant incidence angles of 14° and 37°. If not due to observational effects, the disappearance of these dark features in the intervening 1.5 years represents surface change. Backscatter models that include a diffuse scattering component in addition to a quasi-specular one are required to explain the incidence angle dependence of empty lakes present in both T39 and T55. These scattering models are consistent with both individual empty lakes observed at multiple incidence angles and the collective set of empty lakes observed to date. However, granular lake features observed in T39 with low backscatter forbid a significant diffusive component to explain their returns. Thus, the increase in radar brightness within the study region suggests the exposure of a diffusively scattering lakebed that was previously hidden by an attenuating liquid medium. Among the potential explanations for the surface change, we consider liquid evaporation and infiltration. If infiltration is responsible, a dynamic hydrologic system with a regionally varying phreatic surface is implied. Methane evaporation is a likely explanation, consistent with the current season on Titan. Current GCM estimates suggest that 1 m of methane can be expected to have evaporated between the observations, an amount sufficient to attenuate the diffuse backscatter according to a simple two-layer model. These observations of lake changes constrain volatile fluxes and hence, the evolution of Titan's hydrologic system.< Réduire