Because of its importance as a solvent and its many unique properties, water is a widely studied substance. Computational modeling can bring insight to detailed mechanisms of the interactions of water with itself or other chemicals. The accuracy of simulations of aqueous systems are limited by the accuracy of the water model used. A wide range of empirical models which accurately reproduce bulk water have been developed, but are limited to nonreactive systems. Ab initio models can accurately reproduce water properties, including reactions, but are limited to small systems. Semiempirical models can access larger systems and include reactions, but the transferability of the models is limited by the parameterization. The self-consistent charge density functional tight binding (SCC-DFTB) model was developed in the 1990s to have greater transferability and reliance on fewer parameters. The DFTB+ implementation of SCC-DFTB allows for second- and third-order expansions of the density fluctuations in the energy. For each of these models the structural, dynamic, and spectroscopic properties of bulk SCC-DFTB water are compared to an empirical model, SPC/E, and experimental properties. Although all of the SCC-DFTB models exhibit some failures characteristic of the parent density functional theory, the third-order models are the best models. The experimental-density third-order model best reproduces the liquid structure and rotational dynamics, while the ambient-density third-order model best reproduces the diffusion and infrared line shape.