The shallow water acoustic channel is challenging to estimate and track due to the ill-conditioned nature of the problem, the need to optimize over a complex field, and the time-varying nature of the channel coefficients. We have previously presented a geometric mixed norm approach to estimate and track the channel delays and delay-Doppler spread function coefficients that exploit the often sparse distribution of the channel coefficients. In this work, we present the effectiveness of our approach over a range of field data collected at 15 m depth over ranges of 60, 200, and 1000 m at various wind conditions. We also compare the performance of the estimation and tracking algorithm against other sparse sensing approaches to shallow water acoustic channel estimation, e.g., the L1-regularized least squares algorithm. We show that the choice of the sparsity factor λ, which is an essential design parameter in all sparse sensing methods, plays a critical role in the effectiveness of sparse reconstruction of real field data, and present novel ways to track the sparsity factor over time in a way that is consistent with the time-varying nature of the shallow water acoustic channel.