Land cover (LC) refers to the physical state of the Earth's surface such as soil, vegetation and water, etc. However, most LC features occur at spatial scales much finer than the resolution of the primary remote sensing satellites. In this paper, we explore the possibility of collaborative sparse unmixing for estimation and quantification of LC classes at subpixel level to obtain abundance maps in both the unconstrained and constrained forms (with abundance nonnegativity and abundance sum-to-one constraints imposed). Firstly, computer simulated noise-free and noisy data (Gaussian noise of different noise variance: 2, 4, 8, 16, 32, 64, 128 and 256) were unmixed with a set of global endmembers (substrate, vegetation and dark objects) in the NASA Earth Exchange. In the second set of experiments, a spectrally diverse collection of 11 cloud free scenes of Landsat-5 TM data representing an agricultural set-up in Fresno, California, USA were unmixed and validated using ground vegetation cover. Finally, Landsat-5 TM data for an area of San Francisco (an urbanized landscape), California, USA were used to assess the algorithms and compared with the fractional estimates of World View-2 data (2 m spatial resolution) for validation. The results were evaluated by using descriptive statistics, correlation coefficient, RMSE, probability of success and bivariate distribution function. With computer-simulated data, both unconstrained and constrained solutions gave excellent results up to a certain noise variance, beyond which the performance in classification gradually decreased. For the agricultural setup, mean absolute error (MAE) of vegetation fraction between actual and estimated abundance values was 0.08 for both unconstrained and constrained case, and for the urban landscape, average MAE of the three classes considered was 0.73 for unconstrained and 0.07 for the constrained solution.