In the United States, the Army takes care of land management for installations consisting of more than 4.8 million ha of land for Military training programs. Various long-term training activities continuously disturb ground cover, damage plants, increase potential of soil erosion, degrade habitats, and fragment landscapes and thus lead to natural resources and land condition degradation. In turn, the degradation of land conditions will limit military land carrying capacity. For both sustainable land condition and carrying capacity, there is a strong need to assess land condition and identify land for repair and restoration. In this study, we used percent ground cover of land to quantify the cumulative off-road traffic impacts at the Installation of Fort Riley and generated a time series of maps that accurately simulate spatial patterns and temporal dynamics of percent ground cover, and uncertainties of estimates from 1989 to 2001 using an image-aided co-simulation algorithm. Then, we defined and applied a set of loss assessment function for allocation of land repair and restoration based on the rule of minimum loss. The loss functions transferred uncertainties of land condition estimates into uncertainties of loss assessment and thus provided the basis for minimum loss in decision-making. Finally, we compared this method with simple threshold value and probability based method for repair and restoration of land.