Determining where a vehicle can and cannot safely drive is a fundamental problem that must be answered for all types of vehicle automation. This problem is more challenging in cold regions. Trafficability characteristics of snow and ice surfaces can vary greatly due to factors such as snow depth, strength, density, and friction characteristics. Current technologies do not detect the type of snow or ice surface and therefore do not adequately predict trafficability of these surfaces. In this paper, we took a first step towards developing a machine vision classifier with an exploratory analysis and classification of cold regions surface images. Specifically, we aimed to discriminate between packed snow, virgin snow, and ice surfaces using a series of classical machine learning and deep learning methods. To train the classifiers, we captured photographs of surfaces in real world environments alongside hyperspectral scans, spectral reflectance measurements, and LIDAR. In this initial analysis, only the photography was assessed. The classifiers were cross-validated with a subset of the data collected for the project. In addition to surface imagery, trafficability metrics were collected for each surface in the study. Vehicles from three different military classes (lightweight ATV, light, and medium/heavy) were tested as a modified Jeep Wrangler. Trafficability tests included draw bar and motion resistance, along with acceleration, deceleration, slalom, lane-change and circle dynamics tests where feasible. Each test surface was characterized alongside the vehicle and terrain sensing measurements and include measurements of density, temperature, and strength where applicable. Results reported here show that winter surface conditions can be classified with 70%+ accuracy using onboard photography. Future work includes incorporating additional sensor data, vehicle, and snow data into the model.