Convolutional neural networks (CNNs) are widely acknowledged in the fields of image and video recognition, face recognition, image analysis, image classification and activity detection. CNNs take images as their input; assign adaptive weights and biases to numerous features of the image; and then assign the various categories to them. The intent of this paper is to establish a model to classify outdoor images to different weather classes. Literature survey about the field related to weather prediction has shown that the best results are obtained while using the CNN models. This paper proposes a method of implementation of convolutional neural networks to classify separate weather conditions into four classes, namely cloudy, rainy, shine and sunrise. In this paper, four CNN models with different number of model layers are implemented and their results are examined.