Module 5 (Supervised)

Was a little disappointed with this. The sandy areas in the coastline jumped into the urban classification. Perhaps I should have used a sandy area for a water example? The areas that weren't classified also jumped into the urban class (specifically the large green area in the NE central area of the original image) and the residential class too. Would capturing these areas and inserting them into the unclassified class have worked? I'm guessing that more classes would streamline these problems though.

Module 5 (Unsupervised)

Discuss the usefulness and pitfalls of image rectification.

Image rectification's main usefulness is being able to use multiple images from different sources to produce one comprehensive output. The main pitfall is human intervention! As we know, automation can produce pinpoint accuracies but the data is dependent on human input to identify and locate suitable control points. Like digitizing, the human factor diminishes the reliability of the data. Of course, the image may not help as it may be difficult to identify accurate points - we all assume there will be nice road intersections and giant 'X marks the spot' runways to set as control, but what if the image is in the wilderness or contains hundreds of water features? Also, unless the image has been corrected beforehand to account for distortions such as terrain, this may account for some further error in accuracy. The future looks bright though: increasingly higher spatial resolution data is being acquired that should ease the rectification process.

Why do the following features appear as they do in this thermal infrared image?

Image taken at 6.45am with a ground level temperature of 12C (54F). Now this is positively comfortable for a snow loving Brit like me! However, the time of day is important in this image as the night has cooled the area and the image largely shows gray tones relating to how well features have retained the daytime heat; the basis of thermal inertia. The storage sheds' roofs in the back yards no doubt get hot during the day but lose all heat at night thus appearing very dark. Same goes for the automobiles except where engines are currently running or just been used. Even with the engine running, the frame of the car will still appear dark. Vegetation also shows dark being cool but may vary in grayshade depending on it's moisture content and the content in the soil. So what's bright? Well, sidewalks and patios are brighter than most features as they take heat in quickly during the day but release it more slowly than others except roads in particular in this image. They are likely covered with asphalt which is an excellent absorber and retainer of heat thus appearing in the brightest shade of gray at night. Possibly only a waterbody could match that emission but there is no obvious one in this image. Finally, roof top bright spots are most likely hot vents from each house's heating system, unless some still have their chimneys active at 6.45am!

Module 2, Q7 Compare Displays

There are at 2 major differences between these displays that I notice.

1. The panchromatic has a clearer spatial resolution. The urban area in the multispectral is noticeably more blurry. The resolution is 10x10m to the multispectral's 20x20m.

2. The obvious difference of color as opposed to grayscale emphasizes the contrast between surface features, specifically the blue areas to the southeast that I'm guessing are grassland swamps. A river flows through it that appears to carry much sediment at the mouth which shows very well in the multispectral.