Tim McGraw, Purdue University, West Lafayette, USA, email@example.com PRIMARY
Aijun Huang, Purdue University, West Lafayette, USA, firstname.lastname@example.org
Sijin Wang, Purdue University, West
Lafayette, USA, email@example.com
Approximately how many hours were spent working on this submission in total?
May we post your submission in the Visual Analytics Benchmark Repository after VAST Challenge 2017 is complete?
1 – Boonsong Lake resides within the preserve and has a length of about 3000 feet (see the Boonsong Lake image file). The image of Boonsong Lake is oriented north-south and is an RGB image (not six channels as in the supplied satellite data). Using the Boonsong Lake image as your guide, analyze and report on the scale and orientation of the supplied six-channel satellite images. How much area is covered by a pixel in these images? Please limit your answer to 3 images and 500 words.
In the RGB reference image the 3000 foot long lake is 232 pixels long, resulting in a scale of 12.93 linear feet per pixel, or 167.2 square feet per pixel. Our application allows the user to rotate, translate and scale any of the satellite images relative to the RGB reference image (see Image 1). Both images are shown overlaid on each other with transparency. By matching visual features such as the lake boundary and the roads to the southeast the user can align the images in a few seconds (see Image 2). We found that the satellite and reference images are angularly aligned (0 degrees relative rotation), and that there is a linear scale factor of about 11.4 between the reference image of the lake and the satellite images. This results in a single pixel in the satellite image covering 21783 square feet.
Image 1 – Our manual image registration application showing the Boonsong Lake reference image.
Image 2 – A satellite image aligned with the reference image when the scale factor is 11.4.
2 – Identify features you can discern in the Preserve area as captured in the imagery. Focus on image features that you are reasonably confident that you can identify (e.g., a town full of houses may be identified with a high confidence level). Please limit your answer to 6 images and 500 words.
Note pixel coordinates are given with respect to an origin point in the upper-left corner of the satellite images.
Lakes: We have labeled the true color image from June 26th 2016 with the locations of 3 large lakes centered near pixel coordinates (10, 562), (119, 376) and (222, 330). They are annotated as 1, 2 and 3 on Image 3. Several smaller lakes can be seen in the image, including Boonsong Lake near (490, 156), annotated as 4.
Image 3 – True color image on June 26th, 2016 with annotated features.
Snow/Ice: During the winter months the lakes are usually covered with ice. Patches of snow can be seen covering some vegetation. In true color (see Image 4) it is easy to see three main lakes were covered by ice, and patches of snow are scattered over the whole image.
Image 4 – True color image on December 20th, 2014 showing ice and snow in the preserve.
Roads: A large road runs roughly north-south through the middle of all images, as pointed to in Image 3. Another major road runs east-west near the top of the images. Smaller roads can be seen throughout the preserve.
Clouds: Small clouds are visible throughout several images. They are white in the true color image, and pointed to by an arrow in Image 3. Images dated November 28th, 2014 and November 15th, 2015, shown below in Images 5 and 6, are dominated by clouds.
Image 5 – True color image on November 28th, 2014 showing heavy cloud cover.
Image 6 – True color image on November 15th, 2015 on an overcast day.
Vegetation: In Summer and Fall months it can be seen from the true color images that much vegetation is thriving throughout the preserve, as evidenced by the green terrain in Image 3.
Town: Near coordinates (50,233) there is a town at the intersection of the 2 major roads, shown in Image 7 below.
Image 7 – True color image of the town near the intersection of two roads.
Farmland: Viewed as a combination of bands 4, 3 and 2, light blue represents exposed land and red represents vegetation. In Image 8 we show images from June and September of 2015 and 2016. The circled area, near coordinates (34, 107) at the top left corner of the image is light blue in June. About three months later, in September, the area turns to a red color. Therefore, we suspect this area could be farmland.
Image 8 – Band 4, 3, 2 false color images in 2015 and 2016. The circle area may be farmland.
3 – There are most likely many features in the images that you cannot identify without additional information about the geography, human activity, and so on. Mitch is interested in changes that are occurring that may provide him with clues to the problems with the Pipit bird. Identify features that change over time in these images, using all channels of the images. Changes may be obvious or subtle, but try not to be distracted by easily explained phenomena like cloud cover. Please limit your answer to 6 images and 750 words.
We used the combination of bands 5, 4 and 2 to identify burned land. According to external information regarding false-color imagery, newly burned areas appear red in this combination of wavelengths. As shown in Image 9 below, the circled area in the image from June 24th, 2015 is apparently a burned area. This area had mostly recovered by September of the same year as shown on the right side of Image 9.
Image 9 – Band 5, 4, 2 false color images on June 24th, 2015 and September 12th, 2015. Burned areas are shown in red and circled.
From observations of the raw image of band 5, which can penetrate thin clouds, we found areas in June 2015 with more soil moisture than June 2016, especially the regions at the bottom of Image 10. We will attempt to explain this phenomenon based on evidence from the images. Under the combination of bands 5, 4 and 2 (see Image 11), clear water appears black and water with sediments or saturated soil appears blue. We speculate that floods would result in sediment-laden water. Therefore, if our false color image contains more blue areas, it could be an indication of earlier flooding. As we compared the two September images, we found that they are similar. We conclude that flooding may have occurred between June and September in 2015.
Image 10 – Band 5 Images on June 24th, 2015 and June 26th, 2016 showing high soil moisture in 2015.
Image 11 – Band 5, 4, 2 false color images on June and September in 2015 and 2016. Heavy sediment may have affected the color of the lakes in June.
We used the combination of bands 4, 3 and 2 to determine the health of plants in the images. Under this combination, the color red reflects the health of plants: the brighter the color is, the healthier the plants are. We compared the early Fall images from August 24th, 2014, September 12th, 2015 and September 6th, 2016, shown in Image 12. It is easy to see that the images of 2014 and 2016 are brighter than the one in 2015. We conclude that the plants in September 2015 are not as healthy as they were in August 2014 and September 2016. In addition, on June 24th, 2015, the vegetation appears pink in the false color image. Although we cannot identify the cause, there appears to be some change detrimental to the plants during that period.
Image 12 – Band 4, 3, 2 false color images in 2014, 2015 and 2016. Something seems to have affected plant health in 2015.
We observed that the color of one of the large lakes was abnormal on June 24th, 2015 in the true-color image. About one year later, on June 26th, 2016, half of the lake turned to an abnormal pink color again (see Image 11). The color returned to normal in subsequent images after three months. We hypothesize that this change in color could be the result of: 1) an algae bloom; or 2) pollution from the town west of the lake or the town to the south of the preserve. Because it appears that the bottom part of the image is actually a highland or at least has a higher elevation than the lake, the pollution carried by rivers or wind from the south could impact the lake.
Image 13 – True color image with histogram equalization processing on June 24th, 2015 and September 12th, 2015 showing some distinct changes in the appearance of the lakes.