Vegetation

NDVI

In this study, NDVI analysis was used to determine loss of live green vegetation over time in and around the Agbogbloshie dump site. Although the changing values of vegetation does not directly mean an increase in E-waste and could have been caused by a number of factors (including clearing the land for the dump-site itself), for the purposes of our study we are assuming that a decrease in green over time correlates with the increase in E-waste from the 1990’s onward, and thus is largely caused by the E-waste itself. The toxic chemicals and heavy metals released into the soil and water sources from the improper disposal of electronic products (polyurethane burning, breaking apart, etc) as well as the build up of the waste itself likely disrupts and prevents the growth of vegetation. This is directly reflected in the NDVI imagery. 

We examined NDVI’s of images taken from Landsat for April 1985, January 1991, February 2003, and January 2011. These images were chosen both because they were the years and months with the clearest and most accessible imagery and because they provide us with a broad idea of how the E-waste developed over time--from before the E-waste dumping, to when it first began and its increase in severity until 2011. The red polygon represents the region of interest (ROI) for which we examined the vegetation levels. We chose to look at an area larger than just the dump site because the most vegetation occurred on the outskirts of the site as well as around Korle Lagoon and the area to south west of the site. We can assume that the accumulation of toxic material and heavy metals in the soil and water reach further than just the dump-site, thus looking at the effects of vegetation on the surrounding areas will reflect the level of E-waste in the site itself. 

The images below are the NDVI (bands 3 & 4) time series, the first from 1985 and last from 2011. The lighter pixels have a higher green reflectance meaning these areas have higher levels of live green vegetation.The clearest images come from 1985 and 2003 between which we can see a significant decrease in green. The site itself as well as the area to the south west both have a considerable amount more light pixels in 1985, and we see that these areas get much darker by 2003. Furthermore, 2011 shows very little green reflectance as the majority of the pixels are relatively dark gray. Korle Lagoon is harder to analyze by simply looking at the color. 

1985 NDVI

Statistics from 1985 NDVI

1991 NDVI

Statistics from 1991 NDVI

 2002 NDVI

Statistics from 2003 NDVI

 2011 NDVI

Statistics from 2011 NDVI

The statistics reveal this change as well: the mean NDVI value for 1985= 0.180, 1991= 0.042, 2003= -0.054, 2011= -0.305. The over all change rate from 1985 to 2011 is a decrease of 0.485.

Values were also taken from various sample pixels in strategic areas: 

• the bottom of Korle Lagoon shows an overall decrease of 0.739

• the Top of Korle Lagoon shows an overall decrease of 0.198

• the SE edge of the dumpsite shows an overall decrease of 0.374

• the area to the SW of the dumpsite show and overall decrease of 0.454

Vegetation Loss in the Visible

Below are Landsat images taken form April 1985 and February 2003. They display the visible bands (1, 2, 3). Although the image from 2003 is much brighter (due to variances in image quality, potential cloud/chemical cover, and the stretch that we applied when manipulating the image) just from the visible, we can see a significant decrease in green in the same areas shown by the NDVI--around the edges of the site itself and the area to the South West of the site. We can also see a decrease in what looks like tree cover around the river leading down to Korle Lagoon and around the top of the Lagoon.

1985 Visible

2003 Visible

 


Change Detection - Green Band
The image below was taken from GoogleEarth and displays GeoEye imagery from Feburary of 2000. At this point, there was little to no ewaste dumped on the site. As one can see, there are healthy-looking mangroves around the river along the eastern border of the dump site.



However, the image below, which is an image from December of 2012 using Digital Globe imagery from GoogleEarth, paints a different picture. At this point, the dump site is heavily utilized and polluted. The mangroves are gone and the surface area looks damaged.

We wanted to do a remote sensing analysis to determine whether there was an overall decrease in vegetation in the area due to groundwater pollution as opposed to there simply having been a removal of the mangroves unrelated to chemical seepage from the dumpsite. We did a change analysis of the green band (band 2) between these two images to try to examine the change in greenness from February 2000 to December 2012.

In the change detection map below, the dark blue areas represent a negative change (e.g. a decrease in greenness) while the red areas represent a positive change (e.g. an increase in greenness). There has been a marked decrease in greenness within the dumpsite itself, which is almost entirely blue.