DMSP-OLS Project Description

In this project I processed DMSP-OLS satellite imagery that covered a 22 year period and will be used in research for measuring economic market activities of a remote region in Brazil. My contribution only consisted of extracting spectral values from the raster images, where the mean and median of these values were used in the temporal study, as well as area growth from 1992-2013.

 

Software: ERDAS Imagine, ArcMap and Excel

The first step involved preparing some base files for this project. Since I'm not familiar with this area, I created a shapefile of points identifying the cities that were of interest to the researchers.  I also labeled them with text so that they could be easily recorded in the attribute table. This was necessary due to  the amount of data that needed to be processed. The next base file was creating an AOI in ERDAS from a shapefile of the area. This further limited the area that needed to be processed and saved a lot of time.

 

Processes

ERDAS:

1. Perform an unsupervised classification of the .tif satelllite image using 10 classes.

2. Recode the classification limiting the process to the AOI.

ArcMap:

1. Used the Raster to Polygon tool to convert the reclassified raster to a polygon.

2. Selected the city areas in the new polygon with the identity tool while holding down Ctrl key. Exported these areas as a new shapefile.

3. Use Zonal Statistics as Table tool to create statistics from the polygon that contained only the city areas in the study.

4.  Open ZonalStats and added a column: CITIES. Edited table identifying shape areas with proper city name.

 

Best practices:

The reason I didn't do the classification in ArcMap is because I found it to be less robust in classifying the data than ERDAS. I ended up with different results in light averages. I decided to use ERDAS even though it took longer. Also, with ERDAS I made the decision to clip the area after performing a classification. Although it added to the processing time, clipping the area before processing yielded a smaller class group or different histogram values in some instances. This was believed to be because the area under study had a lower average of light values in general compared to the rest of the raster image.

 

Deliverables:

Shapefiles of the cities of interest for the study area, an Excel Workbook with statistical data and sheets for each year, a geodatabase for each year containing files for that year, a geodatabase of the zonal statistics only.

 

(shapefiles over a raster image)

 

                                1992                                                                                   2013

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Statistical data exported in CSV format to an Excel sheet.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Copyright © 2018 pixelboro.com