Addison's GIS Journey

This week's lab focused on unsupervised vs. supervised image classification. Image classification is the process of giving pixels in an image unique values. Unsupervised classifies without predefined training data, while supervised classifies guided by known training samples.  The map below shows  a supervised classification of the city of Germantown, Maryland. It was created by using the inquire(legacy), the growing properties, and signature editor features in ERDAS Imagine with the given coordinates. The map shows different types of land use and their coverage. The unique classes chosen were Urban, Road, Deciduous Forest, Mixed Forest, Fallow, Agriculture, and Water. The area of each is shown in the legend in acres. A inset map of the distance image map is also included.

 This lab focused on spatial enhancements which is a preprocessing step in preparing imagery for use. Enhancements modify data and imagery to help users interpret the data beyond what is initially apparent by using a variety of methods. This lab also focused on how to view properties of multispectral data and create band indices.  ArcGIS and ERDAS Imagine were used in this lab to examine an image of the surrounding areas of the Olympic Mountains in Washington State. Most tasks required to complete this lab can be done in either program however using both made the lab faster and was able to have more thorough information. Descriptions of three features were given and the task was to identify that feature within the given image by examining the histogram, examining the image as grayscale, examining the image as multispectral, and using the inquire cursor to find brightness value. As well as use a different band combination for each map. The first feature to identify was: In Laye...

This week was an introduction to ERDAS Imagine 2011 and how to navigate it. I found ERDAS Imagine fairly easy to use and navigate. There are some similarities between ERDAS Imagine and ArcGIS Pro. An image of forested land in Washington State was used to create a map. The Inquire Box tool was used to create the subset map found in the map layout below. A column Area was added to the attribute table to obtain the area of each feature in hectares. The image of the map was then opened in ArcGIS to create a map layout found below. The name of each class was changed to show the hectares of each. 

Land Use Land Cover (LULC) and Ground Truthing were covered this week. LULC classification is used for urban planning, natural resource inventory, and others. There are multiple classification systems to identify features but the USGS Standard Land Use/Land Cover Classification System was used in this lab.  There are three types of ground truthing: calibration, accuracy assessment, and field spectra. In this lab accuracy assessment was used to verify that classification/interpretation was correct. This type of ground truthing is done soon after image acquisition. The map below shows an area of Pascagoula, Mississippi. The different colored and labeled polygons represent different features within the image based on the classification system used. The points represent the ground truthing accuracy assessment that was done using random sampling points; where the green represent points that were accurate and red represents points that were inaccurate.  

 This weeks lab was an introduction to Aerial photo basics and the methods and techniques used to visually interpret them.  Aerial photography provides a unique perspective that is used in multiple fields. Since aerial photographs are taken from an elevated position they can reveal features and patterns that we may not be able to see. Understanding the basic techniques and methods to interpret these images can provide valuable insight to the area.  The recognition elements focused on in this lab include: tone, texture, shape and size, shadows, pattern, and associating. The image below depicts areas of different tones and textures. The red outlined areas represent various tones: very dark, dark, medium, light, and very light. When examining the image for tone it is important to keep in mind the uniformity of the tone within the area. While the purple outlined areas represent various textures: very coarse, coarse, mottled, fine, and very fine. Withing the picture the very f...

This weeks lab focused on the effects of scale and resolution on the properties of spatial data, and the Modifiable Area Unit Problem (MAUP).  Vector data is highly dependent on scale, at larger scales(closer view) features are represented with more detail. While at smaller scales(zoomed out) the data will have less data which may require generalization when analyzing. Vector data should be adapted when using different scales to maintain interpretability. Resolution within raster data refers to the size of the grid cells/pixels used to make the dataset. High resolution raster will have smaller pixels, which creates better detail. Low resolution raster will have larger pixels, which may lead to generalizing the information within the raster. Choosing the best resolution is important since it can directly impact the type of analysis and its accuracy. Gerrymandering is the manipulation of the boundaries of an electoral constituency so as to favor one party or class. District lines are...

Spatial interpolation is the technique using points with known values to estimate other areas/points. It is typically used to create maps for precipitation, elevation, and more. In this lab we used four different types of interpolation methods to create a visual of the water quality in Tampa Bay. The four methods used for the estimation of Biochemical Oxygen Demand (BOD) in mg/L were Thiessen, Inverse Distance Weighted(IDW), Spline regularized, and Spline tension.  The Thiessen method proportionally divide and distribute point coverage into polygon regions. Each polygon contains only one point feature. The disadvantages of this include not being suitable for more complex surfaces and it creates sudden boundaries which is unrealistic for most data sets. Thiessen Interpolation The IDW method determines cell values using a linearly weighted combination of sample points. This method assumes the variable being mapped decreases in influence with distance from sampled location. A disadvan...