Graphical Data Exploration

Initially descriptive statistics such as box plots were used for exploring data.  Due to the spatial nature of the data set these methods were unsuccessful. Two dimensional images of the agricultural field with each of the various variables were created using ArcGIS.  This method was effective for finding errors in the data set (G-1).  On the left hand side of (G-1), there are some extremely high values that do not fit in with the rest of the data. After consultation with supervisor these were deemed errors from experimenting with the model and removed (G-2). Recall that higher EC indicates higher soil moisture and nutrient levels. 

G-1. Image of field displaying EC values with errors.

Picture

G-2. Image of field displaying EC values after errors removed.

Picture
The 2D images were created for each of the variables to visually explore the data (G-3).  For the images colors closer to the red on the color spectrum represent lower values.

G-3. Image of field displaying aspect, gradient, plan curvature, profile curvature, SCA and SDA.

Picture
From the aspect graph, the red areas represent northern facing slopes and green are the more southern facing slopes. The north facing slopes receive less sunlight resulting in higher moisture.  This is apparent when you compare the EC and aspect images.

Gradient is steepness of the slope. Red and yellow colors on the image are steeper areas. In theory, steeper slopes are drier because there is less time for infiltration.  There is no strong, visual  similarities between the gradient image and the EC image.

Profile is the down slope curvature of the slope. The purples areas on the profile graph represent convex slope curvatures which shed water and are minimal in the field. The purple areas on the profile image are found in the red areas on the EC image both representing lower soil moisture. In general, concave, the blue areas, are more dominant throughout the field. 

Plan is similar to profile only it is the across slope curvature. The plan image is more difficult to interpret.  The lighter blue areas are concave, with higher moisture.  There is no strong, visual relationship between the plan image and the EC image.

The SCA and SDA images are a great visual representation for exploring the data. From the SCA, the blue areas have high values so are the areas receiving lots of water.  These blue areas from the SCA image correspond to the greener areas on the SDA image as the two variables have a slight inverse relationship. The green areas on the SCA correspond with the red areas on the EC image. There are visual relationships between the EC, SDA and SCA images.