Embedded Files
Analysis of Understanding Global Spatial Patterns using Different Real-World Scenarios
Analysis of Understanding Global Spatial Patterns using Different Real-World Scenarios
Problem and Objective
Problem and Objective
Spatial patterns analysis provides insights into the spatial distribution of phenomenon and their relationships within a physical space or location. The purpose of this analysis is to explore the quantification of visual patterns to help understand global spatial patterns by examining if there is clustering in false alarm incidents and calls ranked as a high priority as well as the relationships of each feature to all other features within the boundary of Batallion 2. This analysis also aimed to examine the density of calls as well as the distance of density clustering for the fire department.
Analysis Procedures
Analysis Procedures
Strategies: The strategies adopted for this analysis are explained as follows. I used ArcGIS Pro (Version 2.9) software. I used the following Modules/tools for the analysis: Calculate Distance Band from Neighbor Count tool, Multi-Distance Cluster Analysis tool, Spatial Join tool, Calculate Field tool, Average Nearest Neighbor tool, High/Low Clustering tool, and Spatial Autocorrelation tool. Data and data types – The data include mxd files of tutorials and exercises layout maps containing (a) Calls for Service – January 2015 feature (b) Battalion 2 boundary feature (c) Incidence of calls – February 2015 feature (d) 300 feet Grid feature class (e) Calls for Service – February 2015 feature (f) Patron locations feature class and (g) Active and Proposed Stations feature classes. Data sources – GIS 520 Fall Semester, NCSU, and tutorial material is an updated version of David W. Allen (2016)’s chapter 8.
Methods: I imported the first map layout (mxd file) of Batallion 2. I opened the attribute table of the Batallion layer 2 to record the measured area (520175356.110283). I examined the properties of the incident – February 2015 feature class. Then, I created a definition query to select only the incident types that are between 700 and 745 inclusively. After searching for the Average Nearest Neighbor tool, I used this tool to calculate the distance of each feature point in the Incident of false alarm calls layer to its nearest neighbor. I checked the report of the analysis.
Next, I imported the second map layout(an mxd file)for the exercise. I used the Calculate Distance Band from Neighbor Count tool with neighbors set to 7 to access the threshold distance (i.e average distance) of the Calls for Service – Jan 2015 layer. Then, I used the High/Low Clustering tool to calculate the level of clustering of the Calls for Service feature using 6 different threshold distances (700, 800, 900, 1000, 1100, and 1200 feet).
Thirdly, I imported the third map layout (mxd file) for the exercise. I searched and use the Multi-Distance Cluster Analysis tool and which I initially run it without computing the confidence envelope. On the second attempt, I ran the tool using the confidence envelope. I added a new field to the output table from this analysis and I used the calculate field tool to calculate the difference between the observed K and the High Confidence Envelope using a mathematical query. I then created a line chart from this table where the y-axis is the difference and the x-axis is the Expected K.
Finally, I imported the fourth map layout (mxd)for the exercise. I joined the Patron Locations (points) feature class to the 300-foot grid layer using the spatial join tool. Next, I selected only valid grid that has join_counts values that exceed zero by creating a definition query for this layer under its properties. Then, I ran the spatial autocorrelation tool for the joined (Patron Location Grid) feature for 6 different test distances starting from 2800 feet with 200 feet increments.
Workflow diagram of the analysis
Results
Results
The map above represents the result of the nearest neighbor analysis indicating that there is a 90% probability that the clustering of false alarm calls was significant for fire incidence type that is between 700 - 745.
The General G statistic measures high or low values cluster. The result of the analysis shows that the peak z-score corresponds to the distance (900 US-Feet) at which the clustering of values is the strongest as shown in the map above.
The confidence envelope of Ripley's K function provides the basis for comparing the level of clustering and identifying the most significant level. The map above shows that 900 US-Feet is the peak of the graph and it is the distance of the most significant clustering of the priority call for service for January 2015.
The result of the spatial autocorrelation shows that the z-score peaked at a distance of 3,400 feet with less than a 1% likelihood that the clustered pattern exhibited by the library patron location is a result of random chance. Therefore, there is a significant clustered pattern in the library patron location.
Application & Reflection
Application & Reflection
Problem Description: The incidence of wood decay often causes serious loss in the yield of timber. Forest health protections aim to detect insect and disease outbreaks and provide assistance to land managers on how to manage the situation. The purpose of this analysis is to examine whether there is clustering in the recent outbreak of incidence of disease in a particular forest stand and monitor the spread of this disease among the trees in each of the forest stands. This analysis will further provide insights into the dynamics behind the disease outbreaks.
Data Needed: The data needed include (a) the study area (forest land) boundary (b) the point feature of individual trees in the case of diseases and (c) the tree stand boundaries
Analysis Procedures: I will add all feature class data to the map. I will use the Calculate Distance Band from Neighbor Count tool to identify the clustering of high-ranked (ranked from low to high) decaying incidence in the trees using the 7 neighbors value. Then, I will use High/Low Clustering to check the degree of clustering at a specific distance using the range of numbers (distances) above and below the average distance from the previous analysis as the threshold test distances. If there is a significant clustering by looking at the report generated (p-value and z-score), I will use the MultiDistance Spatial Cluster Analysis tool to identify the distance band for clustering as well as the spatial relationships between the affected trees and other trees within the study area. I will set the number of the distance bands for this tool to 10 and the weight field will be the column that represents the ranking of the degree of tree decay while the study area feature class will be the forest land boundary. I will also compute the confidence envelope. A new field will be added to the output table from this analysis and the difference between the Observed K and the High Confidence Envelope will be computed using the calculate field tool. I will create a line chart using this calculated difference and the expected K fields.
Page updated
Report abuse