Introduction Post:
Hi everyone! My name is Steven Sy and I’m a fourth year Integrated Sciences student specializing in Environmental Management. The three main focuses of my degree are Environmental Sustainability, Ecology and Economics of Resources and I believe that GIS can play a crucial role in fulfilling all three of those goals. I’ve always been interested in how we can use GIS to analyze problems and discern revealing patterns in the data that can be used to find creative solutions. In the past few years of my undergraduate degree, I’ve taken a few GIS courses here at UBC, including GEOB 270 - Intro to GIS, 370 – Advanced Issues in GIS and am currently taking GEOB 373 - Remote Sensing. Although I thought I had learned everything I could about GIS, my friend encouraged me to continue on and take this course as well to further my knowledge. He had recommended that this course offered a more hands on approach to understanding the tools and concepts that GIS offered while hearing about some of its real-world applications. Personally, I’m excited to learn more about how GIS has been used in research and am hoping to get a better idea of some of the potential career choices that I could take on in the future. We all know that the concepts behind GIS have been used to create entertainment apps like Pokemon Go and major businesses like Disney have adopted the technology to create interactive maps of their theme parks but there are many more fields that GIS has been used for. In this course, we will be focusing on three major topics: Landscape Ecology, Health Geography and Crime Analysis. Although the term is just getting started I’m excited to learn and investigate the potential of GIS and its applications in research and beyond.
Entry 2 - Why is Geography Important?
Tobler’s Law states that everything is related to one another and that nearer things are more related than far things. This is a fundamental aspect of geography. If we can grasp geography we can start to understand how things impact other things and can start to make powerful connections about the phenomenon that we observe. Everything occurs in a space, time and organization and these interactions complicate things. Likewise, Patterns, Places and Processes interact with one another to empower Geographic Information Systems.
One of the important things that we learned about in class this week about geography is that there may not be a single natural scale. A core principle in Landscape Ecology is that the problem we are trying to solve is actively changing depending at what scale we are looking the problem in. This can be commonly described as the Modifiable Areal Unit Problem. A single tree can be viewed as its own system with unique defining traits and challenges. But if we zoom out and look at the entire forest as a system, the challenges take on a different form of its own. At a high scale, we may be concerned about the health of single particular tree, but at lower scales we could be more concerned about the forest’s composition and its inhabitants. Parameters, the effects of spatial autocorrelation and interactions are all dependent on scale. As we zoom in and out of a landscape, the changing reference frame directly impacts the parameters involved. What is constant at localized high scales, may be variable when we zoom out to observe the entire region. Likewise, the appearance of clusters changes with perspective. Interactions also change. As we zoom out when analyzing a problem, emergent processes from as the interactions get compounded and work together to form new interactions.
One solution that we can use to counter the effects of scale is the process of Gerrymandering the data. This allows us to redraw the boundaries of the dataset, effectively allowing us to analyze the situation at multiple scales.
Entry 3: Understanding Landscape Metrics
Landscape metrics help us to understand the connection between patterns that we observe and processes that are continuously occurring. Patterns can be observed in the form of spatial autocorrelation. Processes are what are causing the patterns to arise. These are divided into two main types based on the patterns cause: first order process involves patterns that are in response to environmental factors and second order processes which revolve around interactions between objects and events. These patterns work at different scales and are can sometimes be incredibly complex.
Landscape ecology is about understanding these processes through the examination of form. The landscape (aspects like topography and direction) informs how abiotic conditions shape up and how biotic interactions react to these conditions. Natural disturbances like wild fires and windstorms as well as human changes like climate change and habitat development affect and change the landscape. These changes then shift the ecosystem’s abiotic conditions and biotic interactions. The environment, organisms and space are all interacting with one another to create the processes that drive the patterns we observe.
Quantifying Pattern
1. Number of Land Classes
2. Texture Measures (fine or coarse)
3. Degree of Patch Compactness
4. Whether patches are linear or planar
5. Patch Perimeters Shape Complexity
Landscape Composition
Relative Richness - proportion of the number of cover types potentially present
Dominance - deviation from maximum possible evenness
Diversity - a reflection of richness and distribution evenness
Connectivity - how connected the patches are
Measures of Spatial Configuration
Probabilities of Adjacency
Contagion - how clumped are landscape patterns
Connectivity - how fragmented a habitat is
Proximity Index - the degree a patch is isolated in a landscape
Area Weighted Average Patch Size - the probability of randomly selecting a patch
Entry 4: Stats Review
Stats are important to help us look for relationships in the data, to assess how accurate our findings are and to help make predictions about the future. Having a proper understanding of the various statistical tools we can use for spatial analyses are a fundamental aspect of using GIS in research.
There are four categories that data can exist as: Nominal, Ordinal, Interval, Ratio. Each of the statistical tools has their own strengths and shortcomings while each of the data categories have certain inherent characteristics. It is important to know what kind of data you are working with so that the right statistical analysis can be applied without impacting or manipulating your findings. While it is important to be mindful of how data is processed and analyzed another key stage to be mindful of is the data collections process. Different Sampling methods can also greatly influence your results. For example, it is crucial that rarer classes get sampled often to ensure that we can be statistically confident that these values are not just outliers in the data set while building a better understanding of that particular class. Using a simple random sampling technique may not yield the best results whereas a stratified random sampling method will ensure that all land classes are equally represented.
Some of the statistical techniques we discussed in class include:
-Grouping Analysis
-Regression Modelling (ie. Ordinary Least Squares (OLS) or Geographically Weighted Regression (GWR))
-Spatial Auto Correlation
-and the use of AICc as an accuracy metric
Entry 5: Health Geography
Health, GIS and geography are all intrinsically linked. Where you were born, live and study in all impact our health experiences: the food we eat, the air we breathe in and the health care we have access to.
The study of health geography has evolved over time from the field of medical geography. Medical geography originated as the study of the spatial distribution of diseases with the amazingly named Jon Snow’s major discovery of mapping water borne diseases. He discovered that the conventional wisdom of how illnesses were spread were wrong and that they were actually being caused by contaminated water sources. His findings revolutionized how doctors combatted diseases and brought the power of spatial analysis to the forefront of the industry.
The advancement of GIS technologies has allowed for analyzing, mapping and integrating different data sets allows for novel data exploration. Today health geography can be viewed from a more contemporary perspective as we apply concepts of place and space into the analysis of medical geography. This approach understands that health problems are complex and are intimately linked to power relations in society.
There are 5 strands of Health Geography that analysts focus on
a) Spatial Patterning of Diseases and Health (Epidemiology)
b) Spatial Patterning of Service Provision (supply and demand of health care)
c) Humanistic Approaches to Medical Geography (illness as a social construct)
d) Structural/Materialist Approaches to Medical Geography (inequalities in healthcare)
e) Cultural Approaches to Medical Geography (therapeutic landscapes)
While these five strands are comprehensive, the world of health geography is constantly evolving. Today many believe that the study of Medical/Health Geography can be divided into two categories:
Geographical Epidemiology which looks at what factors are contributing to ill health
and the Geography of Health care which investigates how adequate provisional services are distributed. A mixed methods approach is becoming more and more prevalent as people are combining the benefits of both approaches to form a more complete view of health geography.
Entry 6: Crime Analysis and GIS
Watching TV shows like Blindspot and Criminal Minds using a combination of spatial analysis and criminal behaviours to analyze existing data of criminal activities to determine patterns and to help stop serial killers and planned heists was always exciting to see. Even high tech sophisticated map displays in Westworld are examples of what GIS can be used for. Learning about how GIS is used in the field of Crime Analysis brought to light how GIS provides the backbone for many of these exciting spatial analyses examples.
Geographic crime analysis isn’t just about visualizing where crimes occur, but also about how geography influences crime events. This field of environmental criminology involves the spatial distribution of offences and offenders. In class, we learned that crimes a human phenomena and so their distribution across space and time is not random. Brian described three theories to help investigate behavioural patterns of environmental criminology:
1. Routine Activity Theory
This theory relies on the crime analysis triangle that joins together the place, the offender and the victim. This theory follows people’s routines, the nodes of activity and that crime revolves around the idea that certain environments are predisposed to certain crimes. At its core this theory allows us to predict if conditions are right for a crime to be committed.
2. Rational Choice Theory
This theory gives us insight into what criminals are thinking at the time the crime event is occurring. This theory follows that people’s actions will based on what will lower their chances of being caught.
3. Criminal Pattern Theory
This theory analyzes how daily routines and familiarity influences crimes. The Criminal Pattern Theory helps us to understand where and when crimes will likely occur.
Two other interesting concepts that stuck out to me was the idea of the Donut Theory and the two type of criminals. The donut theory states that criminals will act within their general vicinity because they would be most familiar with the victims patterns and the possible escape routes. Along the same train of thought, people who live right by the offenders are actually safest because they don’t want to attract attention to themselves. Finally we learned about how criminals can either be classified as Marauders (who operate in their own neighbourhoods) or those that are Commuters (who will travel to other locations to commit crimes). These two concepts really emphasize that the analysis of crimes have a heavy component of People alongside Place and Patterns, and is crucial for effective criminal prevention strategies.
Entry 7: GIS in Fire Departments
Life and death situations depend on how quickly emergency respondents can arrive at a scene. The chances of survival during house fires and heart attack cases can plummet within just a manner of minutes. In this week’s class we learned about how the Calgary Fire Department has adopted the use of GIS technologies and practices to create better fire prevention and response strategies. With 32 stations and around 1,100 employed firefighters, it can be incredibly difficult to coordinate the Fire Department Force to effectively service the entire city of Calgary. Fortunately, GIS has been successfully adopted to improve the fire station’s ability to complete Risk Analysis, Response Mapping and Location Allocation.
In order for fire fighters to successfully plan fire fighting strategies they must be able to know which buildings are at risk of catching fires. The CFD has categorized fire risks and severity into four categories: Green, Yellow, Blue, and Red. Each class denotes the chance of the building catching fire and the extent of damages that it will result in. Having an easy classification system allowed firefighters to visualize risk on a map and to highlight areas of concerns right away.
With this information firefighters then focused on how to best optimize response strategies. They combined information about street widths, intersection data, distance to fire stations and fire hydrant locations to coordinate which fire stations are best equipped to service particular areas. The goal for these firefighters is to have at least 14 fire fighters at the scene within 10 minutes. A threshold of eight minutes determines whether or not a fire will spread beyond the room of origin where damage will exceed 50%.
Lastly we discussed how fire fighting facilities are not always allocated to be equitable. There are many challenges, like balancing future developments projects to existing fire stations and of course the political and human effects that these policy decisions can have. The future is tough to predict but GIS can help these fire fighters make the most informed decisions.
In each of these cases, the fire fighters also understood that scale can be a challenge when looking at data (something that we addressed in our earlier classes). To combat this they built maps across many scales, both regional and localized to help discern patterns hidden in the modifiable area unit problem. Seeing what the Calgary Fire Department was able to accomplish was impressive and it was evident that GIS was the backbone that made it all possible.
Entry 8: GEM Presentations
Over the course of the term we were able to hear some of the presentations about some of the projects that the GEM students were working on, other groups paper reviews as well as the 479 term projects. It was interesting to see how some of both these GEM grad students and fellow peers are using GIS to study their projects. We learned a lot about topics that would normally never be mentioned in class and hearing what they had to say helped to inspire our own work. In this blog post I will be discussing some of the topics that caught my attention:
GIS and Elk Mapping:
Although we have heard about how Elks are endangered species many times before I thought this GEM project was particularly unique because it introduced the concept of the “Step Selection Function”. By creating a suitability surface across the landscape on the habitat’s conditions, this project predicted the probabilities of each cell that an elk would choose it when it takes its next step. This modelled the belief that these animals would always choose to make the optimal decision and was a novel approach at mapping animal behaviours spatially.
Drug Use and Overdoses:
BC has the highest drug overdosing cases well above the national average. This presentation showed the importance of context informing the story told by raw data. Population density was identified as a factor that could greatly influence overdosing results. Areas of high population density could signal that there are more people to see you overdosing and that would mean there are likely more people who could help to save you. Meanwhile, low population density regions could mean that fewer people are getting saved.
Food Desserts:
Another interesting topic that was showcased during class presentations was the issue of food desserts. Food desserts are areas that are unable to properly provide food security. Contrary to what I had in mind, this is actually a growing area of concern in the US. As businesses continue to grow and dominate through economies of scale, smaller enterprises are getting forced out as they are unable to compete with the lower prices. This creates a vacuum where residents are forced to travel further distances to get to the bigger chain grocery stores. Unfortunately travel distances exist as an externality cost and although food prices are cheaper, people are forced to spend more on transportation costs. The presenters highlighted the importance of using distance costs in future analysis of food desserts to help identify where this issue is affecting underprivileged residents’ ability to access food.
Spatial analysis of Suicidal Bridge Jumping in Metro Vancouver from 2006-2014:
This presentation investigated which bridges in metro Vancouver were being chosen to act on suicidal thoughts. It was originally believed that most victims will choose a familiar path that uses the least effort to travel to and will often select bridges based on proximity to their node of activity or their residence. The researchers analyzed the location of the jumps that occurred between 2006-2014 in comparison to their addresses and found 145 cases. They concluded that victims were willing to travel further distances to get to the famous bridges (ie Alex Fraser, Lion’s Gate and Patella bridge). Lion’s gate in particular showed this phenomenon. Of the 145 cases, only six individuals were convenient jumpers that used bridges within their vicinity. Contrary to popular belief this study proved that a different approach to preventative strategies was necessary. In thinking about why the opposite effect was happening, some have suggested that it was because these victims felt that these bridges were more recognizable and that their suicides would be more notorious. An alternative reason that I thought about was that these victims didn’t want to be nearby the people that they were comfortable in and didn’t want to put them in that situation.
Forests are Havens for Criminals:
One of the biggest challenges facing forest law enforcement teams officers are that forests are an ideal haven for criminals. Dense forests are perfect hideouts as they are geographically isolated and understaffed. It is easy for these criminals to hide in the forest where officers can have a difficult time tracking them down. Despite knowing about this issue it was interesting to hear that government bodies don’t yet have an adequate method of approaching this problem as it is tough to assign jurisdiction of the forests to the different government bodies. I will be very cautious next time I decide to hike in the forests.
Machine Learning in Geographically Weighted Regression Analysis:
In our term project presentations, one of the interesting projects that I saw was the use of machine learning softwares to make Geographically Weighted Regression Analysis more efficient. He discussed using an Extreme Gradient Boosting Algorithm and Supervised learning algorithms to help in the decision making process. In simple terms he based the software on decision tree statistics to help split data into groups of categories. Based on these categories you assign specific boosting values to your GWR analysis to show which categories are more important than others. Although I didn’t complete understand some of the concepts involved this was a creative approach to assign more rigour weighting allocations.
All of these presentations underscore the importance of geography and using GIS to Research crucial issues that we are dealing with. The vast variety of topics showed the versatility of using GIS across multiple fields and helped to inspire the methodology I used for my own term project. Although this is just a small sample of the applications of GIS, there is still plenty to learn about.
Entry 9 - Where do we go from here?
4 months, a couple labs and a term project later and now we are wrapping up the course. Geob 479 - Research in GIS is coming to a close. 13 lectures later and we’ve learned about some of GIS’ many applications. From utilizing GIS to map crime hot spots, to investigating illness trends and how to best respond to them, and even researching how ecological issues will respond to a changing landscape. We have just begun to scratch the surface about the potential of using GIS for research purposes. Through all of the lectures, class presentations, lab work and even our term project it is clear that GIS is at the forefront of analyzing data and predicting trends.
Despite all of the information that we gained throughout the term it is clear that I still have plenty to learn. Although I am not graduating this year, I am considering going to grad school once i’m done here at UBC. I’ve been looking at programs abroad about Environmental Management and even UBC’s own GEM program. In the meantime I’m looking at employment opportunities in the consulting industry at companies like Next Environmental Consulting or Triton. I am also considering doing a directed studies in my last year of my undergraduate. I’ve learned a lot about using GIS and I’m excited to apply what I’ve learned in the future throughout my career.
This course has been an enlightening experience, thanks for reading along!