DSAM Practicum
Project Proposal
My goal for the Practicum was to explore other methods of analysis besides the GIS mapping that I had worked on in my two previous DSAM courses. I felt that I had made strong gains in my understanding of GIS mapping over the previous courses and didn’t want to just refine the same process over and over. Two topics that I decided to focus on were learning about digital approaches to and visualizations of soil properties, and using spatial analysis tools in QGIS to explore sugar production within Jamaica.
Iteration One
I focused on soil properties for my first iteration. Based on my banana plantation mapping work in the fall, it appeared that there was a relationship between railroads and the spread of the disease. At this point, I wanted to figure out if there are any other relationships that are noteworthy, such how soil properties could contribute to the spread of the disease. Having no knowledge about anything associated with soil properties, my first task was to figure out if anyone had done research into what types of soil properties are conducive to Panama Disease. I discovered Marianne Bosman’s 2016 Master’s paper at Wageningen University. Bosman conducted a field survey in Costa Rica to determine the relationship between incidence rates of Panama Disease and a variety of soil properties. Among her findings were that Panama Disease incidence rates are higher under alkaline (high pH) conditions, high aluminum leads to high incidence rates, high acidity has high incidence, low potassium and phosphorus lead to high incidence, and high magnesium has high incidence rates.
​
After finding this information, my next task was figuring out how to actually obtain any of this data for Jamaica. Because of the lack of historical data, I decided to focus on contemporary soil conditions for this initial foray. Through my searching, I eventually found the International Soil Reference and Information Center (ISRIC) soil geographic databases. You can search their online database by country to see what data they have available. The database only holds two findings for Jamaica: the soil and terrain database for Latin America and the Caribbean and its associated parameter estimates.
​
After hours of frustration from not being able to make the data resemble anything coherent, I finally discovered that the data contained sub-shapefiles that can be joined to the primary shapefile for all of Latin America and the Caribbean. The main shapefile divides countries into several regions based on soil and terrain properties. The sub-shapefiles can then be used to create color gradations for whichever soil property you wish to examine. I created a few of these categorizations, including for nitrogen, slope, and pH. After creating maps based off of this data in QGIS, my main takeaway was that I really did not have a takeaway. The Shapefile divided Jamaica into five seemingly random regions, and I did not feel that the regions adequately accounted for much of the geography in Jamaica, especially the Blue Mountains. Additionally, I found the regions too broad to take meaningful results from them.
​
Overall, my journey with soil led me to two conclusions. First, I needed to conduct more research to figure out what some of the soil properties actually mean to determine if researching soil will be useful for me. I did not know enough about soil to attempt to use it in a meaningful way. Secondly, I left this iteration thinking that digital methods are probably not the best way to analyze soil in a historical context. Jamaica’s soil has undergone significant changes between the early twentieth century and today, and I question whether modern soil maps adequately convey historical changes.
Slope
Clay %
Nitrogen %
pH Level
Iteration Two
For my second iteration, I decided to look at the history of the sugar industry in Jamaica. As I did with banana geography, I wanted to see what the geography of sugar plantations looked like through mapping sugar plantation locations over a series of years in QGIS. Also as with the banana industry, no specific data on smallholdings is available so I had to rely on plantations to create a point-based geography. The Handbooks of Jamaica that I used to map the banana plantations also contain the same information (parish, acreage under cultivation, estate name) for sugar plantations on the island. I then had to type all of this out in EXCEL to put the material in a format readable by QGIS. After creating my spreadsheets for the years 1897, 1907, 1917, and 1922, I uploaded the files to QGIS. I ended up using a Jenks distribution to create my acreage sizes since QGIS calculates the distributions for you. Because I did not want separate ranges for each year, I put all of the plantations and acreages from the four years into one spreadsheet and calculated the Jenks distribution from there. This provided me with consistency across my four maps. After mapping the plantations, I compared these points to those of the banana industry, which shows how largely separate these two industries were from one another, with the majority of sugar plantations in western Jamaica and the majority of banana plantations in the eastern parishes. However, this is information I already knew, so having the map in this case did not provide much analytical value beyond visualizing this information.
​
After looking at the geography, I wanted to see how the total acreage of Jamaican sugar plantations changed over the years that I mapped (1897, 1907, 1917, and 1922). After adding up the plantation acreages for each year, I saw very little change over time. There were 25,347 acres of plantation sugar in 1897, 20,946 in 1907, 21,620 in 1917, and 22,842 in 1922. Seeing this, I was surprised at how little change there was between these years. However, I went back and looked at the change in overall sugar production, not just the plantations, between 1917 and 1922, the years where the banana industry began to see a large decline in acreage due to Panama Disease. For sugar, the total acreage increased from 33,830 to 53,794 acres between 1917 and 1922. This 20,000 acre increase corresponds to a similar acreage decline in the banana industry, which went from 78,485 acres in 1917 to 59,720 in 1922, a loss of about 20,000 acres. Comparing the increase in sugar acreage overall to the increase in sugar plantation acreage shows that only about 1,000 of that 20,000 acre increase came from plantations, meaning that roughly 95% of that growth came from smallholders. This revelation about smallholders was extremely helpful in shaping my dissertation, as it showed that the geography of where the sugar industry grew and where the plantations were located were two very different things.
​
Although the sugar mapping was useful for shaping my thinking, it did not really introduce any new methodological tools into my repertoire. As a result, I decided to try to learn how to use the spatial analysis features of QGIS. It took some time to debug QGIS, as the version I had installed initially had broken processing tools. After finally getting QGIS to cooperate, it opened up a new range of analysis tools for me to use. However, I have yet to find any that seem helpful for my research. One of the tools, Voronoi analysis, divides areas into cells, with each cell covering the region closest to a center. It is often used for determining where to place infrastructure such as post offices, as it shows the largest empty area among a collection of points. I just am not aware of a way that this would help with my capta. Out of all of the tools I explored, buffer analysis is probably the closest to useful for my project. With it, you can see spatial overlap between plantations that might facilitate easier movement of pathogens between them. The main problem right now is that the distances are based on degrees rather than latitude and longitude. I left this iteration largely frustrated about my lack of success in incorporating spatial analysis tools into my GIS repertoire.
Iteration Three
For my final iteration, I decided to double up on the work I was doing in my environmental modeling class. As I was learning a completely new tool and needed to do a large amount of work learning how to create an environmental modeling class, I felt that my remaining time in Practicum would be best spent making that project as strong as it could be. For more information on environmental modeling, click "See More" below or follow the link at the top of the page.