• No results found

1.4. Research Design and Methodological Approach

1.4.2. Data

In May 2015, following my proposal defense, I began a course of research that included extensive fieldwork in the Colombian Amazon, examining archival materials on regional development, conducting in-depth interviews, and producing an original

geodatabase on cumulative deforestation.18 I decided to focus on the Amazonian foothills because the literature had already identified that region as a site of deforestation, both historically and presently.

During the course of my research, I developed a comprehensive understanding of Amazon deforestation that foregrounds the role of history (Pierson, 2004) and integrates the study of local dynamics and perceptions (a view from below) with large-scale transformations (a view from above). Although my empirical chapters and appendices will provide concrete information about my sources, it is worth mentioning here that the “view from below” is based on materials from three historical archives on regional development,19 90 in-depth interviews with local authorities and leaders,20 and a

systematic analysis of the relevant secondary sources. The “view from above,”

meanwhile, relies on the technical processing of at least 50 Landsat satellite images (see Appendix A).

Although the use of satellite imagery in political science is still uncommon, the number of research projects using remote sensing techniques has increased markedly in recent years.21 As a qualitative researcher, I was at first tempted to discount the use of satellite images. However, because land-use changes are typically measured with satellite images and remote sensing techniques (Busch & Ferretti-Gallon, 2017), I decided to

18 I defended my dissertation project before the FARC demobilized, at a time when the environmental

impact of the guerrillas’ withdrawal was unknown.

19 I reviewed the historical archives of the Corpoamazonia, of Caquetá governor’s office, and of the

National Department of Intendancies and Commissariats (DAINCO).

20 All interviews were conducted during trips to Putumayo, Caquetá, Bogotá, Pasto, and Neiva between

June 2015 and May 2018. Most fieldwork was conducted between mid-2015 and early 2017.

21 See for example Giraudy & Luna (2017), Harbers (2015), Hollenbach, Wibbels & Ward (2012),

invest the additional time and resources to understand them. I have consequently come to believe that interdisciplinary collaboration is easier when researchers proactively utilize contributions from multiple disciplines. The disciplinary division between the social and natural sciences when it comes to studying the environment has a methodological dimension, in which mutual mistrust of unfamiliar methodologies limits our ability to properly understand complex environmental phenomena like tropical deforestation.

By taking advantage of the strengths of remote sensing techniques, my dissertation attempts to mitigate some of the negative effects of sharp disciplinary boundaries. While I build on remote sensing, I also recognize its limitations. This

moderate optimism explains my decision to combine views from above and below. While satellite images can represent large-scale transformations, they cannot make sense of social and political dynamics on the ground. Similarly, qualitative strategies of data collection can shed light on the social complexities of Amazon deforestation, but they are poorly suited to measure large-scale landscape transformations. In addition, satellite images are neither produced by the Colombian government nor mediated by local perceptions, which are important concerns for most data collection efforts (Herrera & Kapur, 2007).22

To be sure, there is an emerging industry dedicated to measuring land-use changes and deforestation with satellite technology.23 These data improvements facilitate state-

22 See Herrera & Kapur (2007) for an interesting analysis of the way politics affect both data collection and substantive conclusions.

23 See, for example, Global Forest Watch’s impressive effort to measure annual deforestation globally since

2001: https://www.globalforestwatch.org/. In Colombia, it is important to recognize the contributions of the

building in the Amazon because standard measurements are basic tools for controlling territories and populations. Official representations like “maps, censuses, cadastral lists, and standard units of measurement” constitute “techniques for grasping a large and complex reality” (Scott, 1998, p. 77). Unfortunately, such efforts tend to assume very limited time horizons and rarely take history into account. Available data on tropical deforestation are typically produced by state authorities attempting to prevent

deforestation, or by environmental activists interested in measuring progress (or the lack thereof). Unsurprisingly, these policy-oriented data-collection efforts are geared towards “monitoring” and the issuing of “alerts.” Because academic research on deforestation is largely dependent on this data, most existing literature on the drivers of Amazon

deforestation focuses on transformations occurring in the twenty-first century. Given that a central contribution of this dissertation is to emphasize the

importance of history, and because available data is rarely historical, I decided to produce a unique historical geodatabase on forest cover for my study area (see Figure 1.4). I have processed and technically classified five sets of ten Landsat images (one set per decade) beginning in the mid-1970s.24

24 See Appendix A for technical details. To build this geodatabase, I first completed three specialized

courses on Geographical Information Systems (GIS), Digital Cartography, and Remote Sensing at the Colombian Geographical Institute (IGAC). I subsequently benefited from the superb research assistance of Nicolás Herrera and the constant advice of Osman Roa, an expert at the IGAC.

Figure 1.4. Processed Landsat images

Data for the Amazon region from SINCHI, for altitude from INDEAM, for Landsat Images from the U.S. Geological Survey (coauthored with Nicolás Herrera).

This database allows me to measure cumulative Amazonian deforestation over time since the mid-1970s. Historical and unprocessed satellite images are accessible on the internet through the U.S. Geological Survey. With images dating back to 1972, Landsat is the world’s oldest continuous data collection project to utilize remote sensing techniques. By taking advantage of this freely-accessible raw historical data, my

foothills of Colombia. To the best of my knowledge, my dissertation is the first effort to measure cumulative deforestation in the area of study over such an extended period of time.