Using Remote Sensing to Uncover Changes in Haiti’s Land Use
Narratives of Haiti’s degraded landscapes are as notorious as narratives of its political instability. How Haitians use their land, and its level and quality of vegetative cover have spurred millions in aid funding and have been the source of intense academic debate surrounding environmental well being of the country. Pressures like land grabs, climate change, mining exploration, and food shortages exacerbate land issues in Northern Haiti and both contribute to and are spurred by land manager decisions. Despite some pushback on these narratives, empirical research linking local practices with measures of land change over time has been lacking, and may not capture the way that land managers react to contemporary contexts of political and economic instability, and how they may actively care for land.
In this project, we use land manager knowledge combined with high-resolution, satellite remote sensing to understand how agricultural and forested lands are changing, why those changes are occurring, and how managers are taking action on their land to forestall climate pressures and risks of land appropriation by international institutions and domestic elites. This project was funded, in part, by a grant from the CECE Faculty Research Incubator Program.
As of September, acquisition of both moderate and high-resolution remote sensing data is complete. That means we have Landsat satellite data (30 m pixels) from 1984 until present and PlanetScope data (3 m pixels) from 2018 to present. The former allows us to establish historical baselines of, say, vegetation cover and health, and to track changes to land cover at an annual time step. We are able to monitor both gradual changes such as increased vegetation cover and abrupt changes from disturbance.
A notable abrupt disturbance occurred on this landscape around 2013 when the Caracol Industrial Park was built on agricultural land (Figure 1). This same development can be viewed in our Landsat data in Figure 2A (at location “2”) and Figure 2D. These figures show that there was some large, negative change to vegetation cover and allow us to fingerprint the timing and the magnitude.
In addition to abrupt disturbances, the PlanetScope data shows gradual changes on individual agricultural fields on a near-daily basis. Prior to 2018, satellite pixel sizes were too large to monitor smallholder fields on a regular timestep, inhibiting our ability to monitor crop type and productivity changes. Figure 2B shows this data: fields are different colors indicating the timing of crop growth and harvest. White lines between field indicate trees that are always green, demarcating field boundaries. Figure 2G is an example of one year time series data from location “4” where this crop reaches peak greenness around day of year 180 (aka July 1). We will use this dataset to track how farmers use their fields in terms of cultivation choices and climate responses such as tree planting. These data coupled with field survey information will address key knowledge gaps regarding land use and land value in Haiti.
These images will help us get a sense of the various types of greenings, brownings, and sharp land changes that have occurred. As we turn to the participatory mapping elements of the research project, we’ll start to get a better sense of how these changes align with land manager experiences during the same periods. This info will set the foundation for working with land managers to collectively understand abrupt land changes, as well as reflecting on what land characteristics might influence long term greening or browning.
Figure 1: The northern portion of our Haiti study area featuring the Caracol Industrial Park
Figure: Example remote sensing datasets used in this project.
A) Greening and browning trends from 1984-2024 within 30 m Landsat pixels;
B) Zooming into the red box from A to view individual field data colored by crop growth timing in 2023;
C) 1 m resolution tree cover data from Meta;
D) The vegetation greening trend occurring at location “1” in panel A from 1984 to present;
E) The Caracol disturbance around 2013 at location 2 in panel A;
F) Example of near-daily imagery in 2023 from field 3, recovering from drought;
G) A crop that peaks in greenness around July 1 (or day of year 180) at field 4.
