Practicum Reflection -- Identifying Erroneous Data Points for Crop Mapping in Kenya

The Topic: Identifying Erroneous Data Points for Crop Mapping in Kenya

• What your practicum site was: My practicum site was helping out a research team at the Department of Geographical Sciences at the University of Maryland, in conjunction with the NASA Harvest Africa program.

• How you found that site and the site supervisor, and what advice you would give to future SGC Scholars on identifying a practicum site (such as how to contact a supervisor): I found this site by reaching out to many different professors and researchers on the staff page for the UMD Geography Department. Most I do not recall getting a response from, or some said they did not have work available, but I persisted with emailing additional professors. Eventually, I got a response from a researcher, found the topic interesting, and the rest is history. For future scholars, when looking for a research opportunity, don't feel discouraged if you don't receive a reply at first, just keep on emailing more professors to ask if they have any spots open for research, and at least one will surely get back to you.

• What tasks you performed at this site: I worked on the Fall Army Worm dataset of ground truth data used for training models for the NASA Harvest Africa program in order to recognize crops in satellite images. At this site, the researchers have found that with many datasets there are some erroneous samples that are incorrectly located on buildings, roads, etc. instead of fields (likely due to GPS error). Thus, I helped with this task manually by going through and analyzing crop data by hand in Google Earth Pro (a GIS tool), renaming points as to whether they were "crop," "non-crop," "not sure," or "nd" (no data). I reviewed a subset of this Fall Army Worm dataset in sub-Saharan Africa, which in total has about 34,000 points.

• What you learned about the science related to your site as part of the experience: I learned about the need for crop security in sub-Saharan Africa due to the severe impacts of climate change and recent events, such as deforestation, droughts, etc. on agriculture. As the impacts of climate change become more severe and visible, the impact on food security and crop production will only grow, unless we can find ways to mitigate or adapt to the threats of climate change. My research is helping to train models to serve in a test dataset for evaluating the automated algorithms for outlier detection in a case study for the DORA (Domain-agnostic Outlier Ranking Algorithms) project recently funded by NASA.

• How doing the work at this site has affected you beyond science. Specifically (answer all of these that apply to you):
  • What appreciation was gained for the practical connections between Science and other parts of Society: I learned how scientific research using data analysis can play a role in addressing the need for crop security and food scarcity in developing nations around the globe. Also, the points I manually cleaned will be used to validate the research team's outlier detections (i.e., to check that the outliers detected by the model are the same points I removed because they are not in fields/cropland). This work will have practical impacts towards addressing food security in developing nations, such as in sub-Saharan Africa.

  • How (if at all) you have modified your future plans at UMCP or your post-bachelors academic plans based on this experience: I have not really changed future plans, but I will use my experience to help me find future research or potentially career opportunities in related fields that could be connected to this research. I could look into continuing with other NASA Harvest programs or other NASA programs outside of Harvest, or other research opportunities with the Department of Geographical Sciences at University of Maryland.

  • How doing this project has affected your future career plans, if at all: My career plans have not really changed. To be honest, I am not fully sure what I want to do yet specifically as to my career at the current moment.

Last modified: 10 May 2021