Guest Scientist Blog by Jacqueline Castellanos, NASA JPL Summer Intern
My summer internship at NASA Jet Propulsion Laboratory exposed me to all parts of the scientific process. I went on a hike in search of mosquitoes, collected data, and I had the opportunity to teach teachers how to use the GLOBE Observer Mosquito Habitat Mapper. In my analysis, I learned how to work with data of varying quality and saw how environmental factors affect mosquito abundance. Most importantly, I learned that data given context and explored in meaningful ways can help us understand the problems we are facing today.
In my final project, I explored correlations between remote sensing variables and mosquito abundance data collected by citizen scientists using the GLOBE Observer Mosquito Habitat Mapper. The graphs below show larval counts for Shumate Middle School, Michigan along with remote sensing measurements of precipitation from GPM IMERG and modeled estimates of air temperature from GLDAS. These initial results suggest that warm spells in the fall and winter and warming temperatures in the spring might be correlated with larval count.
Most people don’t realize all the work that goes into getting to the place where you can actually create a graph and complete an analysis. First, I mapped all the mosquito observations made in the GLOBE Observer application. This map helped me get a general idea of where most of the observations were being made.
Next, I needed to find observations that could be used to create time series and be correlated with remote sensing variables. The observations needed to be taken within close geographic proximity to each other, include a larval count, be taken at roughly equal time intervals, and have 30 or more days of observations at a location. Now that I had a list of criteria, it was time to design a method to find the observations.
I separated the data into smaller groups those that included and did not include the larval count. I looked more closely at the larval count data set to find the organizations that had observations over thirty days or more. Then, I spatially clustered the data for each organization based on radius. Clusters that still had observations over thirty days or more were selected. The longest set of observations spanned 61 days.
These methods helped me find the data that could be used to create time series and be correlated with remote sensing variables. The remote sensing variables that I explored were mosquito habitat conditions such as air temperature (from GLDAS), soil moisture (from SMAP), precipitation (from IMERG), and the vegetation normalized difference vegetation index (NDVI – from MODIS). The initial results were exploratory but longer and continuous time series measurements of larval count (~ >4 months) may help identify clear trends with the remote sensing variables.
Dr. Erika Podest and Dr. Helen M Amos told me about the efforts being made to engage people in citizen science such as training sessions and competitions. I enjoyed learning about the citizen scientists that made the observations, and understanding this context explained some anomalous spikes that I was seeing in the data. Most of the time, spikes in the number of observations could be explained by a training session, competition, or a group of students making observations.
I had an amazing experience this summer interning at the NASA Jet Propulsion Working on my project made me more interested in learning about data science and earth science. The experience made me more inclined to pursue a masters degree or a PhD. It has been an honor working with scientists that are eager to promote science and teach others about the environment and working with data collected by people around the world willing to volunteer their time to help us better understand our environment.