PNNL

As the world’s most urgent problems in science and energy grow in size and complexity, researchers from the Pacific Northwest National Laboratory (PNNL) are ready to meet these challenges with algorithms, software, and artificial intelligence. Recently, Advanced Computing, Mathematics and Data (ACMD) Division staff, Erol Cromwell and Yan Shi, leveraged their software engineering skills to create a program capable of analyzing data linked to climate change. The results of this study were published in Scientific Data.

Assessing the cause and effects of climate change is no easy task. These include many complex factors that play a role in global atmospheric variations. One such factor is the number of aerosols, or fine particles, suspended in the air. These can be from dust, pollution, or wildfire smoke—among other things—and can seriously impact air quality. A purpose of the Atmospheric Radiation Measurement (ARM) user facility is to collect measurements that help researchers better understand atmospheric processes, including those influenced by aerosols. “User facilities, such as ARM, collect a lot of data. Researchers will access the datasets and want to analyze them in a certain way to answer a specific scientific question. Within ACMD, we help scientists achieve this by developing software and data products that meet their needs,” states Cromwell.

Since 1997, ARM staff members have deployed specialized instruments at the Southern Great Plains Central Facility in central Oklahoma to collect data on aerosol optical depth (AOD)—or the amount of sunlight that is blocked by aerosols. Over the years, different instruments collected data with varying quality and consistency. After nearly two decades of data collection, a team from PNNL’s Atmospheric Sciences and Global Change Division wanted to combine these data into a single dataset in such a way to preserve high-quality information and minimize any shortcomings in the records. To do this, they relied on the software development and data analysis expertise within ACMD.

Cromwell assisted the research team with their investigation by developing a software to normalize AOD data and provide statistical analysis. Shi also supported this research by advising on data acquisition and quality control. “Because the data are measured with different instruments, they need to be adjusted to the same scale and grid to be comparable to each other,” states Shi. Together, the researchers were able to produce a first-of-its-kind AOD dataset spanning 21 years.

Despite not being atmospheric scientists, Cromwell and Shi used their backgrounds in software engineering to provide a platform for atmospheric data analysis. The ACMD team has also proven itself more than capable of handling data across diverse scientific fields, from cybersecurity to molecular chemistry. “This type of work highlights the importance of proper scientific data management and the ability of software engineering to enhance the science,” says Angela Norbeck, group leader of the ACMD Computational and Data Engineering group.

She notes that as the significance of their work becomes more apparent, ACMD staff will continue to design next-generation code and algorithms in anticipation of future challenges.

Read more about this research on the PNNL website and ARM website.