Every living thing requires energy. Animals, including humans, gain energy from respiration, where oxygen is exchanged for carbon dioxide; while plants absorb carbon dioxide from the environment and produce oxygen. The net ecosystem exchange (NEE) is how much carbon is put into the environment compared to how much is removed. This concept and our ability to simulate it is central to understanding future strategies related to climate change.
In the last few decades, researchers have simulated NEE mainly through process-based models which are data-intensive, time-consuming and costly, as highly detailed data is needed to develop the model. Babak Safa, DWFI post-doctoral researcher, and colleagues, tested a way to simulate NEE using artificial neural networks (ANNs). These models use existing long-term data to find the best match between historical inputs and corresponding outputs. ANNs provide prediction accuracy and great flexibility to researchers. They can also be used to find patterns and relationships between plants and weather that are non-linear or complex. ANNs are a form of artificial intelligence (or machine learning), whereas traditional models are based in a statistical approach. ANNs may provide an approximate solution to a real problem, rather than an exact solution to an oversimplified one.
Researchers in this study tested two cornfield sites (rain-fed and irrigated) near Mead, Nebraska, to simulate NEE using ANN. Their results showed a high correlation between actual and estimated NEE values, meaning the technique is both reliable and efficient. These findings show of the effectiveness of ANNs in future estimation of net carbon exchange rates by both scientists and policymakers.
View the full journal article.