For years, scientists have been using a method called \u201ctemporal autocorrelation (TAC)\u201d to measure forest resilience using satellite data.<\/p>\n
However, scientists have not used TAC uniformly, and its use as a resilience indicator was based largely on theories and assumptions rather than direct evidence from observations on the ground.<\/p>\n
A new paper from the Global Environmental Remote Sensing Laboratory<\/a> led by Zhe Zhu, associate professor of natural resources and the environment<\/a> in the College of Agriculture, Health and Natural Resources<\/a> (CAHNR), provides clear evidence and methodological guidelines for using TAC as a forest resilience indicator.<\/p>\n This work was led by Kexin Song \u201826 (CAHNR), a former UConn Ph.D. student and current postdoctoral associate at Yale University. It was published in Nature Ecology & Evolution.<\/a><\/p>\n Broadly speaking, resilience indicators reflect how well forests resist and recover from stressors like drought, fire, or human disturbances.<\/p>\n \u201cWe want to understand if there is a particular threshold or tipping point where even a small disturbance can push [a forest] to another state,\u201d Song says. \u201cThat\u2019s one of the reasons we track and monitor resilience, so we can better understand how close a forest may be to that threshold and, ultimately, help prevent it from happening.\u201d<\/p>\n One major limitation of TAC is that it extracts information from the residuals generated around observations of forests rather than from a signal that can be directly observed in the satellite data.<\/p>\n Scientists can measure something like when a forest gets greener or browns with the changing seasons. This is a predictable pattern. TAC, however, uses the residuals in the data around such predictable dynamics. TAC looks at how similar residuals are to those immediately before them to indicate how quickly a forest\u2019s vegetation recover from small stresses and, from that, infer its resilience.<\/p>\n \u201cThere are true ecosystem resilience signals embedded in what looks like a noise component,\u201d Song says. \u201cBut we need to be very carefully about how we \u2018filter\u2019, extract, and interpret those signals.\u201d<\/p>\n Song and Zhu discovered a method that works to accurately extract ecosystem resilience information from noisy data, which they describe in the paper. This helps address another shortcoming with previous applications of TAC, as there has no standard methodology, making it difficult to compare results across studies.<\/p>\n \u201cWe found the correct way,\u201d Zhu says. \u201cYou have to have the specific frequency, time period, and vegetation indicators.\u201d<\/p>\n Another key problem with TAC is that it was not tied to on-the-ground observations assessing forest health. International collaborators on this study collected data on hundreds of trees in the Amazon, representing over 100 species across nine plots.<\/p>\n The researchers matched 30 m-resolution satellite data with the specific plots in the Amazon that they sampled. They correlated the satellite data with plants\u2019 \u201chydraulic safety margin\u201d (HSM). HSM is a key measure of a plant\u2019s ability to tolerate and survive dry conditions.<\/p>\n The Amazon may be just the beginning. The findings open up new questions and research directions, including whether this method can help us understand how different ecosystems respond to a wider range of disturbances.<\/p>\n \u201cWe wanted to make sure we are building something on a concrete foundation, not on sand,\u201d Song says. \u201cThe universal idea is that we need to link our satellite data with ground-based measurements of ecosystem function and plant physiology.\u201d<\/p>\n <\/p>\n This work was supported by funding from the National Science Foundation, NASA, and Eversource Energy. <\/em><\/p>\n This work relates to CAHNR\u2019s Strategic Vision area focused on\u00a0<\/em>Advancing Adaptation and Resilience in a Changing Climate.<\/em><\/a><\/p>\n