Figure 1: Well pad in Jonah Field, featuring shrub overstory and herbaceous understory. (Photographed On June 26th, 2025, by Jianing Tian)
Over the past few months, I have been working on the estimation of aboveground net primary production (ANPP) of herbaceous plants in the Jonah Field, Wyoming. To achieve this, I applied the Monteith model, which relates plant productivity to absorbed photosynthetically active radiation (APAR), incident solar radiation (PAR) and radiation use efficiency (RUE). Specifically, I used field-based biomass harvest data and APAR from the Sentinel-2 normalized difference vegetation index (NDVI) to calibrate RUE values. RUE and APAR were then used to estimate large-scale ANPP.
Why Focus on Herbaceous Plants?
The plant community in dryland ecosystem in Jonah Field is dominated by the shrub Artemisia tridentata, with an understory of bunchgrasses (such as Achnatherum hymenoides, Elymus elymoides) and both perennial and annual forbs (such as Phlox hoodii). Herbaceous plants (grasses and forbs) are preferred by livestock over shrubs due to their higher nutritional quality and palatability. Therefore, quantifying the ANPP of herbaceous plants is important for understanding the availability of forage for grazing animals.
Figure 2: A close-up view of shrub overstory and herbaceous understory within a sampling quadrat in Jonah Field. (Photographed On June 26th, 2025, by Jianing Tian)
Besides, herbaceous plants are generally more responsive to environmental changes and reclamation treatments over shorter timescales due to their shorter growth cycle (often within a year or a few years). These species tend to dominate early successional stages following disturbance and are often the primary targets of restoration and management efforts in dryland ecosystems like the Jonah Field.
From a remote sensing perspective, APAR is collected from NDVI using satellite images, which is best suited for estimating productivity in actively photosynthesizing, short-statured vegetation. Including woody shrubs in these calculations could introduce noise, as their NDVI signals may reflect long-lived, non-photosynthetic woody tissues or evergreen components that do not represent current-year productivity.
LOESS (STL). STL is statistical technique that breaks down a time series into trend, seasonal patterns, and residual (remainder) components. The trend component underlies long-term trend largely driven by shrub cover, which remains relatively the same and changes slowly over years while seasonal patterns show a more dynamic change that largely indicates the annual and perennial changes of herbaceous plant growth, which peaks during growing season and declines rapidly after senescence.
The use of STL decomposition enables me to refine the estimates by distinguishing the contributions of herbaceous species from woody shrubs in the NDVI signal. In this way, I could generate more accurate estimates of ANPP for herbaceous plants on both disturbed and undisturbed sites.
STUDENT RESEARCHER
Jianing Tian – Western Resource Fellow | Jianing is a Master of Environmental Science candidate at the Yale School of the Environment (YSE), where she specializes in dryland ecology and ecosystem recovery. She is particularly interested in the carbon cycle, including carbon emissions, soil carbon sequestration, and overall ecosystem carbon balance. Before YSE, Jianing earned a dual B.S. degree in Environmental Science from Duke Kunshan University and Duke University. During her undergraduate studies, she conducted research on agricultural carbon emissions, and she now aims to apply geospatial analysis and remote sensing to study vegetation recovery and carbon dynamics in semi-arid landscapes. In her free time, Jianing enjoys jogging, hiking, and playing musical instruments. See what Jianing has been up to.
