We are excited to announce that our lab’s latest research, led by Haoran Liu with contributions from Fa Li, Hamid Dashti, and Min Chen, has been published in Remote Sensing of Environment:
“Hyperspectral surface reflectance improves GPP estimation in terrestrial biosphere modeling using model-data fusion”.
Gross Primary Productivity (GPP), the total carbon uptake by plants through photosynthesis, is a cornerstone of the global carbon cycle and a key buffer against rising atmospheric CO₂. Yet, terrestrial biosphere models (TBMs) struggle with large uncertainties in GPP estimates due to limited information on vegetation traits. Remote sensing offers a way forward, but most existing approaches rely on derived products like Leaf Area Index (LAI), which introduce significant errors.
In this study, we tested whether surface reflectance, particularly hyperspectral data, can more directly and effectively constrain biosphere models. Using our Terrestrial Ecosystem Carbon cycle simulator (TECs) with an embedded radiative transfer model, we carried out both simulation-based (OSSEs) and real-data (OSEs) assimilation experiments at Harvard Forest. We compared three data streams:
-
Hyperspectral reflectance from the PRISMA,
-
Multispectral reflectance from MODIS,
-
MODIS-derived LAI.
Key Findings
-
Hyperspectral reflectance outperformed multispectral reflectance and LAI, cutting GPP estimation errors by nearly half under controlled experiments.
-
Real-world tests confirmed these advantages, showing hyperspectral data better captured seasonal dynamics of canopy structure and leaf chlorophyll content.
-
Both hyperspectral and multispectral reflectance outperformed LAI, demonstrating the value of using raw reflectance data to “cut out the middleman” and avoid compounding uncertainties.
Looking Ahead
With the rapid expansion of hyperspectral satellite missions—such as NASA’s PACE and EMIT, and ESA’s CHIME—our findings underscore the transformative potential of hyperspectral remote sensing for reducing uncertainty in carbon cycle modeling. This work moves us closer to more accurate predictions of ecosystem responses to climate change.
📄 Full article available here: Remote Sensing of Environment