Developing novel GPP estimation for crops at field-level using new-generation satellite data in the US Corn Belt

With rising demands of food, feed and fiber from a growing global population, the agricultural landscape plays an increasingly important role in the global carbon cycle. Gross primary production (GPP) is the amount of carbon uptake for plant growth that directly determines crop productivity, and it is also the largest carbon flux in terrestrial ecosystems. Accurate monitoring of GPP is critical for designing effective management practices and policies and that can contribute to increasing crop yield and to stabilizing atmospheric CO2 concentrations. NASA’s solar-induced fluorescence (SIF) satellites have revealed that crops in the US Corn Belt have the highest peak photosynthesis activity on the Earth. However, SIF-based and other existing satellite-derived GPP products are characterized by coarse spatial resolution (≥ 500 m) at which crop fields are mostly mixed. The lack of high-spatial-and-temporal-resolution crop GPP dataset has hampered global carbon cycle studies and agricultural applications. To boost the productivity and sustainability of the agro-ecosystem, it is essential to monitor crop GPP at field-level, regional-scale, and sub-week-frequency.

To fill this big gap, this project proposes to: 1) develop and evaluate a new algorithm for 5-day, 30 m resolution GPP estimation for corn and soybean in the US Corn Belt integrating absorbed photosynthetic active radiation (APAR, defined as the product of photosynthetic active radiation (PAR) and fraction of absorbed PAR (FPAR)) and canopy chlorophyll content (CCC, defined as the product of leaf chlorophyll content (LCC) and leaf area index (LAI)) retrieved from new-generation satellite data such as NASA’s Harmonized Landsat and Sentinel-2 (HLS) and the PI Kaiyu Guan’s Landsat-MOIDS fused STAIR surface reflectance products; 2) quantify variations of GPP for rainfed/irrigated corn and soybean, and investigate how they respond to climate variability and technical and managerial changes in space and time, and how they are linked with variations of crop yield. Specifically, I propose three tasks to address three scientific questions:

Question 1: Among radiative transfer model (RTM), machine learning (ML) and vegetation index (VI) approaches, which method performs the most robust in estimating CCC and FPAR from HLS and STAIR data, respectively?

Task 1 (Canopy variables retrieval): Compare and evaluate CCC and FPAR retrieval algorithms for corn and soybean from new-generation satellite data.

Question 2: Is the proposed crop GPP model integrating APAR and CCC a scalable solution for corn and soybean GPP estimation in the US Corn Belt?

Task 2 (GPP algorithm evaluation): Develop and evaluate the new algorithm for high spatiotemporal resolution GPP of corn and soybean in the US Corn Belt.

Question 3: How do field-level GPP for rainfed/irrigated corn and soybean vary across the US Corn Belt and over the past two decades?

Task 3 (GPP variations investigation): Investigate the spatial and temporal variations of crop GPP, its natural drivers, and its impacts on crop yield.

To implement and address the above tasks and questions, I will fully take the advantages of new-generation satellite data, ground measurements, radiative transfer models, machine learning models and vegetation index-based models, and cloud computing.

This study will support NASA Earth Science Research Program and NASEM 2017 Decadal Survey for Earth Observation from Space by expanding our knowledge on how and why crop productivity changes, and thoroughly utilizing NASA’s multi-satellite datasets. Furthermore, policymakers/farmers will be able to apply the proposed research to improve crop management, which is an essential element of NASA’s Applied Science Program for the benefit of society.