Radiance-based NIR_v as a proxy for GPP of corn and soybean

Abstract

Substantial uncertainty exists in daily and sub-daily gross primary production (GPP) estimation, which dampens accurate monitoring of the global carbon cycle. Here we find that near-infrared radiance of vegetation (NIR v,Rad ), defined as the product of observed NIR radiance and normalized difference vegetation index, can accurately estimate corn and soybean GPP at daily and half-hourly time scales, benchmarked with multi-year tower-based GPP at three sites with different environmental and irrigation conditions. Overall, NIR v,Rad explains 84% and 78% variations of half-hourly GPP for corn and soybean, respectively, outperforming NIR reflectance of vegetation (NIR v,Ref ), enhanced vegetation index (EVI), and far-red solar-induced fluorescence (SIF 760 ). The strong linear relationship between NIR v,Rad and absorbed photosynthetically active radiation by green leaves (APAR green ), and that between APAR green and GPP, explain the good NIR v,Rad -GPP relationship. The NIR v,Rad -GPP relationship is robust and consistent across sites. The scalability and simplicity of NIR v,Rad indicate a great potential to estimate daily or sub-daily GPP from high-resolution and/or long-term satellite remote sensing data.

NIRvP: a robust structural proxy for sun-induced chlorophyll fluorescence and photosynthesis across scales

“…S8b) cannot be easily explained by invoking SIF retrieval noise and therefore other potential reasons for these differences should be investigated. Our results on the direct comparison of SIF and NIRVP for GPP estimation confirm and considerably extend previous findings from site-level studies 22,40,41 . Among other things, we confirmed that previous site-level results showing better GPP estimation performance of NIRVP compared to SIF in crops 22,40 also held in a Mediterranean grassland site which experiences drought (Fig.…”

“…S1a). Regarding the correlations of SIF and NIRVP to GPP, two previous studies 22,40 already reported higher or similar correlation of NIRVP for the rice, wheat, soybean and the two corn datasets we used. In addition, we found that NIRVP outperformed SIF also for the grassland site (R 2 = 0.60 vs. 0.49 for half-hourly data; Fig.…”

“…Differences between NIRVP and NIRV regarding their correlation to SIF and GPP are expected to become more important when using data from upcoming geostationary satellite missions such as GeoCarb, TEMPO and Sentinel-4 2 . As canopy-level PAR cannot be directly observed from airplanes or satellites, either a simple approach that approximates PAR via the downwelling NIR radiance 22,37,40 , or more complex methods involving atmospheric radiative transfer modelling 38,55 or machine learning 56 can be used. The radiance-based approach has previously been shown to have comparable performance with direct PAR observations at the site level 22 .…”

“…NIRVP was estimated from FLUO at-sensor radiance data to ensure minimal differences compared to SIF retrievals in terms of sensor and processing aspects. The approach of using NIR radiance as proxy for the product of NIR reflectance times PAR was previously introduced as NIRVR 22 and was found to show good performance in terms of correlation to SIF and GPP at the site level 22,40,41 . Those results included observations in cloudy conditions where the largest differences between PAR and NIR radiance are expected, the discrepancies between the two variables should be even smaller in the clear sky conditions under which the airborne campaign was conducted.…”

“…However, direct comparisons of SIF and NIRVP for GPP estimation using observations from the same sensors (or at least from the same viewing geometries) have been limited to a small set of crop sites 22,40,41 .…”