Measuring Bi-Directional Reflectance for Gross Primary Productivity with a Constellation of SmallSats

Abstract:

The "missing carbon" problem has plagued the carbon cycle field for over 30 years. A newly proposed constellation of satellites promises to finally close the gap and find the missing carbon. This constellation would measure vegetation from multiple angles at solar wavelengths, essentially measuring the bidirectional reflectance (BRDF), and from this retrieve the Gross Primary Productivity (GPP), something that has eluded space remote sensing community up until now, showing up to 40% uncertainty.

The science value of such a BRDF retrieval approach has been demonstrated using multi-angle, multi-spectral measurements from various deployments of the Cloud Absorption Radiometer (CAR) as the “gold standard” data for BRDF estimation. CAR is an airborne instrument operated by NASA Goddard Space Flight Center. Initial observing system simulations (OSSE) with four satellites launched as secondary payloads and operating in different imaging modes show BRDF error estimates of less than 12% when compared to CAR measurements, a 50% improvement to the worst case BRDF error produced by corresponding monoliths.

However, GPP products require estimating the BRDF of photochemical reflectance index (PRI), which needs angular measurements at the xanthophyll sensitive band (533nm) – unavailable in CAR. The satellite OSSEs will be repeated using AMPSEC tower measurements. AMPSEC is a Unispec-DC (PP Systems, Amesbury,MA, USA) spectroradiometer with 256 contiguous bands with a nominal band spacing of 3 nm and a nominal range of operation between 350 and 1200 nm. The data will be used to estimate parameters of the widely-used Rahman–Pinty–Verstraete (RPV) and RossThin-LiSparseReciprocal (RTnLS) BRDF models. Since AMPSEC reflectance data is obtained at 360 view-azimuth directions and 90 view-zenith directions, satellite clusters will be able to sample only a part of this angular space. To make best use of the satellite-cluster BRDF data, a heuristic optimization method is used to find the best angular sub-sampling. Also, different closed loop formation-flying geometries are considered. We will show the effect of these formation-flying architectures on BRDF, PRI and GPP estimation errors and identify an optimal baseline architecture that will reduce errors when compared to existing spaceborne instruments such as MODIS and MISR.