Publications

Welch, H; Brodie, S; Jacox, MG; Robinson, D; Wilson, C; Bograd, SJ; Oliver, MJ; Hazen, EL (2020). Considerations for transferring an operational dynamic ocean management tool between ocean color products. REMOTE SENSING OF ENVIRONMENT, 242, 111753.

Abstract
Satellite remote sensing data are critical for assessing ecosystem state and evaluating long-term trends and shifts in ecosystem components. Many operational tools rely on continuous streams of remote sensing data, and when a satellite sensor reaches the end of its designed lifespan, these tools must be transferred to a more reliable data stream. Transferring between data streams can produce discontinuities in tool products, and it is important to quantify these downstream impacts and understand the mechanisms that cause discontinuity. To illustrate the complexities of tool transfer, we compare five products for ocean chlorophyll-a, which is a proxy for phytoplankton biomass and an important input for tools that monitor marine biophysical processes. The five chlorophyll-a products included three blended products and two single sensor products from MODIS and VIIRS. We explored the downstream impacts of tool transfer using EcoCast: an operational dynamic ocean management tool that combines real-time predictions from target and bycatch species distribution models to produce integrated surfaces of fishing suitability. EcoCast was operationalized using MODIS chlorophyll-a, and we quantify the impacts of transferring to the intended replacement of MODIS, VIIRS, and test if impacts can be minimized by using a blended chlorophyll-a product instead. Differences between chlorophyll products did not linearly propagate through to the species model predictions and the integrated fishing suitability surfaces. Instead, differences in species model predictions were determined by the shape of chlorophyll-a response curves in the species models relative to chlorophyll-a differences between sensors. However, differences in the integrated fishing suitability surfaces were reduced by canceling of differences from individual species model predictions. Differences in the integrated fishing suitability surfaces were not reduced by transferring to a blended product, highlighting the complexity of transferring operational tools between different remote sensing data products. These results contribute to our general understanding of the mechanisms by which transferring between data streams impacts downstream products. To aid decision-making regarding tool transfer, we developed an interactive web application that allows end-users to explore differences in chlorophyll products within times period and regions of interest.

DOI:
10.1016/j.rse.2020.111753

ISSN:
0034-4257