We propose to develop, implement, and maintain a cloud-based, open-source software system capable of producing the parameters needed by NASA for the official discharge product for every river in the…

The University of Massachusetts (UMass) was awarded a $779,966.00 Grant for Research contract from NASA's Shared Services Center to develop, implement, and maintain a cloud-based, open-source software system called "Confluence" that will produce the parameters needed by NASA for the official discharge product for every river in the SWOT River Database. The contract, which has no set-aside designation, will fund the deployment of five mass conserved flow law inversion (MCFLI) discharge algorithms developed by the previous SWOT Science Team at the global scale, as well as the addition of necessary diagnostics, consensus algorithms, and evaluation modules. The Confluence system is intended to serve as the computational backbone of SWOT discharge for NASA and the entire SWOT Science Team. UMass will partner with a SWOT Science Team proposal led by Michael Durand of The Ohio State University to develop select Confluence modules and lead discharge evaluation activities.

We propose to develop, implement, and maintain a cloud-based, open-source software system capable of producing the parameters needed by NASA for the official discharge product for every river in the swot river database (sword). We name this system confluence, as IT represents a merging of a decade of previous theoretical work, swot data, science team (ST) products, and an ST community approach to estimating discharge. Confluence will deploy five mass conserved flow law inversion (mcfli) discharge algorithms developed by the previous ST at the global scale. We will add numerous and necessary novel diagnostics pre- and post- mcfli, combine mcfli algorithms into a consensus product, include evaluation modules, and provide discharge parameters to the ST and to NASA. We anticipate additional inputs of prior data from other ST members that will improve our discharge estimates and thus the discharge product. Our discharge output will also be readily available to the ST as inputs for other projects seeking to merge swot discharge with models. In SUM, we are seeking to provide the computational backbone of swot discharge for JPL and the entire ST, consistent with the call for exactly such proposals. Creating our proposed system is a substantial undertaking, and in order to fully develop and distribute any proposed data products using proposed resources, we believe that developing, testing, and distributing a global discharge product will occupy the entirety of the pre-launch period and require a substantial percentage of total proposed resources, leaving little time pre-launch for the science of discharge. This is grounded in our experience developing a working prototype of confluence over the past calendar year. We will thus partner with an ST proposal submitted by michael durand of the ohio state university, who (if funded) will develop a select few novel confluence modules and lead discharge evaluation pre- and post-launch, tasks beyond available resources for this proposal. Also following our prototype experience, we will hire a dedicated software engineer for this pre-launch period to professionally develop the software on time and on budget. The software engineer will also optimize existing mcfli algorithms, hopefully leading to large gains in computing speed. Post-launch (and post-software development), we are motivated by a driving science question: what is the best way to combine prior databases and swot observations to estimate global river discharge? The computer engineering personnel herein are experts in machine learning and big data for infrastructure and environmental systems. Confluence will have access to all global swot data, a rich database of prior information from the ST, and numerous discharge estimates from different physical inversion platforms: an unprecedented volume of data. Thus, our computer engineering expertise will be brought to bear to assess whether the wealth of global swot data collated in confluence could more effectively estimate discharge using machine learning techniques rather than with mcfli as envisioned by hydrologists during the past decade of swot theory work by the ST. We ultimately propose to be the final stop for all information needed to produce discharge at the global scale. We hope to provide a platform to amalgamate all ST discharge knowledge and insight into an efficient, open source, and comprehensive software platform. Having this single software system will prevent redundancies, improve efficiency, and ensure the best discharge product possible is delivered on time to NASA and the public. Our open source and containerized software can run on any operating system, cluster, or cloud, and thus individual ST members, NASA, and the public can pursue their own discharge goals on their own terms while we, at the same time, deliver on mission discharge requirements at the global scale.