Title: Computational and open source enhancements to NASA LIS to enable the next generation digital twins
Presenting Author:  David Mocko / Sujay Kumar
Organization: Goddard Space Flight Center
Co-Author(s): James Geiger, Jules Kouatchou, Craig Pelissier

Abstract:
The NASA Land Information System (LIS; lis.gsfc.nasa.gov) is an open source software infrastructure designed to facilitate the efficient utilization of terrestrial hydrological observations. The significant investments from ESTO have enabled the development of major computational tools and capabilities in LIS including a large suite of land surface models, data assimilation algorithms, inverse modeling, radiative transfer models, and model verification. These capabilities have enabled LIS to be a foundational terrestrial hydrology modeling digital twin (DT), while enabling additional interoperable capabilities in the ecosystem of DT systems. Though LIS has been demonstrated for informing mission requirements and interpreting and utilizing NASA Earth science data, our current ability to maximize the information extraction from remote sensing measurements is limited by computational requirements. In particular, techniques such as data assimilation, inverse modeling, and uncertainty estimation rely on sets of multiple model runs called ensembles, which are computationally demanding to employ at global or even regional scales. This presentation will provide updates on the computational, I/O, enhancements in LIS that allow for overcoming these limitations. We will also present updates on the containerization of a large modeling system such as LIS, which has enhanced the accessibility of the system through cloud environments. This capability is also important for developing a viable pathway for enabling interoperable Earth system digital twins and the use of large data volumes expected from future missions such as NISAR, where local data access and reliance on on-premise computing resources may not be practical.