During geomagnetic storms and substorms, a vast amount of magnetospheric energy is deposited into the ionosphere-thermosphere (I-T) system via the convergence of Poynting flux and energetic particle precipitation. This energy deposition leads to significant heating at high latitudes, changes in global circulation, and disturbs neutral and plasma densities. The changes in the I-T system have a significant impact on the near-Earth space environment such as satellite drag, orbiting, and communication. Largescale storm-time responses have been well studied using I-T models driven by statistical maps of aurora and electric potential, while mesoscale and small-scale processes and cross-scale interaction were much less understood. This is partially attributed to the lack of sufficient multi-scale ionospheric observations that can be used to constrain models. To improve the accuracy of space weather modeling for the I-T system, we need continuous observations of high-latitude forcing (e.g., particle precipitation and electric fields) with sufficient resolution that can separate temporal and spatial structures/variabilities, as well as simultaneous observations of I-T responses such as neutral winds, neutral/plasma temperatures and densities. Such multi-parameter and multi-scale observations will allow one to monitor different scales of forcing and consequences at the same time so as to understand the global, multi-scale dynamics of the I-T system and its responses to external driving conditions, as well as constrain I-T models to improve space weather predictability. This white paper addresses the current challenges for a more realistic space weather simulation and solicits observational and modeling requirements.