There is growing recognition within the subseasonal-to-seasonal (S2S) climate forecast community that the land surface state (i.e., soil moisture, soil temperature, snow, and vegetation) provides a significant source of predictability a shorter time scales. Potentially the largest source regions for land-forced climate variability are elevated regions, including the Tibetan Plateau and Rocky Mountain Cordillera. In these locations, the land is providing thermal and dynamic forcing in the middle troposphere (i.e., 500-hPa) that has been shown to enforce and/or modulate a Rossby wavenumber-5 train. The recent international Impact of Initialized Land Temperature and Snowpack on Sub-seasonal to Seasonal Prediction Experiment (LS4P)-Phase 1 revealed that May surface heating anomalies over the Tibetan Plateau explain more June precipitation variability over the western U.S. than ocean SST anomalies. The Tibetan Plateau–North American teleconnection is strongest during April–May, before the North Pacific jet has shifted poleward, and before Pacific and Atlantic subtropical highs emerge. What this means is: in order to fully harvest the potential 2–4 week lead predictability embedded in the Tibetan Plateau–North American teleconnection, Earth system models must accurately represent spring–summer shoulder season land–atmosphere interactions, including snow cover/snowmelt dynamics.

We propose a comprehensive, E3SMv2 modeling and model evaluation project to fully diagnose Tibetan Plateau and Rocky Mountain surface heating anomalies as sources of climate predictability. Through our investigation, we will learn: 1) how dominant central U.S. precipitating event types are affected by remote Tibetan Plateau surface heating anomalies, 2) how local Rocky Mountain and remote Tibetan Plateau surface heating forcings interact and effect central U.S. precipitation, and 3) how accurately E3SMv2 represents the constituent multi-scale land–atmosphere interactions. We will conduct land-only ELMv2 simulations to isolate terrestrial process representation errors, AMIP simulations to benchmark coupled land–atmosphere interactions, and idealized AMIP simulations to isolate ENSO, PDO, and land surface heating as sources of climate variability. With the benefit of new insights and new diagnostic tools (i.e., DOE Coordinated Model Evaluation Capabilities) from these preliminary activities, we will investigate S2S reforecasts of a low-level jet–mesoscale convective system event, a low-level jet–atmospheric river event, and a tropical cyclone predecessor event.  Furthermore, we will contribute E3SMv2 simulations to LS4P Phase 2. Our project will critically advance understanding of sources of U.S. summer precipitation predictability and serve to identify systematic process-level errors and biases in E3SMv2’s representation of spring–summer land surface and land–atmosphere interactions in an S2S forecast context. The proposed research directly addresses the DOE Earth and Environmental Systems Modeling Regional Global Model Analysis (RGMA) Water Cycle Area goal to demonstrate how biases in the representation of the water cycle affect Earth system predictability, while concurrently addressing broad RGMA goals of: 1) advancing understanding of the Earth system, 2) improving models, and 3) reducing uncertainties that exist in current Earth system model projections.