Generating Novel Climate Scenarios with the STITCHES Python Package
STITCHES enhances ESM emulation by creating new scenario realizations efficiently, leveraging CMIP6 outputs to generate gridded, multivariate output data for novel scenarios for multisectoral dynamics studies.
Image courtesy of Eberhard Grossgasteiger on Pexels
Understanding the coupled human-Earth system is complex. Scientists use Earth System Models (ESMs) to predict future climate scenarios, but these models are expensive and time-consuming to run. This work documents the open-source, FAIR-principles compliant python package, STITCHES, a tool that quickly creates new climate scenarios using existing model data. It combines pieces of past data to form new predictions, saving time and resources. This helps scientists explore more scenarios without needing to run full models each time, making multisector dynamic research more efficient and accessible.
Our research introduces the Python package STITCHES, a FAIR, open-source tool that helps scientists understand how the interactions occur in the human-Earth system. STITCHES quickly creates new climate scenarios by leveraging existing data, saving time and resources. It allows researchers to explore new possibilities without running expensive models.
We present STITCHES, a Python package designed to enhance the flexibility and efficiency of Earth System Model (ESM) emulation. STITCHES addresses the computational challenges of modeling human-Earth system interactions by enabling the creation of new scenario realizations without the extensive computational costs associated with full ESM runs. By leveraging existing ESM outputs from the CMIP6/ScenarioMIP archive, STITCHES recombines these outputs into new, multivariate, spatially resolved scenarios based on target global surface air temperature trajectories. This top-down approach preserves the statistical characteristics of the original ESM outputs and allows for the joint emulation of most variables outputs by ESM, a feature that distinguishes it from other emulation methods.
STITCHES integrates seamlessly with the Pangeo Cloud catalog, allowing users to access and emulate variables from ESMs participating in ScenarioMIP without pre-downloading data, which has historically been a huge workload in itself. The package's outputs are structured identically to CMIP6 data, making them compatible with many existing impact models. By enabling the generation of new realizations of novel scenarios in minutes, STITCHES offers a practical tool for multisector dynamic and impact modelers to explore novel scenarios and enrich existing datasets, potentially doubling or tripling the number of available runs with which to drive their models. This capability enhances the scope and depth of scenario analysis, providing a valuable resource for researchers.