What problem does it solve? Running the COSIPY glacier/snowpack energy and mass balance model requires deep operational knowledge: exact input units (Kelvin, hPa, W/m2), netCDF formatting, TOML configuration, and recovery from silent failures like Celsius temperatures forcing all precipitation to rain. This Skill packages that expertise so an AI agent can prepare forcing data, execute the real COSIPY model, and validate results without manual trial and error. ## Core Features & Use Cases - Forcing Conversion: Convert ERA5, AWS, CSV, or NASA POWER meteorological data into COSIPY netCDF input with automatic unit conversions, bounds checking, and unit-trap warnings. - Model Execution with Preflight Checks: Run the actual COSIPY v2.0.2 model via run_cosipy.py, with a preflight_check.py that verifies the Python environment, source tree, and input data before execution. - Output Parsing and Validation: Extract mass balance, snow height, albedo, and energy fluxes from result netCDF files and compare against observations (SNOTEL SWE, snow depth, stake data) with NSE/KGE/r/PBIAS metrics. - Diagnostic Recovery: A triplets.yaml database maps 23 known failure symptoms (e.g., broken netCDF4 backend, thin-column IndexError) to validated remedies. - Use Case: Simulate seasonal snow depth at a point station using NASA POWER forcing, then validate the modeled SNOWHEIGHT against the Canadian Historical Daily Snow Depth Database. ## Quick Start Run preflight_check.py in this directory, then ask the agent to simulate glacier mass balance for Zhadang Glacier using the bundled ERA5 2009 test data and report the MB output.
