agentsclimarketplace

Glm lake calibration

Skill cxcscmu/SkillLearnBench/skills/b4-skill-creator-claude-opus-4-6/temperature-simulation/glm-lake-calibration

Calibrate the General Lake Model (GLM) for lake temperature simulation. Use this skill whenever calibrating GLM parameters, running GLM simulations, or tuning Kw, coef_mix_hyp, wind_factor, lw_factor, or ch to match observed water temperature profiles. Covers Lake Mendota and similar dimictic lakes.From its SKILL.md

Install
npx -y skills add cxcscmu/SkillLearnBench --skill glm-lake-calibration

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

2.8 KB, 673 tokens by cl100k_base, as published. Nobody here has run it

GLM Lake Calibration

Key Calibration Parameters and Their Effects

These five parameters most influence vertical temperature structure:

ParameterRangeEffect
Kw[0.1, 0.5]Light extinction coefficient. Higher values trap more heat near surface, warming epilimnion and cooling hypolimnion. Controls thermocline sharpness.
coef_mix_hyp[0.3, 0.7]Hypolimnetic mixing coefficient. Higher values increase deep mixing, warming the hypolimnion and reducing stratification strength.
wind_factor[0.7, 1.3]Scales wind speed input. Higher values deepen the mixed layer and increase surface cooling via latent/sensible heat.
lw_factor[0.7, 1.3]Scales incoming longwave radiation. Higher values warm the surface. Affects overall heat budget.
ch[0.0005, 0.002]Bulk transfer coefficient for sensible heat. Higher values increase sensible heat exchange, cooling the surface in summer.

Calibration Strategy for Lake Mendota

Lake Mendota is a large dimictic lake (43°N, max depth ~25m). Key calibration targets:

  1. Overall RMSE: Get surface and mid-depth temperatures right first via wind_factor and lw_factor
  2. Deep temperature (≥13m): Controlled primarily by Kw and coef_mix_hyp
  3. Summer deep: Most sensitive to coef_mix_hyp (hypolimnetic mixing during stratification)

Recommended Calibration Approach

  1. Start with reasonable defaults: Kw=0.30, coef_mix_hyp=0.5, wind_factor=1.0, lw_factor=1.0, ch=0.0013
  2. Adjust Kw first — affects light penetration and thermal structure
  3. Tune coef_mix_hyp — controls deep water warming during stratification
  4. Fine-tune wind_factor and lw_factor for overall bias
  5. Adjust ch last — fine-tunes surface heat exchange

Parameter Interactions

  • Increasing Kw + decreasing coef_mix_hyp = stronger stratification
  • wind_factor and ch both affect surface heat loss but through different mechanisms
  • lw_factor primarily shifts mean temperature of the whole lake

Running GLM

cd /root && mkdir -p output && glm --xdisp
# Or simply: glm (if no X display needed, use glm 2>&1)

GLM reads glm3.nml from the current directory. Output goes to output/output.nc.

Modifying Parameters in glm3.nml

Parameters are in Fortran namelist format. Key sections:

  • &light: contains Kw
  • &mixing: contains coef_mix_hyp
  • &meteorology: contains wind_factor, lw_factor, ch

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.