Improving a snow energy and mass balance model with observations of albedo decay and recovery following Colorado’s 2020 Cameron Peak Wildfire
Data files
Jun 09, 2026 version files 40.49 MB
-
caldwell_dryad_repository.zip
40.49 MB
-
README.md
4.50 KB
Abstract
Wildfires have grown in both intensity and extent, with recent events increasingly affecting seasonal snow zones that accumulate deep, persistent snowpacks. Post-fire impacts—such as forest canopy loss and deposition of light-absorbing impurities (e.g., charred debris and soot)—increase net shortwave radiation to the snowpack by diminishing canopy shading and reducing snow albedo, resulting in accelerated snowmelt and earlier snow disappearance. While previous work has shown that snow albedo can gradually recover, the rate and magnitude of recovery remain poorly constrained. This study analyzes observations of albedo decay and recovery following Colorado’s 2020 Cameron Peak Fire, the largest wildfire in state history. Station-measured albedo indicates that median ablation-season albedo increased by ~40 % during the study period (0.38 in WY 2022 to 0.53 in WY 2025). Event-scale analysis indicates faster and more variable albedo decay following snowfall at burned sites, with median decay rates up to ~2–3 times more negative than unburned sites. These observations inform updates to post-fire albedo decay parameterization in iSnobal, a snow energy and mass balance model. Compared to the base configuration, implementing exponential albedo decay improved agreement with observed albedo distributions by 50–63 % during water years 2021–2023 and shifted modeled snow disappearance 8–14 days closer to observations. iSnobal simulations demonstrate that canopy loss exerted a stronger control on net shortwave energy absorption than post-fire albedo changes. Accounting for post-wildfire albedo changes in snow models is critical for improving predictions of snowmelt timing and water yield in fire-affected regions.
Dataset DOI: 10.5061/dryad.zs7h44jr9
Description of the data and file structure
Snow Albedo, SWE, and Melt Modeling Dataset (Burned and Unburned Forest Sites, Colorado)
Overview:
This dataset contains modeled and observed snow water equivalent (SWE), snow albedo, and melt metrics derived from multiple snow modeling approaches and field observations across burned and unburned forested sites in northern Colorado, USA. The dataset supports analysis of post-fire snow hydrologic response, including changes in melt timing, energy balance, and seasonal snow evolution.
Data span water years 2021–2025 and include multiple sites representing different topographic and disturbance conditions.
Study Sites:
- Burned Flat (BF)
- Burned North (BN)
- Burned South (BS)
- Unburned Flat (UF)
- Unburned North (UN)
- Joe Wright / Snow Telemetry (JW / SNOTEL)
Files and variables
Snow Water Equivalent (SWE) from three sources:
- AWS observations (field instruments)
- Base iSnobal model output
- Exponential decay model output
- Combined and Landfire (LF) model outputs
Units: millimeters (mm SWE)
Notes:
- AWS SWE originally in cm and converted to mm
- Model SWE derived from snow thickness and density where explicit SWE was not provided:
- SWE = thickness × snow_density
Snow Albedo:
- Model outputs
- AWS station measurements
Units: dimensionless (0–1)
Processing:
- Values > 1 removed as invalid
- Summer periods masked out in model outputs
- AWS albedo filtered to snow-on periods only
- Corrected for terrain effects from slope and aspect
- Median daily albedo derived from median hourly AWS albedo between 11 am and 2 pm MST. Also applied to model outputs for hourly albedo.
- Decay rate analysis: exponential regression equation fit to median daily albedo following new storm event (>3 cm) for at least 3 days without a new storm reset
Melt Rate Metrics:
Daily melt rates were derived from SWE time series as:
- Positive daily change in SWE loss:
- - melt = -ΔSWE
Two melt windows are provided:
- Peak SWE → Snow Disappearance Date (SDD)
- Sustained melt period (manually defined seasonal window)
Units: mm day -1
Water year (WY) = October 1 – September 30
Model Output Files:
- baseRun_output/
- final_exponential/
- final_combined/
- lf/
AWS Observations:
- AWS_daily_MST.csv
- AWS_hourly_MST.csv
Contains:
- SWE
- Snow presence flag (snow on)
- Albedo (Median_Albedo_Corrected)
Code/software
Matlab (license required)
- applyMedianWindowAlbedo.m – Creates daily albedo using median hourly albedo between 11 am and 2 pm MST for observations.
- plotDecayLoopAllYrs.m – Main script for exponential regression fit to albedo decay events
- plotModelSWnetOutput.m – Creates cumulative sum for net SW energy from model outputs and observations.
- sweMeltRate.m – Daily melt rates for modeled and observed data.
- decayRateControls.m – Plots days relative to peak SWE and prior albedo versus albedo decay rates.
- plotModelOutput.m – Plots modeled SWE output from each iSnobal model run
- albedoRecovery.m – Creates boxplot of ablation season albedo values
- medianAlbedoSnobal.m – Creates daily albedo using median hourly albedo between 11 am and 2 pm MST for model outputs.
- plotModelAlbedoOutput.m – Plots iSnobal modeled albedo and creates cumulative distribution function plots with K-S statistics.
- slopeCorrectionAlbedo.m – Corrects hourly albedo for slope and aspect effects, requires SolarAzEl.m in the present working directory.
- ModelSDD_allSites.mat – Supporting file for snow disappearance dates in plotModelOutput.m
- SolarAzEl.m – Supporting function for calculating solar positions used in slopeCorrectionAlbedo.m
Python (Pycharm interpreter used for this project)
- Requires arcpy, os, rioxarray, xarray libraries.
- Scripts:
- makeMask.py: Creates a binary burn mask raster using MTBS burn severity product.
- make_topo.nc: Creates a new topo.nc file with burn mask raster from above script for input in iSnobal.
Access information
Other publicly accessible locations of the data:
Data was derived from the following sources: