Tutorial: MicaSense → Landsat Harmonization

This tutorial demonstrates how to harmonize drone-scale multispectral reflectance (e.g., MicaSense RedEdge) into Landsat-equivalent band values using the cross-sensor-cal regression workflow.

---

Overview

You will learn how to:

1. prepare MicaSense reflectance inputs
2. apply band mapping and wavelength matching
3. use regression tables to harmonize reflectance
4. inspect harmonized Landsat-style products

This allows direct comparison between drone and satellite observations.

---

1. Inputs

You need reflectance values from a calibrated drone multispectral system. These can be:

• stacked reflectance TIFFs
• ENVI-formatted band images
• Parquet tables exported from your workflow

Each band should have known center wavelengths.

---

2. Harmonization workflow

cross-sensor-cal uses regression relationships linking MicaSense band values to Landsat OLI bands. These regressions are derived from calibrated field and NEON comparisons.

Run the following command:

```bash cscal-micasense-to-landsat \ --input your_micasense_input \ --output ms_to_ls_output \ --regression-table data/regression/micasense_to_landsat.csv Outputs include: Landsat-equivalent reflectance table diagnostic statistics optional ENVI export 3. Inspecting harmonized reflectance Example: import pandas as pd

df = pd.read_parquet("ms_to_ls_output/micasense_landsat_harmonized.parquet") df.head() Columns will match Landsat bands: Blue Green Red NIR SWIR1 SWIR2 4. NDVI sanity check df["ndvi"] = (df["NIR"] - df["Red"]) / (df["NIR"] + df["Red"]) df["ndvi"].describe() If harmonization worked correctly, NDVI should fall within typical vegetation ranges (0.1–0.9 for most cases). 5. Next steps Integrate drone → NEON → Landsat comparisons Combine harmonized products with NEON-derived spectra Use the merged output in ecological modeling workflows See also: NEON → Landsat tutorial Pipeline outputs

---