Retrieval (Model inversion)

Warning

Github of the project: https://github.com/Prikaziuk/retrieval_rtmo

Highlights

  1. Numerical optimization of RTMo (optical) module of SCOPE model

    • Top of canopy (TOC) reflectance

  2. Sensors

    • ground (ASD, FLOX)

    • airborne (HyPlant)

    • satellite (Sentinel-2 MSI, Sentinel-3 OLCI)

    • custom - just add yourself into input/sensors.xlsx

  3. Available options
    • Hyperspectral instruments:

      • data format - text

      • parallel computing (parfor)

      • time-series (angles, radiation)

      • calculation of observation geometry from coordinates and time

      • plotting of validation data (1 : 1 plots)

      • output - .xlsx or .csv

    • Multispectral instruments (Satellites):

      • data format - NetCDF4

      • parallel computing (parfor)

      • retrieval on an image subset (K x K pixels)

      • output - .xlsx + NetCDF4

Definition

Typically you provide vegetation parameters and SCOPE gives you expected reflectance spectra. For retrieval you provide the reflectance spectra and SCOPE gives you expected vegetation parameters, resulting in similar spectra.

That is why it is called model inversion. We somehow want to get the input (parameters) from the output (spectra).

There are two basic inversion approaches:
  • look-up table

    • spectra are produced beforehand in the desired ranges of values

    • for example you change chlorophyll content (Cab) from 0 to 100 ug cm-2 in steps of 5.

    • advantage fast

    • disadvantage coarse: if you tune 6 parameters giving only 10 steps for each you end up in a look-up table with 1.000.000 simulations

  • numerical optimization

    • spectra are calculated on the spot (for each measurement during the run)

    • algorithm is minimizing the cost-function (difference between modelled and measured data)

    • advantage sharp values

    • disadvantage slow, local minimum trap

Directory structure

demonstrator-master.zip
├── src
│   ├── +helpers
│   ├── +io
│   ├── +models
│   ├── +plot
│   ├── +sat
│   ├── +to_sensor
│   ├── +ts
│   ├── COST_4SAIL_common.m
│   ├── fit_spectra.m
│   ├── Input_data.xlsx
│   ├── main.m
│   ├── main_sat.m
│   └── propagate_uncertainty.m
│
├── input
│   ├── fluspect_data
│   ├── radiationdata
│   ├── soil_spectrum
│   ├── PC_flu.xlsx
│   └── sensors.xlsx
│
├── measured
│   ├── airborne
│   ├── canopy
│   ├── leaf
│   ├── synthetic
│   └── Sentinel-2_sample.nc
│
└── output
    ├── sample_S2
    └── sample_synthetic

Acknowledgement

The project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 721995.

_images/trustee_logo.png