Note

Go to the end to download the full example code.

Two-dimensional NMR processing

Temporarily excluded from the public gallery.

This source is preserved for the future 2D characterization campaign, but the current public NMR workflow intentionally supports only validated 1D processing. On the merged public branch, scp.nmr.Experiment.process() raises NotImplementedError for multi-dimensional datasets, so this example is not published while 2D support remains under separate scientific review.

This example complements sphx_glr_plugins_spectrochempy-nmr_examples_processing_nmr_plot_processing_nmr.py:

  • plot_processing_nmr.py starts from a TopSpin 2rr dataset that is already processed.

  • This example starts from the raw TopSpin ser file and performs the main processing steps inside SpectroChemPy.

The pipeline is intentionally split in two stages:

  1. F2 (direct dimension): apodization -> FFT -> auto-phase

  2. F1 (indirect dimension): zero-filling -> apodization -> FFT

It illustrates:

  • reading raw 2D SER data;

  • automatic handling of Bruker encodings such as STATES-TPPI;

  • quaternion-aware 2D FFT processing;

  • how to obtain a spectrum comparable to the processed 2rr reference.

Requires the official spectrochempy-nmr plugin. Install with: pip install spectrochempy[nmr].

Import API

import spectrochempy as scp

Read raw 2D SER data

The raw SER file contains time-domain data together with the acquisition metadata needed to decode the hypercomplex encoding in F1.

nmrdir = scp.preferences.datadir / "nmrdata" / "bruker" / "tests" / "nmr"
ser = scp.nmr.read(nmrdir / "topspin_2d", expno=1, remove_digital_filter=True)

/home/runner/work/spectrochempy/spectrochempy/plugins/spectrochempy-nmr/src/spectrochempy_nmr/extern/nmrglue/_bruker.py:441: UserWarning: (196608,)cannot be shaped into(147, 1024)
  _, data = read_binary(f, shape=shape, cplex=cplex, big=big, isfloat=isfloat)

Print dataset summary — both dimensions are in time domain

ser
NDDataset [topspin_2d expno:1 procno:1 (SER)] — quaternion, shape: (y:96, x:948), count
name
:
topspin_2d expno:1 procno:1 (SER)
author
:
runner@runnervmvrwv9
created
:
2026-08-03 18:28:09+00:00
history
:
2026-08-03 18:28:09+00:00> Imported from TopSpin dataset
Data
title
:
intensity
values
:
RR[[ -0.2238 -0.1985 ... -0.1662 0.03262]
[-0.006566 0.0282 ... -0.02949 -0.06717]
...
[ 0 0 ... 0 0]
[ 0 0 ... 0 0]] count
RI[[ 0.06219 0.1467 ... 0.04565 0.03068]
[-0.05969 -0.08752 ... -0.05134 -0.05994]
...
[ -0 -0 ... -0 -0]
[ -0 -0 ... -0 -0]] count
IR[[ -0.1623 -0.0563 ... 0.02654 -0.01094]
[ 0.1344 -0.006515 ... -0.08239 0.00516]
...
[ 0 0 ... 0 0]
[ 0 0 ... 0 0]] count
II[[-0.003312 -0.001535 ... 0.02067 -0.08058]
[-0.05685 0.1174 ... 0.05831 -0.003414]
...
[ -0 -0 ... -0 -0]
[ -0 -0 ... -0 -0]] count
shape
:
(y:96(complex), x:948(complex))
Dimension `x`
size
:
948
title
:
F2 acquisition time
coordinates
:
[ 0 48 ... 4.541e+04 4.546e+04] µs
Dimension `y`
size
:
96
title
:
F1 acquisition time
coordinates
:
[ 0 74 ... 6956 7030] µs


Plot the raw time-domain data (quaternion representation)

_ = ser.plot_map()
plot processing nmr 2d

For comparison, load the TopSpin-processed reference spectrum. This is similar to the dataset used in plot_processing_nmr.py.

reference = scp.nmr.read(nmrdir / "topspin_2d" / "1" / "pdata" / "1" / "2rr")

Plot the processed TopSpin reference

_ = reference.plot_map()
plot processing nmr 2d

Stage 1: process F2 (direct dimension)

Experiment.process() applies the usual direct-dimension workflow: exponential multiplication, FFT, then automatic phasing along F2.

Here we also set the F2 zero-filled size to match the processed TopSpin spectrum more closely.

from spectrochempy_nmr import Experiment

exp = Experiment(ser)
f2_processed = exp.process(apodization="em", lb=2.0, size=2048)

# After F2 processing: F2 is in frequency domain, F1 is still time domain
f2_processed

Traceback (most recent call last):
  File "/home/runner/work/spectrochempy/spectrochempy/build/~gallery_examples/processing/nmr/plot_processing_nmr_2d.py", line 86, in <module>
    f2_processed = exp.process(apodization="em", lb=2.0, size=2048)
  File "/home/runner/work/spectrochempy/spectrochempy/plugins/spectrochempy-nmr/src/spectrochempy_nmr/experiment.py", line 545, in process
    raise NotImplementedError(msg)
NotImplementedError: Public NMR processing currently supports only validated 1D experiments. Multi-dimensional NMR processing remains out of public scope pending further scientific characterization.

Plot after F2 processing — only the horizontal axis is transformed

_ = f2_processed.plot_map()

Stage 2: process F1 (indirect dimension)

F1 still needs the usual indirect-dimension steps. We keep them explicit here because they are often tuned case by case in practice.

The values below are chosen so that the resulting spectrum is close to the processed TopSpin 2rr reference shipped with the test data.

f1 = f2_processed.zf_size(size=1024, dim="y")
f1 = f1.em(lb=5.0, dim="y")
f1 = f1.fft(dim="y")

f1

Plot the fully processed 2D spectrum

_ = f1.plot_map()

Extract and inspect a 1D slice

Once the 2D spectrum is processed we can reuse the same style of operations shown in plot_processing_nmr.py.

slice_f2 = f1[-27.6, :]
_ = slice_f2.plot()

Compare the same slice against the TopSpin 2rr reference. We normalize both traces before overlaying them because the absolute intensity depends on the exact processing chain (window functions, phasing, scaling conventions, etc.).

reference_slice_f2 = reference[-27.6, :]
slice_f2_norm = slice_f2.normalize(method="max", dim="x")
reference_slice_f2_norm = reference_slice_f2.normalize(method="max", dim="x")
_ = slice_f2_norm.plot(color="black")
_ = reference_slice_f2_norm.plot(clear=False, color="red", linestyle="--")

Total running time of the script: (0 minutes 3.228 seconds)