def write_segy_based_on(hf, segy_y_reader, data, endian='>'):
"""
Purpose:
make it easy to write new (processed) data back to a new segy file based on the original
segy file
Args:
fh: A file-like object open for binary write.
segy_y_reader: a SegYReader instance from which the segy headers and other parameters can be retreived
data: a numpy array containing trace samples organised in (trace, inline, crossline order)
"""
encoding = segy_y_reader.encoding
if not is_supported_encoding(encoding):
raise UnsupportedEncodingError("Writing SEG Y", encoding)
write_textual_reel_header(fh, segy_y_reader.textual_reel_header, encoding)
write_binary_reel_header(fh, segy_y_reader.binary_reel_header, endian)
write_extended_textual_headers(fh, segy_y_reader.extended_textual_header, encoding)
trace_header_packer = HeaderPacker(trace_header_format, endian)
il_xl = list(itertools.product(segy_y_reader.inline_range(), segy_y_reader.xline_range()))
x_y = list(itertools.product(range(0,segy_y_reader.num_inlines()), range(0,segy_y_reader.num_xlines())))
for ix,ij in zip(il_xl, x_y):
idx = reader.trace_index((ix[0],ix[1]))
write_trace_header(fh, segy_y_reader.trace_header(idx), trace_header_packer)
write_trace_samples(fh, data[:,ij[0],ij[1]], segy_y_reader.data_sample_format, endian=endian)
def write_segy(fh,
dataset,
encoding=None,
trace_header_format=TraceHeaderRev1,
endian='>',
progress=None):
"""
Args:
fh: A file-like object open for binary write, positioned to write the textual reel header.
dataset: An object implementing the interface of segpy.dataset.Dataset, such as a SegYReader.
trace_header_format: The class which defines the layout of the trace header. Defaults to TraceHeaderRev1.
encoding: Optional encoding for text data. Typically 'cp037' for EBCDIC or 'ascii' for ASCII. If omitted, the
seg_y_data object will be queries for an encoding property.
endian: Big endian by default. If omitted, the seg_y_data object will be queried for an encoding property.
progress: A unary callable which will be passed a number
between zero and one indicating the progress made. If
provided, this callback will be invoked at least once with
an argument equal to one.
Raises:
UnsupportedEncodingError: If the specified encoding is neither ASCII nor EBCDIC
UnicodeError: If textual data provided cannot be encoded into the required encoding.
"""
progress_callback = progress if progress is not None else lambda p: None
if not callable(progress_callback):
raise TypeError("write_segy(): progress callback must be callable")
encoding = encoding or (hasattr(dataset, 'encoding')
and dataset.encoding) or ASCII
if not is_supported_encoding(encoding):
raise UnsupportedEncodingError("Writing SEG Y", encoding)
write_textual_reel_header(fh, dataset.textual_reel_header, encoding)
write_binary_reel_header(fh, dataset.binary_reel_header, endian)
write_extended_textual_headers(fh, dataset.extended_textual_header,
encoding)
trace_header_packer = make_header_packer(trace_header_format, endian)
num_traces = dataset.num_traces()
for trace_index in dataset.trace_indexes():
write_trace_header(fh, dataset.trace_header(trace_index),
trace_header_packer)
write_trace_samples(fh,
dataset.trace_samples(trace_index),
dataset.data_sample_format,
endian=endian)
progress_callback(trace_index / num_traces)
progress_callback(1)
def write_segy(fh,
seg_y_data,
encoding=None,
trace_header_format=TraceHeaderRev1,
endian='>',
progress=None):
"""
Args:
fh: A file-like object open for binary write, positioned to write the textual reel header.
seg_y_data: An object from which the headers and trace_samples data can be retrieved. Requires the following
properties and methods:
seg_y_data.textual_reel_header
seg_y_data.binary_reel_header
seg_y_data.extended_textual_header
seg_y_data.trace_indexes
seg_y_data.trace_header(trace_index)
seg_y_data.trace_samples(trace_index)
seg_y_data.encoding
seg_y_data.endian
One such legitimate object would be a SegYReader instance.
trace_header_format: The class which defines the layout of the trace header. Defaults to TraceHeaderRev1.
encoding: Optional encoding for text data. Typically 'cp037' for EBCDIC or 'ascii' for ASCII. If omitted, the
seg_y_data object will be queries for an encoding property.
endian: Big endian by default. If omitted, the seg_y_data object will be queried for an encoding property.
progress: An optional progress bar object.
Raises:
UnsupportedEncodingError: If the specified encoding is neither ASCII nor EBCDIC
UnicodeError: If textual data provided cannot be encoded into the required encoding.
"""
encoding = encoding or (hasattr(seg_y_data, 'encoding')
and seg_y_data.encoding) or ASCII
if not is_supported_encoding(encoding):
raise UnsupportedEncodingError("Writing SEG Y", encoding)
write_textual_reel_header(fh, seg_y_data.textual_reel_header, encoding)
write_binary_reel_header(fh, seg_y_data.binary_reel_header, endian)
write_extended_textual_headers(fh, seg_y_data.extended_textual_header,
encoding)
trace_header_packer = make_header_packer(trace_header_format, endian)
for trace_index in seg_y_data.trace_indexes():
write_trace_header(fh, seg_y_data.trace_header(trace_index),
trace_header_packer)
write_trace_samples(fh,
seg_y_data.trace_samples(trace_index),
seg_y_data.data_sample_format,
endian=endian)
def test_roundtrip(self, write_header, endian):
with BytesIO() as write_stream:
write_binary_reel_header(write_stream, write_header, endian)
written_stream = write_stream.getvalue()
with BytesIO(written_stream) as read_stream:
read_header = read_binary_reel_header(read_stream, endian)
assert are_equal(write_header, read_header)
def test_roundtrip(self, write_header, endian):
with BytesIO() as write_stream:
write_binary_reel_header(write_stream, write_header, endian)
written_stream = write_stream.getvalue()
with BytesIO(written_stream) as read_stream:
read_header = read_binary_reel_header(read_stream, endian)
self.assertTrue(are_equal(write_header, read_header))
def write_segy(fh,
seg_y_data,
encoding=None,
trace_header_format=TraceHeaderRev1,
endian='>',
progress=None):
"""
Args:
fh: A file-like object open for binary write, positioned to write the textual reel header.
seg_y_data: An object from which the headers and trace_samples data can be retrieved. Requires the following
properties and methods:
seg_y_data.textual_reel_header
seg_y_data.binary_reel_header
seg_y_data.extended_textual_header
seg_y_data.trace_indexes
seg_y_data.trace_header(trace_index)
seg_y_data.trace_samples(trace_index)
seg_y_data.encoding
seg_y_data.endian
seg_y_data.data_sample_format
One such legitimate object would be a SegYReader instance.
trace_header_format: The class which defines the layout of the trace header. Defaults to TraceHeaderRev1.
encoding: Optional encoding for text data. Typically 'cp037' for EBCDIC or 'ascii' for ASCII. If omitted, the
seg_y_data object will be queries for an encoding property.
endian: Big endian by default. If omitted, the seg_y_data object will be queried for an encoding property.
progress: An optional progress bar object.
Raises:
UnsupportedEncodingError: If the specified encoding is neither ASCII nor EBCDIC
UnicodeError: If textual data provided cannot be encoded into the required encoding.
"""
encoding = encoding or (hasattr(seg_y_data, 'encoding') and seg_y_data.encoding) or ASCII
if not is_supported_encoding(encoding):
raise UnsupportedEncodingError("Writing SEG Y", encoding)
write_textual_reel_header(fh, seg_y_data.textual_reel_header, encoding)
write_binary_reel_header(fh, seg_y_data.binary_reel_header, endian)
write_extended_textual_headers(fh, seg_y_data.extended_textual_header, encoding)
trace_header_packer = make_header_packer(trace_header_format, endian)
for trace_index in seg_y_data.trace_indexes():
write_trace_header(fh, seg_y_data.trace_header(trace_index), trace_header_packer)
write_trace_samples(fh, seg_y_data.trace_samples(trace_index), seg_y_data.data_sample_format, endian=endian)
def write_segy(fh,
dataset,
encoding=None,
trace_header_format=TraceHeaderRev1,
endian='>',
progress=None):
"""
Args:
fh: A file-like object open for binary write, positioned to write the textual reel header.
dataset: An object implementing the interface of segpy.dataset.Dataset, such as a SegYReader.
trace_header_format: The class which defines the layout of the trace header. Defaults to TraceHeaderRev1.
encoding: Optional encoding for text data. Typically 'cp037' for EBCDIC or 'ascii' for ASCII. If omitted, the
seg_y_data object will be queries for an encoding property.
endian: Big endian by default. If omitted, the dataset object will be queried for an encoding property.
progress: A unary callable which will be passed a number
between zero and one indicating the progress made. If
provided, this callback will be invoked at least once with
an argument equal to one.
Raises:
UnsupportedEncodingError: If the specified encoding is neither ASCII nor EBCDIC
UnicodeError: If textual data provided cannot be encoded into the required encoding.
"""
progress_callback = progress if progress is not None else lambda p: None
if not callable(progress_callback):
raise TypeError("write_segy(): progress callback must be callable")
encoding = encoding or (hasattr(dataset, 'encoding') and dataset.encoding) or ASCII
if not is_supported_encoding(encoding):
raise UnsupportedEncodingError("Writing SEG Y", encoding)
write_textual_reel_header(fh, dataset.textual_reel_header, encoding)
write_binary_reel_header(fh, dataset.binary_reel_header, endian)
write_extended_textual_headers(fh, dataset.extended_textual_header, encoding)
trace_header_packer = make_header_packer(trace_header_format, endian)
num_traces = dataset.num_traces()
for trace_index in dataset.trace_indexes():
write_trace_header(fh, dataset.trace_header(trace_index), trace_header_packer)
write_trace_samples(fh, dataset.trace_samples(trace_index), dataset.data_sample_format, endian=endian)
progress_callback(trace_index / num_traces)
progress_callback(1)
def write_segy(fh,
dataset,
encoding=None,
trace_header_format=TraceHeaderRev1,
endian='>',
progress=None):
"""
Args:
fh: A file-like object open for binary write, positioned to write the textual reel header.
dataset: An object implementing the interface of segpy.dataset.Dataset, such as a SegYReader.
trace_header_format: The class which defines the layout of the trace header. Defaults to TraceHeaderRev1.
encoding: Optional encoding for text data. Typically 'cp037' for EBCDIC or 'ascii' for ASCII. If omitted, the
seg_y_data object will be queries for an encoding property.
endian: Big endian by default. If omitted, the seg_y_data object will be queried for an encoding property.
progress: An optional progress bar object.
Raises:
UnsupportedEncodingError: If the specified encoding is neither ASCII nor EBCDIC
UnicodeError: If textual data provided cannot be encoded into the required encoding.
"""
encoding = encoding or (hasattr(dataset, 'encoding') and dataset.encoding) or ASCII
if not is_supported_encoding(encoding):
raise UnsupportedEncodingError("Writing SEG Y", encoding)
write_textual_reel_header(fh, dataset.textual_reel_header, encoding)
write_binary_reel_header(fh, dataset.binary_reel_header, endian)
write_extended_textual_headers(fh, dataset.extended_textual_header, encoding)
trace_header_packer = make_header_packer(trace_header_format, endian)
for trace_index in dataset.trace_indexes():
write_trace_header(fh, dataset.trace_header(trace_index), trace_header_packer)
write_trace_samples(fh, dataset.trace_samples(trace_index), dataset.data_sample_format, endian=endian)
binary_reel_header.sample_interval = int(
args.sample_rate * 1000) # samples length microseconds
binary_reel_header.num_samples = len(itimes) # number of samples
binary_reel_header.data_traces_per_ensemble = 0 # len(data) # also must be # not sure how to work this out, maybe I need to scan the file first or after insert it
binary_reel_header.auxiliary_traces_per_ensemble = 0
binary_reel_header.trace_sorting = 4 # http://oivdoc91.vsg3d.com/APIS/RefManCpp/struct_so_v_r_segy_file_header.html#a612dab3b4d9c671ba6554a8fb4a88057
binary_reel_header.format_revision_num = 256 # 0 or 1 TODO move to 1
binary_reel_header.data_sample_format = args.dtype # see binary header def.py file as it changes by revision. 1 is always ibm float
if args.use_extended_header:
binary_reel_header.num_extended_textual_headers = len(
extended_textual_header
) # see binary header def.py file as it changes by revision. 1 is always ibm float
# Write headers
toolkit.write_textual_reel_header(fo, textual_reel_header, segyencoding)
toolkit.write_binary_reel_header(fo, binary_reel_header, endian)
if args.use_extended_header:
toolkit.write_extended_textual_headers(fo, extended_textual_header,
segyencoding)
# Pre-Format trace headerf
trace_header_format = toolkit.make_header_packer(
segpy.trace_header.TraceHeaderRev1, endian)
if args.seismic_dimensions == 3:
# either iterate over the grid for 3d
xxlines, iinlines = np.meshgrid(xlines, inlines)
trace_iter = np.vstack([iinlines.flat, xxlines.flat]).T
else:
# or for 2d just iterate over cdp an sp
trace_iter = np.vstack([inlines_r, xlines_r]).T
def write_binary_reel_header(self, binary_reel_header):
self._fh.seek(REEL_HEADER_NUM_BYTES)
write_binary_reel_header(self._fh, binary_reel_header, self.endian)
def write_binary_reel_header(self,binary_reel_header):
self._fh.seek(REEL_HEADER_NUM_BYTES)
write_binary_reel_header(self._fh, binary_reel_header, self.endian)
def write_segy_file(data: pd.DataFrame, outfile: str):
metadata = data.groupby(data.itrace).first() # one row for each trace
# lets work out the whole inline xline grid
inlines_r = pd.Series(data.inline.unique())
xlines_r = pd.Series(data.xline.unique())
# choose interval as most common inline step, should I use min but ignore zero and nan?
inline_int = inlines_r.diff().mode().values
xline_int = xlines_r.diff().mode().values
inlines = np.arange(inlines_r.min(), inlines_r.max(), inline_int)
xlines = np.arange(xlines_r.min(), xlines_r.max(), xline_int)
# find inline and xline with most data
xline_lens = metadata.xline.groupby(metadata.xline).count()
longest_xline_n = xline_lens.argmax()
longest_xline = metadata[metadata.xline == longest_xline_n]
inline_lens = metadata.inline.groupby(metadata.inline).count()
longest_inline_n = inline_lens.argmax()
longest_inline = metadata[metadata.inline == longest_inline_n]
if SEISMIC_DIMENSIONS == 3:
# work out bins spacing
ibins = pylab.distances_along_curve(
longest_xline[['x', 'y']]) / longest_xline.inline.diff().iloc[1:]
ibin = ibins.mean()
xbins = pylab.distances_along_curve(
longest_inline[['x', 'y']]) / longest_inline.xline.diff().iloc[1:]
xbin = xbins.mean()
inline_angs = angles(longest_inline[['x', 'y']].values)
inline_rot = inline_angs.mean()
pi = metadata.iloc[
0] # reference point, might not be origin as we don't have that yet
pj = metadata.iloc[100] # use this for a test
gd = Grid(pi, inline_rot, ibin, xbin)
x, y = gd.ix2xy(pj.inline, pj.xline)
# check the predicted coords are right withing a meter
np.testing.assert_allclose(pj.x, x, atol=1)
np.testing.assert_allclose(pj.y, y, atol=1)
logger.info("Writing segy {}".format(OUTFILE))
tmin = data.t.min()
# times to interpolate onto
itimes = np.arange(tmin, data.t.max() + SAMPLE_RATE, SAMPLE_RATE)
template_trace = segpy.trace_header.TraceHeaderRev1()
template_binary_reel_header = segpy.binary_reel_header.BinaryReelHeader()
header = []
textual_reel_header = get_header(header)
metadata = data.groupby(data.itrace).first()
# Format text header
if USE_EXTENDED_HEADER:
logger.debug("Formating extended textual header")
extended_textual_header = toolkit.format_extended_textual_header(
'', DEFAULT_SEGY_ENCODING)
else:
extended_textual_header = toolkit.format_extended_textual_header(
''.join(header), DEFAULT_SEGY_ENCODING)
# Format binary header
binary_reel_header = template_binary_reel_header
# samples length microseconds
binary_reel_header.sample_interval = int(SAMPLE_RATE * 1000)
# number of samples
binary_reel_header.num_samples = len(itimes)
# len(data) # also must be # not sure how to work this out,
# maybe I need to scan the file first or after insert it
binary_reel_header.data_traces_per_ensemble = 0
binary_reel_header.auxiliary_traces_per_ensemble = 0
# http://oivdoc91.vsg3d.com/APIS/RefManCpp/struct_so_v_r_segy_file_header.html#a612dab3b4d9c671ba6554a8fb4a88057
binary_reel_header.trace_sorting = 4
# 0 or 1 TODO move to 1
binary_reel_header.format_revision_num = 256
# see binary header def.py file as it changes by revision. 1 is always ibm float
binary_reel_header.data_sample_format = DTYPE
if USE_EXTENDED_HEADER:
# see binary header def.py file as it changes by revision. 1 is always ibm float
binary_reel_header.num_extended_textual_headers = len(
extended_textual_header)
# Pre-Format trace headerf
trace_header_format = toolkit.make_header_packer(
segpy.trace_header.TraceHeaderRev1, ENDIAN)
if SEISMIC_DIMENSIONS == 3:
# either iterate over the grid for 3d
xxlines, iinlines = np.meshgrid(xlines, inlines)
trace_iter = np.vstack([iinlines.flat, xxlines.flat]).T
else:
# or for 2d just iterate over cdp an sp
trace_iter = np.vstack([inlines_r, xlines_r]).T
i = 0
tracen = 1
with open(outfile, 'wb') as fo:
# Write headers
toolkit.write_textual_reel_header(fo, textual_reel_header,
DEFAULT_SEGY_ENCODING)
toolkit.write_binary_reel_header(fo, binary_reel_header, ENDIAN)
if USE_EXTENDED_HEADER:
toolkit.write_extended_textual_headers(fo, extended_textual_header,
DEFAULT_SEGY_ENCODING)
for inline, xline in trace_iter:
i += 1
if ((metadata.inline == inline) * (metadata.xline == xline)):
trace = data[(data.inline == inline) * (data.xline == xline)]
metatrace = metadata[(metadata.inline == inline) *
(metadata.xline == xline)]
x = metatrace.x
y = metatrace.y
times = trace.t.values
vels = trace.v.values
elif SEISMIC_DIMENSIONS == 3:
x, y = gd.ix2xy(inline, xline)
times = itimes
vels = np.zeros(itimes.shape)
else:
logger.warning("inline/xline or cdp/sp not found", inline,
xline)
continue
cdp = inline
sp = xline
if i % 1000 == 0:
logger.debug("Writing trace: "
"{: 8.0f}/{}, {: 6.2f}%".format(
i, len(trace_iter),
(1.0 * i / len(trace_iter)) * 100))
if len(vels) == 0:
logger.error("Error no vels on trace", i)
continue
# interpolate data
samples = np.interp(itimes, times, vels)
# ensure datatype is ok
samples = np.require(samples, dtype='d')
# Format trace header
trace_header = template_trace
trace_header.line_sequence_num = 1000 + tracen
trace_header.field_record_num = tracen
trace_header.trace_num = tracen
if SEISMIC_DIMENSIONS == 3:
trace_header.file_sequence_num = inline
trace_header.ensemble_num = xline
trace_header.inline_number = inline
trace_header.crossline_number = xline
else:
trace_header.file_sequence_num = 1000 + tracen
trace_header.ensemble_num = cdp
trace_header.shotpoint_number = sp
trace_header.num_samples = len(samples) # number of samples
trace_header.sample_interval = SAMPLE_RATE # sample interval
trace_header.cdp_x = x
trace_header.cdp_y = y
trace_header.source_x = x
trace_header.source_y = y
# write trace header and data
toolkit.write_trace_header(fo,
trace_header,
trace_header_format,
pos=None)
toolkit.write_trace_samples(fo,
samples=samples,
ctype=SEG_Y_TYPE,
endian=ENDIAN,
pos=None)
tracen += 1
logger.debug("Writing trace: {: 8.0f}/{}, {: 6.2f}%".format(
i, len(trace_iter), (1.0 * i / len(trace_iter)) * 100))
