def test_check_dataid(): """test check dataid.""" datin = [Data(), Data()] dat = dataprep.check_dataid(datin) assert dat[0].dataid == '(1)' assert dat[1].dataid == '(2)'
def test_fuzzy(): """ test fuzzy cluster """ dat1 = Data() dat1.data = np.ma.identity(3) dat1.data.mask = np.zeros([3, 3]) dat1.data.data[0, 0] = 1.1 dat2 = Data() dat2.data = np.ma.ones([3, 3]) dat2.data.mask = np.zeros([3, 3]) tmp = fuzzy_clust.FuzzyClust(None) tmp.indata = {'Raster': [dat1, dat2]} tmp.spinbox_minclusters.setValue(2) tmp.spinbox_maxclusters.setValue(2) tmp.combobox_alg.setCurrentIndex(0) tmp.settings(True) datout2 = tmp.outdata['Cluster'][0].data.data datout = np.array([[1, 2, 2], [2, 1, 2], [2, 2, 1]]) if datout2[0, 0] == 2: datout = np.abs(datout - 3) np.testing.assert_array_equal(datout2, datout)
def gdal_to_dat(dest, bandid='Data'): """ GDAL to Data format Parameters ---------- dest - GDAL format GDAL format bandid - str band identity """ dat = Data() gtr = dest.GetGeoTransform() rtmp = dest.GetRasterBand(1) dat.data = rtmp.ReadAsArray() nval = rtmp.GetNoDataValue() dat.data = np.ma.masked_equal(dat.data, nval) dat.data.set_fill_value(nval) dat.data = np.ma.fix_invalid(dat.data) dat.nrofbands = dest.RasterCount dat.tlx = gtr[0] dat.tly = gtr[3] dat.dataid = bandid dat.nullvalue = nval dat.rows = dest.RasterYSize dat.cols = dest.RasterXSize dat.xdim = abs(gtr[1]) dat.ydim = abs(gtr[5]) dat.wkt = dest.GetProjection() dat.gtr = gtr return dat
def test_normalisation(): """Tests for normalisation.""" datin = Data() datin.data = np.ma.array([[1., 2.], [1., 2.]]) tmp = normalisation.Normalisation(None) tmp.indata = {'Raster': [datin, datin]} tmp.radiobutton_interval.setChecked(True) tmp.settings(True) datout = np.ma.array([[0., 1.], [0., 1.]]) np.testing.assert_array_equal(tmp.outdata['Raster'][0].data, datout) tmp.radiobutton_mean.setChecked(True) tmp.settings(True) datout = np.ma.array([[-1., 1.], [-1., 1.]]) np.testing.assert_array_equal(tmp.outdata['Raster'][0].data, datout) tmp.radiobutton_median.setChecked(True) tmp.settings(True) datout = np.ma.array([[-1., 1.], [-1., 1.]]) np.testing.assert_array_equal(tmp.outdata['Raster'][0].data, datout) tmp.radiobutton_8bit.setChecked(True) tmp.settings(True) datout = np.ma.array([[0., 255.], [0., 255.]]) np.testing.assert_array_equal(tmp.outdata['Raster'][0].data, datout)
def smalldata(): """Small test dataset.""" dat = Data() dat.data = np.ma.array([[29000., 29000.], [29000., 29000.]], mask=[[0, 0], [0, 0]]) dat.extent = (25, 25, -28, -27) # left, right, bottom, top return dat
def test_rtp(): """test rtp.""" datin = Data() datin.data = np.ma.array([[1, 2], [1, 2]]) dat2 = [[0.8763800720931049, 2.123619927906895], [0.8763800720931049, 2.123619927906895]] dat = dataprep.rtp(datin, 60, 30) np.testing.assert_array_equal(dat.data, dat2)
def run(self): """ Entrypoint to start this routine """ bandid = [] self.proj.set_current(self.indata['Raster'][0].wkt) for i in self.indata['Raster']: bandid.append(i.dataid) self.banddata[i.dataid] = Data() tmp = self.banddata[i.dataid] self.dataid[i.dataid] = i.dataid tmp.tlx = i.tlx tmp.tly = i.tly tmp.xdim = i.xdim tmp.ydim = i.ydim tmp.rows = i.rows tmp.cols = i.cols tmp.nullvalue = i.nullvalue tmp.wkt = i.wkt tmp.min = i.data.min() tmp.max = i.data.max() tmp.mean = i.data.mean() try: tmp.units = i.units except AttributeError: tmp.units = '' self.combobox_bandid.currentIndexChanged.disconnect() self.combobox_bandid.addItems(bandid) indx = self.combobox_bandid.currentIndex() self.oldtxt = self.combobox_bandid.itemText(indx) self.combobox_bandid.currentIndexChanged.connect(self.update_vals) idata = self.banddata[self.oldtxt] self.lbl_cols.setText(str(idata.cols)) self.lbl_rows.setText(str(idata.rows)) self.txt_null.setText(str(idata.nullvalue)) self.dsb_tlx.setText(str(idata.tlx)) self.dsb_tly.setText(str(idata.tly)) self.dsb_xdim.setText(str(idata.xdim)) self.dsb_ydim.setText(str(idata.ydim)) self.lbl_min.setText(str(idata.min)) self.lbl_max.setText(str(idata.max)) self.lbl_mean.setText(str(idata.mean)) self.led_units.setText(str(idata.units)) self.update_vals() tmp = self.exec_() if tmp == 1: self.acceptall() tmp = True return tmp
def test_equation(): """tests equation editor.""" datin = Data() datin.data = np.ma.array([[1., 2.], [1., 2.]]) datout = datin.data * 2 tmp = equation_editor.EquationEditor() tmp.indata = {'Raster': [datin, datin]} tmp.settings('i0+i1') np.testing.assert_array_equal(tmp.outdata['Raster'][0].data, datout)
def test_trimraster(): """test trim raster.""" datin = Data() datin.data = np.ma.masked_equal( [[0, 0, 0, 0], [0, 1, 2, 0], [0, 1, 2, 0], [0, 0, 0, 0]], 0) datin.nullvalue = 0 dat2 = [[1, 2], [1, 2]] dat = dataprep.trim_raster([datin]) np.testing.assert_array_equal(dat[0].data, dat2)
def init_grid(self, data): """ Initializes raster variables in the Data class Args: data (numpy masked array): masked array containing raster data.""" grid = Data() grid.data = data grid.xdim = self.dxy grid.ydim = self.dxy grid.extent = [self.xrange[0], self.xrange[1], self.yrange[0], self.yrange[1]] return grid
def test_crisp(): """ test crisp cluster """ dat1 = Data() dat1.data = np.ma.identity(3) dat1.data.mask = np.zeros([3, 3]) dat2 = Data() dat2.data = np.ma.ones([3, 3]) dat2.data.mask = np.zeros([3, 3]) tmp = crisp_clust.CrispClust(None) tmp.indata = {'Raster': [dat1, dat2]} tmp.spinbox_minclusters.setValue(2) tmp.spinbox_maxclusters.setValue(2) tmp.settings(True) datout2 = tmp.outdata['Cluster'][0].data.data datout = np.array([[1, 2, 2], [2, 1, 2], [2, 2, 1]]) if datout2[0, 0] == 2: datout = np.abs(datout - 3) np.testing.assert_array_equal(datout2, datout)
def run(self): """Entrypoint to start this routine.""" bandid = [] self.proj.set_current(self.indata['Raster'][0].wkt) for i in self.indata['Raster']: bandid.append(i.dataid) self.banddata[i.dataid] = Data() tmp = self.banddata[i.dataid] self.dataid[i.dataid] = i.dataid tmp.xdim = i.xdim tmp.ydim = i.ydim tmp.nullvalue = i.nullvalue tmp.wkt = i.wkt tmp.extent = i.extent tmp.data = i.data tmp.units = i.units self.combobox_bandid.currentIndexChanged.disconnect() self.combobox_bandid.addItems(bandid) indx = self.combobox_bandid.currentIndex() self.oldtxt = self.combobox_bandid.itemText(indx) self.combobox_bandid.currentIndexChanged.connect(self.update_vals) idata = self.banddata[self.oldtxt] irows = idata.data.shape[0] icols = idata.data.shape[1] self.lbl_cols.setText(str(icols)) self.lbl_rows.setText(str(irows)) self.txt_null.setText(str(idata.nullvalue)) self.dsb_tlx.setText(str(idata.extent[0])) self.dsb_tly.setText(str(idata.extent[-1])) self.dsb_xdim.setText(str(idata.xdim)) self.dsb_ydim.setText(str(idata.ydim)) self.lbl_min.setText(str(idata.data.min())) self.lbl_max.setText(str(idata.data.max())) self.lbl_mean.setText(str(idata.data.mean())) self.led_units.setText(str(idata.units)) self.update_vals() tmp = self.exec_() if tmp == 1: self.acceptall() tmp = True return tmp
def init_grid(self, data): """ Initializes raster variables in the Data class Args: data (numpy masked array): masked array containing raster data.""" grid = Data() grid.data = data grid.cols = self.numx grid.rows = self.numy grid.xdim = self.dxy grid.ydim = self.dxy grid.tlx = self.xrange[0] grid.tly = self.yrange[1] return grid
def test_IGRF(smalldata): """Tests IGRF Calculation.""" datin2 = Data() datin2.data = np.ma.array([[0., 0.], [0., 0.]], mask=[[0, 0], [0, 0]]) datin2.extent = (25, 25, -28, -27) # left, right, bottom, top dat2 = [[940.640983, 864.497698], [1164.106631, 1079.494023]] tmp = igrf.IGRF() tmp.indata = {'Raster': [smalldata, datin2]} tmp.dateedit.setDate(QtCore.QDate(2000, 1, 1)) tmp.dsb_alt.setValue(0.) tmp.settings(True) dat = tmp.outdata['Raster'][-1].data np.testing.assert_array_almost_equal(dat, dat2)
def rtp(data, I_deg, D_deg): """ Reduction to the Pole """ datamedian = np.ma.median(data.data) ndat = data.data - datamedian ndat.data[ndat.mask] = 0 fftmod = np.fft.fft2(ndat) ny, nx = fftmod.shape nyqx = 1 / (2 * data.xdim) nyqy = 1 / (2 * data.ydim) kx = np.linspace(-nyqx, nyqx, nx) ky = np.linspace(-nyqy, nyqy, ny) KX, KY = np.meshgrid(kx, ky) I = np.deg2rad(I_deg) D = np.deg2rad(D_deg) alpha = np.arctan2(KY, KX) filt = 1 / (np.sin(I) + 1j * np.cos(I) * np.cos(D - alpha))**2 zrtp = np.fft.ifft2(fftmod * filt) zrtp = zrtp.real + datamedian # Create dataset dat = Data() dat.data = np.ma.masked_invalid(zrtp) dat.data.mask = np.ma.getmaskarray(data.data) dat.rows, dat.cols = zrtp.shape dat.nullvalue = data.data.fill_value dat.dataid = data.dataid dat.tlx = data.tlx dat.tly = data.tly dat.xdim = data.xdim dat.ydim = data.ydim return dat
def acceptall(self): """Accept.""" dxy = self.dsb_dxy.value() nullvalue = self.dsb_null.value() data = self.indata['Point'][0] newdat = [] for data in self.pbar.iter(self.indata['Point']): if data.dataid != self.dataid.currentText(): continue filt = (data.zdata != nullvalue) x = data.xdata[filt] y = data.ydata[filt] z = data.zdata[filt] for i in [x, y, z]: filt = np.logical_not(np.isnan(i)) x = x[filt] y = y[filt] z = z[filt] tmp = quickgrid(x, y, z, dxy, showtext=self.parent.showprocesslog) mask = np.ma.getmaskarray(tmp) gdat = tmp.data # Create dataset dat = Data() dat.data = np.ma.masked_invalid(gdat[::-1]) dat.data.mask = mask[::-1] dat.nullvalue = nullvalue dat.dataid = data.dataid dat.xdim = dxy dat.ydim = dxy dat.extent = [x.min(), x.max(), y.min(), y.max()] newdat.append(dat) self.outdata['Raster'] = newdat self.outdata['Point'] = self.indata['Point']
def gdal_to_dat(dest, bandid='Data'): """ GDAL to Data format. Parameters ---------- dest - GDAL format GDAL format bandid - str band identity """ dat = Data() gtr = dest.GetGeoTransform() nbands = dest.RasterCount if nbands == 1: rtmp = dest.GetRasterBand(1) dat.data = rtmp.ReadAsArray() else: dat.data = [] for i in range(nbands): rtmp = dest.GetRasterBand(i + 1) dat.data.append(rtmp.ReadAsArray()) dat.data = np.array(dat.data) dat.data = np.moveaxis(dat.data, 0, -1) nval = rtmp.GetNoDataValue() dat.data = np.ma.masked_equal(dat.data, nval) dat.data.set_fill_value(nval) dat.data = np.ma.fix_invalid(dat.data) dat.extent_from_gtr(gtr) dat.dataid = bandid dat.nullvalue = nval dat.wkt = dest.GetProjection() return dat
def init_grid(self, data): """ Initializes raster variables in the Data class. Parameters ---------- data : numpy array Masked array containing raster data. Returns ------- grid : PyGMI Data PyGMI raster dataset. """ grid = Data() grid.data = data grid.xdim = self.dxy grid.ydim = self.dxy grid.extent = [ self.xrange[0], self.xrange[1], self.yrange[0], self.yrange[1] ] return grid
def test_tilt(): """test tilt depth.""" datin = Data() datin.data = np.ma.array([[0, 0, .1, .5, 1], [0, .1, .5, 1, .5], [.1, .5, 1, .5, .1], [.5, 1, .5, .1, 0], [1, .5, .1, 0, 0]]) tmp = tiltdepth.TiltDepth(None) tmp.indata = {'Raster': [datin]} tmp.dsb_dec.setValue(0.) tmp.dsb_inc.setValue(90.) tmp.settings(True) tmp.change_band1() datout2 = tmp.depths datout = np.array([[3.93612464, -1.99438548, 1., 0.32962923], [3.49438548, -2.49438548, 1., 0.34958333], [2.99438548, -2.99438548, 1., 0.34958333], [1.98888969, -1.98888969, 2., 0.36451351], [2.48888969, -1.48759916, 2., 0.3654272]]) np.testing.assert_array_almost_equal(datout2, datout)
def tiltdepth(self, data): """ Calculate tilt depth """ self.pbar.setValue(0) self.pbar.setMaximum(4) # RTP inc = self.dsb_inc.value() dec = self.dsb_dec.value() zout = dataprep.rtp(data, inc, dec) # Tilt self.pbar.setValue(1) nr, nc = zout.data.shape dy, dx = np.gradient(zout.data) dxtot = np.sqrt(dx**2 + dy**2) dz = cooper.vertical(zout.data) t1 = np.arctan(dz / dxtot) self.pbar.setValue(2) # A negative number implies we are straddling 0 # Contour tilt x = zout.extent[0] + np.arange(nc) * zout.xdim + zout.xdim / 2 y = zout.extent[-1] - np.arange(nr) * zout.ydim - zout.ydim / 2 X, Y = np.meshgrid(x, y) Z = np.rad2deg(t1) self.X = X self.Y = Y self.Z = Z cnt0 = self.axes.contour(X, Y, Z, [0]) cnt45 = self.axes.contour(X, Y, Z, [45], alpha=0) cntm45 = self.axes.contour(X, Y, Z, [-45], alpha=0) self.pbar.setValue(3) gx0, gy0, cgrad0, cntid0 = vgrad(cnt0) gx45, gy45, _, _ = vgrad(cnt45) gxm45, gym45, _, _ = vgrad(cntm45) g0 = np.transpose([gx0, gy0]) self.pbar.setValue(4) dmin1 = [] dmin2 = [] for i, j in self.pbar.iter(g0): dmin1.append(distpc(gx45, gy45, i, j, 0)) dmin2.append(distpc(gxm45, gym45, i, j, 0)) dx1 = gx45[dmin1] - gx0 dy1 = gy45[dmin1] - gy0 dx2 = gxm45[dmin2] - gx0 dy2 = gym45[dmin2] - gy0 grad = np.arctan2(dy1, dx1) * 180 / pi grad[grad > 90] -= 180 grad[grad < -90] += 180 gtmp1 = np.abs(90 - np.abs(grad - cgrad0)) grad = np.arctan2(dy2, dx2) * 180 / pi grad[grad > 90] -= 180 grad[grad < -90] += 180 gtmp2 = np.abs(90 - np.abs(grad - cgrad0)) gtmp = np.logical_and(gtmp1 <= 10, gtmp2 <= 10) gx0 = gx0[gtmp] gy0 = gy0[gtmp] cntid0 = cntid0[gtmp] dx1 = dx1[gtmp] dy1 = dy1[gtmp] dx2 = dx2[gtmp] dy2 = dy2[gtmp] dist1 = np.sqrt(dx1**2 + dy1**2) dist2 = np.sqrt(dx2**2 + dy2**2) dist = np.min([dist1, dist2], 0) self.x0 = gx0 self.x1 = dx1 + gx0 self.x2 = dx2 + gx0 self.y0 = gy0 self.y1 = dy1 + gy0 self.y2 = dy2 + gy0 self.depths = np.transpose([gx0, gy0, cntid0.astype(int), dist]) tmp = dataprep.quickgrid(gx0, gy0, dist, data.xdim, showtext=self.showtext) mask = np.ma.getmaskarray(tmp) gdat = tmp.data dat = Data() dat.data = np.ma.masked_invalid(gdat[::-1]) dat.data.mask = mask[::-1] dat.nullvalue = dat.data.fill_value dat.dataid = data.dataid dat.xdim = data.xdim dat.ydim = data.xdim dat.extent = [gx0.min(), gx0.max(), gy0.min(), gy0.max()] self.outdata['Raster'] = [dat]
def get_aster_ged_bin(ifile): """ Get ASTER GED binary format Emissivity_Mean_Description: Mean Emissivity for each pixel on grid-box using all ASTER data from 2000-2010 Emissivity_SDev_Description: Emissivity Standard Deviation for each pixel on grid-box using all ASTER data from 2000-2010 Temperature_Mean_Description: Mean Temperature (K) for each pixel on grid-box using all ASTER data from 2000-2010 Temperature_SDev_Description: Temperature Standard Deviation for each pixel on grid-box using all ASTER data from 2000-2010 NDVI_Mean_Description: Mean NDVI for each pixel on grid-box using all ASTER data from 2000-2010 NDVI_SDev_Description: NDVI Standard Deviation for each pixel on grid-box using all ASTER data from 2000-2010 Land_Water_Map_LWmap_Description: Land Water Map using ASTER visible bands Observations_NumObs_Description: Number of values used in computing mean and standard deviation for each pixel. Geolocation_Latitude_Description: Latitude Geolocation_Longitude_Description: Longitude ASTER_GDEM_ASTGDEM_Description: ASTER GDEM resampled to NAALSED Parameters ---------- ifile : str filename to import Returns ------- dat : PyGMI raster Data dataset imported """ dat = [] nval = -9999 bandid = {} bandid[0] = 'Emissivity_mean_band_10' bandid[1] = 'Emissivity_mean_band_11' bandid[2] = 'Emissivity_mean_band_12' bandid[3] = 'Emissivity_mean_band_13' bandid[4] = 'Emissivity_mean_band_14' bandid[5] = 'Emissivity_std_dev_band_10' bandid[6] = 'Emissivity_std_dev_band_11' bandid[7] = 'Emissivity_std_dev_band_12' bandid[8] = 'Emissivity_std_dev_band_13' bandid[9] = 'Emissivity_std_dev_band_14' bandid[10] = 'Temperature_mean' bandid[11] = 'Temperature_std_dev' bandid[12] = 'NDVI_mean' bandid[13] = 'NDVI_std_dev' bandid[14] = 'Land_water_map' bandid[15] = 'Observations' bandid[16] = 'Latitude' bandid[17] = 'Longitude' bandid[18] = 'ASTER GDEM' scale = [ 0.001, 0.001, 0.001, 0.001, 0.001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.01, 0.01, 0.01, 0.01, 1, 1, 0.001, 0.001, 1 ] units = [ '', '', '', '', '', '', '', '', '', '', 'Kelvin', 'Kelvin', '', '', '', 'Number per pixel', 'degrees', 'degrees', 'meters' ] data = np.fromfile(ifile, dtype=np.int32) rows_cols = int((data.size / 19)**0.5) data.shape = (19, rows_cols, rows_cols) lats = data[16] * scale[16] lons = data[17] * scale[17] latsdim = (lats.max() - lats.min()) / (lats.shape[0] - 1) lonsdim = (lons.max() - lons.min()) / (lons.shape[0] - 1) tlx = lons.min() - abs(lonsdim / 2) tly = lats.max() + abs(latsdim / 2) for i in range(19): dat.append(Data()) dat[i].data = data[i] * scale[i] dat[i].dataid = bandid[i] dat[i].nullvalue = nval * scale[i] dat[i].xdim = lonsdim dat[i].ydim = latsdim dat[i].units = units[i] rows, cols = dat[i].data.shape xmin = tlx ymax = tly ymin = ymax - rows * dat[i].ydim xmax = xmin + cols * dat[i].xdim dat[i].extent = [xmin, xmax, ymin, ymax] dat.pop(17) dat.pop(16) return dat
def get_aster_ged(ifile): """ Gets ASTER GED data Parameters ---------- ifile : str filename to import Returns ------- dat : PyGMI raster Data dataset imported """ dat = [] ifile = ifile[:] dataset = gdal.Open(ifile, gdal.GA_ReadOnly) subdata = dataset.GetSubDatasets() latentry = [i for i in subdata if 'Latitude' in i[1]] subdata.pop(subdata.index(latentry[0])) dataset = None dataset = gdal.Open(latentry[0][0], gdal.GA_ReadOnly) rtmp = dataset.GetRasterBand(1) lats = rtmp.ReadAsArray() latsdim = (lats.max() - lats.min()) / (lats.shape[0] - 1) lonentry = [i for i in subdata if 'Longitude' in i[1]] subdata.pop(subdata.index(lonentry[0])) dataset = None dataset = gdal.Open(lonentry[0][0], gdal.GA_ReadOnly) rtmp = dataset.GetRasterBand(1) lons = rtmp.ReadAsArray() lonsdim = (lons.max() - lons.min()) / (lons.shape[0] - 1) tlx = lons.min() - abs(lonsdim / 2) tly = lats.max() + abs(latsdim / 2) i = -1 for ifile2, bandid2 in subdata: dataset = None dataset = gdal.Open(ifile2, gdal.GA_ReadOnly) bandid = bandid2 units = '' if 'ASTER_GDEM' in bandid2: bandid = 'ASTER GDEM' units = 'meters' if 'Land_Water_Map' in bandid2: bandid = 'Land_water_map' if 'Observations' in bandid2: bandid = 'Observations' units = 'number per pixel' rtmp2 = dataset.ReadAsArray() if rtmp2.shape[-1] == min(rtmp2.shape) and rtmp2.ndim == 3: rtmp2 = np.transpose(rtmp2, (2, 0, 1)) nbands = 1 if rtmp2.ndim == 3: nbands = rtmp2.shape[0] for i2 in range(nbands): nval = -9999 i += 1 dat.append(Data()) if rtmp2.ndim == 3: dat[i].data = rtmp2[i2] else: dat[i].data = rtmp2 dat[i].data = np.ma.masked_invalid(dat[i].data) dat[i].data.mask = (np.ma.getmaskarray(dat[i].data) | (dat[i].data == nval)) if dat[i].data.mask.size == 1: dat[i].data.mask = (np.ma.make_mask_none(dat[i].data.shape) + np.ma.getmaskarray(dat[i].data)) dat[i].data = dat[i].data * 1.0 if 'Emissivity/Mean' in bandid2: bandid = 'Emissivity_mean_band_' + str(10 + i2) dat[i].data = dat[i].data * 0.001 if 'Emissivity/SDev' in bandid2: bandid = 'Emissivity_std_dev_band_' + str(10 + i2) dat[i].data = dat[i].data * 0.0001 if 'NDVI/Mean' in bandid2: bandid = 'NDVI_mean' dat[i].data = dat[i].data * 0.01 if 'NDVI/SDev' in bandid2: bandid = 'NDVI_std_dev' dat[i].data = dat[i].data * 0.01 if 'Temperature/Mean' in bandid2: bandid = 'Temperature_mean' units = 'Kelvin' dat[i].data = dat[i].data * 0.01 if 'Temperature/SDev' in bandid2: bandid = 'Temperature_std_dev' units = 'Kelvin' dat[i].data = dat[i].data * 0.01 dat[i].dataid = bandid dat[i].nullvalue = nval dat[i].xdim = abs(lonsdim) dat[i].ydim = abs(latsdim) dat[i].units = units rows, cols = dat[i].data.shape xmin = tlx ymax = tly ymin = ymax - rows * dat[i].ydim xmax = xmin + cols * dat[i].xdim dat[i].extent = [xmin, xmax, ymin, ymax] srs = osr.SpatialReference() srs.ImportFromWkt(dataset.GetProjection()) srs.AutoIdentifyEPSG() dat[i].wkt = srs.ExportToWkt() dataset = None return dat
def get_aster(ifile): """ Gets ASTER Data Parameters ---------- ifile : str filename to import Returns ------- dat : PyGMI raster Data dataset imported """ dat = [] ifile = ifile[:] dataset = gdal.Open(ifile, gdal.GA_ReadOnly) subdata = dataset.GetSubDatasets() latentry = [i for i in subdata if 'Latitude' in i[1]] subdata.pop(subdata.index(latentry[0])) dataset = None dataset = gdal.Open(latentry[0][0], gdal.GA_ReadOnly) rtmp = dataset.GetRasterBand(1) lats = rtmp.ReadAsArray() latsdim = ((lats.max() - lats.min()) / (lats.shape[0] - 1)) / 2 lonentry = [i for i in subdata if 'Longitude' in i[1]] subdata.pop(subdata.index(lonentry[0])) dataset = None dataset = gdal.Open(lonentry[0][0], gdal.GA_ReadOnly) rtmp = dataset.GetRasterBand(1) lons = rtmp.ReadAsArray() lonsdim = ((lons.max() - lons.min()) / (lons.shape[1] - 1)) / 2 lonsdim = latsdim tlx = lons.min() - abs(lonsdim / 2) tly = lats.max() + abs(latsdim / 2) cols = int((lons.max() - lons.min()) / lonsdim) + 1 rows = int((lats.max() - lats.min()) / latsdim) + 1 newx2, newy2 = np.mgrid[0:rows, 0:cols] newx2 = newx2 * lonsdim + tlx newy2 = tlx - newy2 * latsdim subdata = [i for i in subdata if 'ImageData' in i[0]] i = -1 for ifile2, bandid2 in subdata: dataset = None dataset = gdal.Open(ifile2, gdal.GA_ReadOnly) rtmp2 = dataset.ReadAsArray() tmpds = gdal.AutoCreateWarpedVRT(dataset) rtmp2 = tmpds.ReadAsArray() gtr = tmpds.GetGeoTransform() tlx, lonsdim, _, tly, _, latsdim = gtr nval = 0 i += 1 dat.append(Data()) dat[i].data = rtmp2 if dat[i].data.dtype.kind == 'i': if nval is None: nval = 999999 nval = int(nval) elif dat[i].data.dtype.kind == 'u': if nval is None: nval = 0 nval = int(nval) else: if nval is None: nval = 1e+20 nval = float(nval) dat[i].data = np.ma.masked_invalid(dat[i].data) dat[i].data.mask = dat[i].data.mask | (dat[i].data == nval) if dat[i].data.mask.size == 1: dat[i].data.mask = (np.ma.make_mask_none(dat[i].data.shape) + dat[i].data.mask) dat[i].dataid = bandid2 dat[i].nullvalue = nval dat[i].xdim = abs(lonsdim) dat[i].ydim = abs(latsdim) rows, cols = dat[i].data.shape xmin = tlx ymax = tly ymin = ymax - rows * dat[i].ydim xmax = xmin + cols * dat[i].xdim dat[i].extent = [xmin, xmax, ymin, ymax] srs = osr.SpatialReference() srs.ImportFromWkt(dataset.GetProjection()) srs.AutoIdentifyEPSG() dat[i].wkt = srs.ExportToWkt() if dat == []: dat = None dataset = None return dat
def get_modis(ifile): """ Gets MODIS data Parameters ---------- ifile : str filename to import Returns ------- dat : PyGMI raster Data dataset imported """ dat = [] ifile = ifile[:] dataset = gdal.Open(ifile, gdal.GA_ReadOnly) subdata = dataset.GetSubDatasets() latentry = [i for i in subdata if 'Latitude' in i[1]] subdata.pop(subdata.index(latentry[0])) dataset = None dataset = gdal.Open(latentry[0][0], gdal.GA_ReadOnly) rtmp = dataset.GetRasterBand(1) lats = rtmp.ReadAsArray() latsdim = ((lats.max() - lats.min()) / (lats.shape[0] - 1)) / 2 lonentry = [i for i in subdata if 'Longitude' in i[1]] subdata.pop(subdata.index(lonentry[0])) dataset = None dataset = gdal.Open(lonentry[0][0], gdal.GA_ReadOnly) rtmp = dataset.GetRasterBand(1) lons = rtmp.ReadAsArray() lonsdim = ((lons.max() - lons.min()) / (lons.shape[1] - 1)) / 2 lonsdim = latsdim tlx = lons.min() - abs(lonsdim / 2) tly = lats.max() + abs(latsdim / 2) cols = int((lons.max() - lons.min()) / lonsdim) + 1 rows = int((lats.max() - lats.min()) / latsdim) + 1 newx2, newy2 = np.mgrid[0:rows, 0:cols] newx2 = newx2 * lonsdim + tlx newy2 = tlx - newy2 * latsdim tmp = [] for i in subdata: if 'HDF4_EOS:EOS_SWATH' in i[0]: tmp.append(i) subdata = tmp i = -1 for ifile2, bandid2 in subdata: dataset = None dataset = gdal.Open(ifile2, gdal.GA_ReadOnly) rtmp2 = dataset.ReadAsArray() if rtmp2.shape[-1] == min(rtmp2.shape) and rtmp2.ndim == 3: rtmp2 = np.transpose(rtmp2, (2, 0, 1)) nbands = 1 if rtmp2.ndim == 3: nbands = rtmp2.shape[0] for i2 in range(nbands): rtmp = dataset.GetRasterBand(i2 + 1) bandid = rtmp.GetDescription() nval = rtmp.GetNoDataValue() i += 1 dat.append(Data()) if rtmp2.ndim == 3: dat[i].data = rtmp2[i2] else: dat[i].data = rtmp2 newx = lons[dat[i].data != nval] newy = lats[dat[i].data != nval] newz = dat[i].data[dat[i].data != nval] if newx.size == 0: dat[i].data = np.zeros((rows, cols)) + nval else: tmp = quickgrid(newx, newy, newz, latsdim) mask = np.ma.getmaskarray(tmp) gdat = tmp.data dat[i].data = np.ma.masked_invalid(gdat[::-1]) dat[i].data.mask = mask[::-1] if dat[i].data.dtype.kind == 'i': if nval is None: nval = 999999 nval = int(nval) elif dat[i].data.dtype.kind == 'u': if nval is None: nval = 0 nval = int(nval) else: if nval is None: nval = 1e+20 nval = float(nval) dat[i].data = np.ma.masked_invalid(dat[i].data) dat[i].data.mask = (np.ma.getmaskarray(dat[i].data) | (dat[i].data == nval)) if dat[i].data.mask.size == 1: dat[i].data.mask = (np.ma.make_mask_none(dat[i].data.shape) + np.ma.getmaskarray(dat[i].data)) dat[i].dataid = bandid2 + ' ' + bandid dat[i].nullvalue = nval dat[i].xdim = abs(lonsdim) dat[i].ydim = abs(latsdim) rows, cols = dat[i].data.shape xmin = tlx ymax = tly ymin = ymax - rows * dat[i].ydim xmax = xmin + cols * dat[i].xdim dat[i].extent = [xmin, xmax, ymin, ymax] srs = osr.SpatialReference() srs.ImportFromWkt(dataset.GetProjection()) srs.AutoIdentifyEPSG() dat[i].wkt = srs.ExportToWkt() dataset = None return dat
def get_raster(ifile, nval=None): """ This function loads a raster dataset off the disk using the GDAL libraries. It returns the data in a PyGMI data object. Parameters ---------- ifile : str filename to import nval : float, optional No data/null value. The default is None. Returns ------- dat : PyGMI raster Data dataset imported """ dat = [] bname = ifile.split('/')[-1].rpartition('.')[0] + ': ' ifile = ifile[:] ext = ifile[-3:] custom_wkt = None # Envi Case if ext == 'hdr': ifile = ifile[:-4] tmp = glob.glob(ifile + '.dat') if tmp: ifile = tmp[0] if ext == 'ers': with open(ifile) as f: metadata = f.read() if 'STMLO' in metadata: clong = metadata.split('STMLO')[1][:2] orig = osr.SpatialReference() if 'CAPE' in metadata: orig.ImportFromEPSG(4222) orig.SetTM(0., float(clong), 1., 0., 0.) orig.SetProjCS(r'Cape / TM' + clong) custom_wkt = orig.ExportToWkt() elif 'WGS84' in metadata: orig.ImportFromEPSG(4148) orig.SetTM(0., float(clong), 1., 0., 0.) orig.SetProjCS(r'Hartebeesthoek94 / TM' + clong) custom_wkt = orig.ExportToWkt() dataset = gdal.Open(ifile, gdal.GA_ReadOnly) if dataset is None: return None gtr = dataset.GetGeoTransform() for i in range(dataset.RasterCount): rtmp = dataset.GetRasterBand(i + 1) bandid = rtmp.GetDescription() if nval is None: nval = rtmp.GetNoDataValue() dat.append(Data()) dat[i].data = rtmp.ReadAsArray() if dat[i].data.dtype.kind == 'i': if nval is None: nval = 999999 nval = int(nval) elif dat[i].data.dtype.kind == 'u': if nval is None: nval = 0 nval = int(nval) else: if nval is None: nval = 1e+20 nval = float(nval) if ext == 'ers' and nval == -1.0e+32: dat[i].data[np.ma.less_equal(dat[i].data, nval)] = -1.0e+32 # Note that because the data is stored in a masked array, the array ends up # being double the size that it was on the disk. dat[i].data = np.ma.masked_invalid(dat[i].data) dat[i].data.mask = (np.ma.getmaskarray(dat[i].data) | (dat[i].data == nval)) if dat[i].data.mask.size == 1: dat[i].data.mask = (np.ma.make_mask_none(dat[i].data.shape) + np.ma.getmaskarray(dat[i].data)) dat[i].extent_from_gtr(gtr) if bandid == '': bandid = bname + str(i + 1) dat[i].dataid = bandid if bandid[-1] == ')': dat[i].units = bandid[bandid.rfind('(') + 1:-1] dat[i].nullvalue = nval if custom_wkt is None: srs = osr.SpatialReference() srs.ImportFromWkt(dataset.GetProjection()) srs.AutoIdentifyEPSG() dat[i].wkt = srs.ExportToWkt() else: dat[i].wkt = custom_wkt dataset = None return dat
def get_ascii(ifile): """ Import ascii raster dataset Parameters ---------- ifile : str filename to import Returns ------- dat : PyGMI raster Data dataset imported """ isESRI = False with open(ifile, 'r') as afile: adata = afile.read() adata = adata.split() if adata[0] == 'ncols': isESRI = True if isESRI: nbands = 1 ncols = int(adata[1]) nrows = int(adata[3]) xdim = float(adata[9]) ydim = float(adata[9]) nval = float(adata[11]) ulxmap = float(adata[5]) ulymap = float(adata[7]) + ydim * nrows if 'center' in adata[4].lower(): ulxmap = ulxmap - xdim / 2 if 'center' in adata[6].lower(): ulymap = ulymap - ydim / 2 adata = adata[12:] else: with open(ifile[:-3] + 'hdr', 'r') as hfile: tmp = hfile.readlines() xdim = float(tmp[0].split()[-1]) ydim = float(tmp[1].split()[-1]) ncols = int(tmp[2].split()[-1]) nrows = int(tmp[3].split()[-1]) nbands = int(tmp[4].split()[-1]) ulxmap = float(tmp[5].split()[-1]) ulymap = float(tmp[6].split()[-1]) nval = -9999.0 bandid = ifile[:-4].rsplit('/')[-1] adata = np.array(adata, dtype=float) adata.shape = (nrows, ncols) if nbands > 1: warnings.warn('PyGMI only supports single band ASCII files. ' 'Only first band will be exported.') dat = [Data()] i = 0 dat[i].data = np.ma.masked_equal(adata, nval) if dat[i].data.mask.size == 1: dat[i].data.mask = (np.ma.make_mask_none(dat[i].data.shape) + dat[i].data.mask) dat[i].dataid = bandid dat[i].nullvalue = nval dat[i].xdim = xdim dat[i].ydim = ydim rows, cols = dat[i].data.shape xmin = ulxmap ymax = ulymap ymin = ymax - rows * ydim xmax = xmin + cols * xdim dat[i].extent = [xmin, xmax, ymin, ymax] return dat
def get_geosoft(hfile): """ Get geosoft file Parameters ---------- ifile : str filename to import Returns ------- dat : PyGMI Data Dataset imported """ f = open(hfile, mode='rb') es = np.fromfile(f, dtype=np.int32, count=1)[0] # 4 sf = np.fromfile(f, dtype=np.int32, count=1)[0] # signf ncols = np.fromfile(f, dtype=np.int32, count=1)[0] # ncol nrows = np.fromfile(f, dtype=np.int32, count=1)[0] # nrow kx = np.fromfile(f, dtype=np.int32, count=1)[0] # 1 dx = np.fromfile(f, dtype=np.float64, count=1)[0] # dx dy = np.fromfile(f, dtype=np.float64, count=1)[0] # dy x0 = np.fromfile(f, dtype=np.float64, count=1)[0] # xllcor y0 = np.fromfile(f, dtype=np.float64, count=1)[0] # yllcor rot = np.fromfile(f, dtype=np.float64, count=1)[0] # rot zbase = np.fromfile(f, dtype=np.float64, count=1)[0] # zbase zmult = np.fromfile(f, dtype=np.float64, count=1)[0] # zmult label = np.fromfile(f, dtype='a48', count=1)[0] mapno = np.fromfile(f, dtype='a16', count=1)[0] proj = np.fromfile(f, dtype=np.int32, count=1)[0] unitx = np.fromfile(f, dtype=np.int32, count=1)[0] unity = np.fromfile(f, dtype=np.int32, count=1)[0] unitz = np.fromfile(f, dtype=np.int32, count=1)[0] nvpts = np.fromfile(f, dtype=np.int32, count=1)[0] izmin = np.fromfile(f, dtype=np.int32, count=1)[0] izmax = np.fromfile(f, dtype=np.int32, count=1)[0] izmed = np.fromfile(f, dtype=np.int32, count=1)[0] izmea = np.fromfile(f, dtype=np.int32, count=1)[0] zvar = np.fromfile(f, dtype=np.float64, count=1)[0] prcs = np.fromfile(f, dtype=np.int32, count=1)[0] temspc = np.fromfile(f, dtype='a324', count=1)[0] if es == 2: nval = -32767 data = np.fromfile(f, dtype=np.int16, count=nrows * ncols) elif es == 4: data = np.fromfile(f, dtype=np.float32, count=nrows * ncols) nval = -1.0E+32 else: return None data = np.ma.masked_equal(data, nval) data = data / zmult + zbase data.shape = (nrows, ncols) data = data[::-1] f.close() dat = [] dat.append(Data()) i = 0 dat[i].data = data dat[i].dataid = hfile[:-4] dat[i].nullvalue = nval dat[i].xdim = dx dat[i].ydim = dy xmin = x0 ymax = y0 + dy * nrows ymin = y0 xmax = xmin + ncols * dx dat[i].extent = [xmin, xmax, ymin, ymax] return dat
def get_geopak(hfile): """ GeoPak Import Parameters ---------- hfile : str filename to import Returns ------- dat : PyGMI raster Data dataset imported Returns ------- dat : PyGMI Data PyGMI raster dataset. """ with open(hfile, 'rb') as fin: fall = fin.read() off = 0 fnew = [] while off < len(fall): off += 1 breclen = np.frombuffer(fall, dtype=np.uint8, count=1, offset=off)[0] if breclen == 130: break reclen = breclen if breclen == 129: reclen = 128 off += 1 fnew.append(fall[off:off + reclen]) off += reclen fnew = b''.join(fnew) header = np.frombuffer(fnew, dtype=np.float32, count=32, offset=0) # Lines in grid 1 # Points per line 2 # Grid factor 3 # Grid base value 4 # Grid X origin 5 # Grid Y origin 6 # Grid rotation 7 # Grid dummy value 8 # Map scale 9 # Cell size (X) 10 # Cell size (Y) 11 # Inches/unit 12 # Grid X offset 13 # Grid Y offset 14 # Grid hdr version 15 # # Lines in grid 17 # Points per line 18 # Grid factor 21 # Grid base value 22 # Z maximum 23 # Z minimum 24 # # Grid dummy value 26 nrows = int(header[0]) ncols = int(header[1]) gfactor = header[2] gbase = header[3] x0 = header[4] y0 = header[5] # rotation = header[6] nval = header[7] # mapscale = header[8] dx = header[9] dy = header[10] # inches_per_unit = header[11] # xoffset = header[12] # yoffset = header[13] # hver = header[14] # zmax = header[22] # zmin = header[23] data = np.frombuffer(fnew, dtype=np.int16, count=(nrows * ncols), offset=128) data = np.ma.masked_equal(data, nval) data = data / gfactor + gbase data.shape = (nrows, ncols) data = data[::-1] dat = [] dat.append(Data()) i = 0 dat[i].data = data dat[i].dataid = hfile[:-4] dat[i].nullvalue = nval dat[i].xdim = dx dat[i].ydim = dy xmin = x0 ymax = y0 + dy * nrows ymin = y0 xmax = xmin + ncols * dx dat[i].extent = [xmin, xmax, ymin, ymax] return dat
def settings(self): """ Settings. Returns ------- bool True if successful, False otherwise. """ if 'Raster' not in self.indata: print('Error: You must have a multi-band raster dataset in ' 'addition to your cluster analysis results') return False self.map.data = self.indata['Raster'] bands = [i.dataid for i in self.indata['Raster']] self.combo.clear() self.combo.addItems(bands) self.combo.currentIndexChanged.connect(self.on_combo) self.map.init_graph() self.map.polyint() self.map.polyi.polyi_changed.connect(self.updatepoly) self.map.update_graph() tmp = self.exec_() if tmp == 0: return False classifier, lbls, datall, _, _ = self.init_classifier() rows, cols, bands = datall.shape datall.shape = (rows * cols, bands) yout = classifier.predict(datall) yout.shape = (rows, cols) datall.shape = (rows, cols, bands) data = copy.copy(self.indata['Raster']) dat_out = [Data()] dat_out[-1].metadata['Cluster']['input_type'] = [] for k in data: dat_out[-1].metadata['Cluster']['input_type'].append(k.dataid) zonal = np.ma.array(yout, mask=self.map.data[0].data.mask) if self.parent is None: plt.imshow(zonal) plt.show() i = len(lbls) dat_out[-1].data = zonal dat_out[-1].nullvalue = zonal.fill_value dat_out[-1].metadata['Cluster']['no_clusters'] = i dat_out[-1].metadata['Cluster']['center'] = np.zeros([i, len(data)]) dat_out[-1].metadata['Cluster']['center_std'] = np.zeros( [i, len(data)]) # if cfit.labels_.max() > 0: # dat_out[-1].metadata['Cluster']['vrc'] = skm.calinski_harabasz_score(X, cfit.labels_) m = [] s = [] for i2 in lbls: m.append(datall[yout == i2].mean(0)) s.append(datall[yout == i2].std(0)) dat_out[-1].metadata['Cluster']['center'] = np.array(m) dat_out[-1].metadata['Cluster']['center_std'] = np.array(s) # self.log = ('Cluster complete' + ' (' + self.cltype+')') # dat_out[-1].xdim = data[0].xdim dat_out[-1].ydim = data[0].ydim dat_out[-1].dataid = 'Clusters: ' + str( dat_out[-1].metadata['Cluster']['no_clusters']) dat_out[-1].nullvalue = data[0].nullvalue dat_out[-1].extent = data[0].extent for i in dat_out: i.data += 1 i.data = i.data.astype(np.uint8) i.nullvalue = 0 i.data.data[i.data.mask] = 0 print('Cluster complete') self.outdata['Cluster'] = dat_out self.outdata['Raster'] = self.indata['Raster'] return True
def run(self): """ Run. Returns ------- None. """ data = copy.copy(self.indata['Raster']) self.update_vars() cltype = self.cltype cov_constr = self.constrain no_runs = self.runs max_iter = self.max_iter term_thresh = self.term_thresh no_clust = np.array([self.min_cluster, self.max_cluster]) de_norm = self.denorm expo = self.fexp print('Fuzzy Clustering started') # ############################################################################# # Section to deal with different bands having different null values. masktmp = data[0].data.mask # Start with the first entry # Add the masks to this.This promotes False values to True if necessary for i in data: masktmp += i.data.mask for i, _ in enumerate(data): # Apply this to all the bands data[i].data.mask = masktmp # ############################################################################# dat_in = np.array([i.data.compressed() for i in data]).T if self.radiobutton_manual.isChecked() is True: ext = ('ASCII matrix (*.txt);;' 'ASCII matrix (*.asc);;' 'ASCII matrix (*.dat)') filename = QtWidgets.QFileDialog.getOpenFileName( self.parent, 'Read Cluster Centers', '.', ext) if filename == '': return False os.chdir(os.path.dirname(filename)) ifile = str(filename) dummy_mod = np.ma.array(np.genfromtxt(ifile, unpack=True)) [row, col] = np.shape(dummy_mod) ro1 = np.sum(list(range(no_clust[0], no_clust[1] + 1))) if dat_in.shape[1] != col or row != ro1: QtWidgets.QMessageBox.warning(self.parent, 'Warning', ' Incorrect matrix size!', QtWidgets.QMessageBox.Ok, QtWidgets.QMessageBox.Ok) cnt = -1 for i in range(no_clust[0], no_clust[1] + 1): smtmp = np.zeros(i) for j in range(i): cnt = cnt + 1 smtmp[j] = dummy_mod[cnt] startmdat = {i: smtmp} startmfix = {i: []} filename = QtWidgets.QFileDialog.getOpenFileName( self.parent, 'Read Cluster Center Constraints', '.', ext) if filename == '': QtWidgets.QMessageBox.warning( self.parent, 'Warning', 'Running cluster analysis without constraints', QtWidgets.QMessageBox.Ok, QtWidgets.QMessageBox.Ok) else: ifile = str(filename) dummy_mod = np.ma.array(np.genfromtxt(ifile, unpack=True)) [row, col] = np.shape(dummy_mod) ro1 = np.sum(list(range(no_clust[0], no_clust[1] + 1))) if dat_in.shape[1] != col or row != ro1: QtWidgets.QMessageBox.warning(self.parent, 'Warning', ' Incorrect matrix size!', QtWidgets.QMessageBox.Ok, QtWidgets.QMessageBox.Ok) cnt = -1 for i in range(no_clust[0], no_clust[1] + 1): smtmp = np.zeros(i) for j in range(i): cnt = cnt + 1 smtmp = dummy_mod[cnt] startmfix = {i: smtmp} cnt = -1 dat_out = [Data() for i in range(no_clust[0], no_clust[1] + 1)] for i in range(no_clust[0], no_clust[1] + 1): print('Number of Clusters:' + str(i)) cnt = cnt + 1 if self.radiobutton_datadriven.isChecked() is True: print('Initial guess: data driven') no_samp = dat_in.shape[0] dno_samp = no_samp / i idx = np.arange(0, no_samp + dno_samp, dno_samp) idx[0] = 1 startmdat = {i: np.zeros([i, dat_in.shape[1]])} dat_in1 = dat_in smtmp = np.zeros([i, dat_in.shape[1]]) for k in range(dat_in.shape[1]): # this is same as matlab sortrows # dat_in1[dat_in1[:, k].argsort()] for j in range(i): smtmp[j, k] = np.median(dat_in1[idx[j]:idx[j + 1], k]) startmdat = {i: smtmp} startmfix = {i: np.array([])} # inserted 'np.array' del dat_in1 clu, clcent, clobj_fcn, clvrc, clnce, clxbi = self.fuzzy_means( dat_in, i, startmdat[i], startmfix[i], max_iter, term_thresh, expo, cltype, cov_constr) elif self.radiobutton_manual.isChecked() is True: print('Initial guess: manual') clu, clcent, clobj_fcn, clvrc, clnce, clxbi = self.fuzzy_means( dat_in, i, startmdat[i], startmfix[i], max_iter, term_thresh, expo, cltype, cov_constr) elif self.radiobutton_random.isChecked() is True: print('Initial guess: random') clobj_fcn = np.array([np.Inf]) for j in range(no_runs): print('Run ' + str(j + 1) + ' of' + str(no_runs)) xmins = np.minimum(dat_in, 1) xmaxs = np.maximum(dat_in, 1) startm1dat = { i: np.random.uniform(xmins[np.zeros(i, int), :], xmaxs[np.zeros(i, int), :]) } startmfix = {i: np.array([])} clu1, clcent1, clobj_fcn1, clvrc1, clnce1, clxbi1 = \ self.fuzzy_means(dat_in, i, startm1dat[i], startmfix[i], max_iter, term_thresh, expo, cltype, cov_constr) if clobj_fcn1[-1] < clobj_fcn[-1]: clu = clu1 clcent = clcent1 clobj_fcn = clobj_fcn1 clnce = clnce1 clxbi = clxbi1 clvrc = clvrc1 startmdat = {i: startm1dat[i]} clalp = np.array(clu).max(0) # 0 means row wise max? clidx = np.array(clu).argmax(0) clalp = clalp - (1.0 / clcent.shape[0]) clalp = clalp / clalp.max() clalp[clalp > 1] = 1 clalp[clalp < 0] = 0 zonal = np.ma.zeros(data[0].data.shape) - 9999.0 alpha = np.ma.zeros(data[0].data.shape) - 9999.0 alpha1 = (data[0].data.mask == 0) zonal[alpha1 == 1] = clidx alpha[alpha1 == 1] = clalp zonal.mask = data[0].data.mask alpha.mask = data[0].data.mask cent_std = np.array( [np.std(dat_in[clidx == k], 0) for k in range(i)]) # den_cent = clcent # den_cent_std = np.array(cent_std, copy=True) # den_cent_std1 = np.array(cent_std, copy=True) if de_norm == 1: pass dat_out[cnt].metadata['Cluster']['input_type'] = [] for k in data: dat_out[cnt].metadata['Cluster']['input_type'].append(k.dataid) dat_out[cnt].data = np.ma.array(zonal) dat_out[cnt].metadata['Cluster']['no_clusters'] = i dat_out[cnt].metadata['Cluster']['center'] = clcent dat_out[cnt].metadata['Cluster']['center_std'] = cent_std dat_out[cnt].metadata['Cluster']['memdat'] = [] for k in range(clcent.shape[0]): dummy = np.ones(data[0].data.shape) * np.nan alpha1 = (data[0].data.mask == 0) dummy[alpha1 == 1] = clu[k, :] dat_out[cnt].metadata['Cluster']['memdat'].append(dummy) dat_out[cnt].metadata['Cluster']['vrc'] = clvrc dat_out[cnt].metadata['Cluster']['nce'] = clnce dat_out[cnt].metadata['Cluster']['xbi'] = clxbi dat_out[cnt].metadata['Cluster']['obj_fcn'] = clobj_fcn # dat_out[cnt].type = self.type # dat_out[cnt].algorithm = cltype # dat_out[cnt].initialization = init_type # dat_out[cnt].init_mod = startmdat[i] # dat_out[cnt].init_constrains = startmfix[i] # dat_out[cnt].runs = no_runs # dat_out[cnt].max_iterations = max_iter # dat_out[cnt].denormalize = de_norm # dat_out[cnt].term_threshold = term_thresh # dat_out[cnt].shape_constrain = cov_constr # dat_out[cnt].zonal = zonal # dat_out[cnt].alpha = alpha # dat_out[cnt].memdat = np.ma.array(dat_out[cnt].memdat) # dat_out[cnt].memdat.mask = dat_out[cnt].alpha.mask # dat_out[cnt].xxx = data[0].xxx # dat_out[cnt].yyy = data[0].yyy # dat_out[cnt].denorm_center = den_cent # dat_out[cnt].denorm_center_stdup = den_cent_std # dat_out[cnt].denorm_center_stdlow = den_cent_std1 # dat_out[cnt].iterations = clobj_fcn.size # dat_out[cnt].fuzziness_exp = expo # gaugetmp.crisplogtxt.Value += '\n'+logtxt for i in dat_out: i.xdim = data[0].xdim i.ydim = data[0].ydim i.dataid = 'Fuzzy Cluster: ' + str( i.metadata['Cluster']['no_clusters']) i.nullvalue = data[0].nullvalue i.extent = data[0].extent i.data += 1 print('Fuzzy Cluster complete' + ' (' + self.cltype + ' ' + self.init_type + ')') self.outdata['Cluster'] = dat_out self.outdata['Raster'] = self.indata['Raster'] return True