Documentation page for AR_tracer.py¶

Functions to compute Hausdorff distance between AR axes pairs and link ARs across time steps to form tracks.

Author: guangzhi XU (xugzhi1987@gmail.com) Update time: 2020-06-05 22:46:19.

class AR_tracer.AR(id, data)¶

Ojbect representing an AR entity

Hausdorff(lats, lons)¶

Compute modified Hausdorff distance from the lastest record to given axis

Parameters:	lats (ndarray) – 1d array, the target axis’s latitude coordinates. lons (ndarray) – 1d array, the target axis’s longitude coordinates.
Returns:	float – modified Hausdorff distance from this AR to the given axis.

__init__(id, data)¶

Parameters:	id (int) – a numeric id for each AR. data (pandas.DataFrame) – DataFrame storing an AR’s records.

anchor_lats¶

1d array, get the latitude coordinates from roughly evenly spaced points from the AR axis.

Type:	ndarray

anchor_lons¶

1d array, get the longitude coordinates from roughly evenly spaced points from the AR axis.

Type:	ndarray

append(ar)¶: Add new records to the AR track

backwardHausdorff(lats, lons)¶

Compute backward Hausdorff distance from the lastest record to given axis

Parameters:	lats (ndarray) – 1d array, the target axis’s latitude coordinates. lons (ndarray) – 1d array, the target axis’s longitude coordinates.
Returns:	float – backward Hausdorff distance from this AR to the given axis.

coor¶

(Nx3) ndarray, (time, lat_centroid, lon_centroid) coordinates of an AR track.

Type:	ndarray

duration¶

track duration in hours.

Type:	int

forwardHausdorff(lats, lons)¶

Compute forward Hausdorff distance from the lastest record to given axis

Parameters:	lats (ndarray) – 1d array, the target axis’s latitude coordinates. lons (ndarray) – 1d array, the target axis’s longitude coordinates.
Returns:	float – forward Hausdorff distance from this AR to the given axis.

latest¶

the AR record of the latest time point in an AR track.

Type:	Series

lats¶

1d array, the latitude coordinates of the AR axis in the latest record in an AR’s track.

Type:	ndarray

lons¶

1d array, the longitude coordinates of the AR axis in the latest record in an AR’s track.

Type:	ndarray

rdp_lats¶

1d array, the latitude coordinates of the simplified AR axis in the latest record in an AR’s track.

Type:	ndarray

rdp_lons¶

1d array, the longitude coordinates of the simplified AR axis in the latest record in an AR’s track.

Type:	ndarray

times¶

sorted time stamps of an AR track.

Type:	Series

AR_tracer.filterTracks(tr_list, min_duration, min_nonrelax, verbose=True)¶

Filter tracks

Keyword Arguments:
Parameters:	tr_list (list) – list of AR objects, found tracks. min_duration (int) – min duration in hrs to keep a track. min_nonrelax (int) – min number of non-relaxed records in a track to keep a track.
	verbose (bool) – print some messages or not.
Returns:	tr_list (list) – list of AR objects, filtered tracks.

Tracks that are filtered:

tracks that are too short, controlled by ‘min_duration’
tracks that consist of solely relaxed records.

AR_tracer.forwardHausdorff(lats1, lons1, lats2, lons2)¶

Compute forward Hausdorff distance betweem 2 tracks

Parameters:	lats1 (list or 1D array) – latitudes of track1. lons1 (list or 1D array) – longitudes of track1. lats2 (list or 1D array) – latitudes of track2. lons2 (list or 1D array) – longitudes of track2.
Returns:	forward Hausdorff distance in km.

AR_tracer.getAnchors(arr, num_anchors=7)¶

Get anchor points along from an axis.

Parameters:	arr (ndarray) – 1D array from which to sample the anchor points.
Returns:	(ndarray) – 1D array of the sampled anchor points from <arr>.

AR_tracer.getDistMatrix(tr_list, newlats, newlons)¶

Compute distance matrix among track axis anchors

Parameters:

tr_list (list) – list of AR objs, existing systems at time t=t.
newlats (list or 1D array) – latitudes at t=t+1.
newlons (list or 1D array) – longitudes at t=t+1.

Returns:

dists (ndarray) –

n*m matrix consisting distances between existing: and new tracks. Rows as new records at tnow, columns as existing tracks.

AR_tracer.matchCenters(tr_list, newrec, time_gap_allow, max_dist_allow, track_scheme='simple', isplot=False, plot_dir=None, verbose=True)¶

Match and link nearby centers at 2 consecutive time steps

Parameters:

tr_list (list) – list of AR objs, existing systems at time t=t.
newrec (DataFrame) – new center data at time t=t+1.
time_gap_allow (int) – max allowed gap between 2 records, in number of hours.
max_dist_allow (float) – max allowed Hausdorff distance allowed between 2 records, in km.

Keyword Arguments:

track_scheme (str) – tracking scheme. ‘simple’: all tracks are simple
'full' (paths.) – use the network scheme, tracks are connected by their
points. (joint) –
isplot (bool) – create schematic plot or not.
plot_dir (str) – folder to save schematic plot. Only used if isplot=True.
verbose (bool) – print some messages or not.

Returns:

tr_list (list) –

list of AR objs, ARs with new matching records: appended at the end.
allocated_recs (list): list of ints, ids of new records that are: attributed to existing systems during the process.

Matching is based on geo-distances and uses nearest neighbour strategy.

AR_tracer.plotHD(y1, x1, y2, x2, timelabel=None, linkflag='', ax=None, show=True)¶

Plot Hausdorff links

Keyword Arguments:
Parameters:	y1,x1 (ndarray) – 1d array, y, x coordinates of AR axis A. y2,x2 (ndarray) – 1d array, y, x coordinates of AR axis B.
	timelabel (str or None) – string of the time stamp. If given, plot as subplot title. linkflag (str) – a single char to denote the type of linking, used in generated plot. ‘’ for initial linking, ‘M’ for a merging, ‘S’ for a splitting. ax (plt axis obj) – if not give, create a new axis to plot with. show (bool) – whether to show the figure or not.

AR_tracer.readCSVRecord(abpath_in)¶

Read in individual AR records from .csv file

Parameters:	abpath_in (str) – absolute file path to AR record file.
Returns:	ardf (pandas.DataFrame) – record saved in DataFrame.

New in v2.0.

AR_tracer.trackARs(record, time_gap_allow, max_dist_allow, track_scheme='simple', isplot=False, plot_dir=None, verbose=True)¶

Track ARs at consecutive time points to form tracks

Keyword Arguments:
Parameters:	record (DataFrame) – AR records at different time slices. time_gap_allow (int) – max allowed gap between 2 records, in number of hours. max_dist_allow (float) – max allowed Hausdorff distance allowed between 2 records, in km.
	track_scheme (str) – tracking scheme. ‘simple’: all tracks are simple 'full' (paths.) – use the network scheme, tracks are connected by their points. (joint) – isplot (bool) – whether to create schematic plots of linking. plot_dir (str) – folder to save schematic plot. verbose (bool) – print some messages or not.
Returns:	finished_list (list) – list of AR objs. Found tracks.