This object is used to examine a network of channels in chi space. More...

#include <LSDChiNetwork.hpp>

Public Member Functions
	LSDChiNetwork (string channel_network_fname)
	Crate routine to make a LSDChiNetwork object. More...

	LSDChiNetwork (LSDFlowInfo &FlowInfo, int SourceNode, int OutletNode, LSDRaster &Elevation, LSDRaster &FlowDistance, LSDRaster &DrainageArea)

	LSDChiNetwork (LSDFlowInfo &FlowInfo, int SourceNode, int OutletNode, LSDRaster &Elevation, LSDRaster &FlowDistance, LSDRaster &DrainageArea, LSDRaster &Chi)

int	get_n_channels ()

int	get_NRows () const

int	get_NCols () const

float	get_XMinimum () const

float	get_YMinimum () const

float	get_DataResolution () const

int	get_NoDataValue () const

void	print_channel_details_to_screen (int channel_number)
	Print channel details to screen for bug checking. More...

void	print_channel_details_to_file (string fname, float A_0, float m_over_n)
	Print channel details to file for bug checking. More...

void	print_channel_details_to_file_full_fitted (string fname)
	This function prints the details of all channels to a file. More...

void	print_channel_details_to_file_full_fitted_for_ArcMap (string fname)
	This function prints the details of all channels to a csv file that can be ingested by ArcMap. More...

void	print_channel_details_to_file_full_fitted (string fname, int target_nodes, int minimum_segment_length)
	This function prints the details of all channels to a file. More...

void	extend_tributaries_to_outlet ()
	This extends the tributary channels all the way to the outlet. More...

Array2D< float >	data_to_array (int data_member)
	Routine for returning calculated data to an array. More...

void	slope_area_extraction_vertical_intervals (float interval, float area_thin_fraction, string fname)
	Extract slope over fixed vertical intervals. More...

void	slope_area_extraction_horizontal_intervals (float interval, float area_thin_fraction, string fname)
	Extract slope over fixed horizontal intervals. More...

void	calculate_chi (float A_0, float m_over_n)
	This function calculates the chi values for the channel network using the rectangle rule. More...

float	calculate_optimal_chi_spacing (int target_nodes)
	This function calucaltes the chi spacing of the main stem channel (the longest channel). More...

int	calculate_skip (int target_nodes)
	This function calucaltes the skip parameter of the main stem (the longest channel). More...

int	calculate_skip (int target_nodes, vector< float > &sorted_chis)
	This function calucaltes the skip parameter based on a vector of chi values. More...

int	calculate_skip (int target_nodes, int channel_number)
	This function calucaltes the skip parameter of a give channel. More...

void	find_most_likeley_segments (int channel, int minimum_segment_length, float sigma, int N, vector< float > &b_vec, vector< float > &m_vec, vector< float > &r2_vec, vector< float > &DW_vec, vector< float > &thinned_chi, vector< float > &thinned_elev, vector< float > &fitted_elev, vector< int > &node_reference, vector< int > &these_segment_lengths, float &this_MLE, int &this_n_segments, int &n_data_nodes, float &this_AIC, float &this_AICc)
	This function gets the most likely channel segments for a particular channel. More...

void	find_most_likeley_segments_dchi (int channel, int minimum_segment_length, float sigma, float dchi, vector< float > &b_vec, vector< float > &m_vec, vector< float > &r2_vec, vector< float > &DW_vec, vector< float > &thinned_chi, vector< float > &thinned_elev, vector< float > &fitted_elev, vector< int > &node_reference, vector< int > &these_segment_lengths, float &this_MLE, int &this_n_segments, int &n_data_nodes, float &this_AIC, float &this_AICc)
	This function gets the most likely channel segments for a particular channel. More...

void	find_most_likeley_segments_monte_carlo (int channel, int minimum_segment_length, float sigma, int mean_skip, int skip_range, vector< float > &b_vec, vector< float > &m_vec, vector< float > &r2_vec, vector< float > &DW_vec, vector< float > &thinned_chi, vector< float > &thinned_elev, vector< float > &fitted_elev, vector< int > &node_reference, vector< int > &these_segment_lengths, float &this_MLE, int &this_n_segments, int &n_data_nodes, float &this_AIC, float &this_AICc)
	This gets the most likely segments but uses the monte carlo data thinning method. More...

void	find_most_likeley_segments_monte_carlo_dchi (int channel, int minimum_segment_length, float sigma, float mean_dchi, float variation_dchi, vector< float > &b_vec, vector< float > &m_vec, vector< float > &r2_vec, vector< float > &DW_vec, vector< float > &thinned_chi, vector< float > &thinned_elev, vector< float > &fitted_elev, vector< int > &node_reference, vector< int > &these_segment_lengths, float &this_MLE, int &this_n_segments, int &n_data_nodes, float &this_AIC, float &this_AICc)
	This gets the most likely segments but uses the monte carlo data thinning method. More...

float	search_for_best_fit_m_over_n_dchi (float A_0, int n_movern, float d_movern, float start_movern, int minimum_segment_length, float sigma, int target_nodes_mainstem, string fname)
	The master routine for calculating the best fit m over n values for a channel network, based on a fixed value of dchi. More...

float	search_for_best_fit_m_over_n (float A_0, int n_movern, float d_movern, float start_movern, int minimum_segment_length, float sigma, int target_nodes_mainstem, string fname)
	The master routine for calculating the best fit m over n values for a channel network. More...

float	search_for_best_fit_m_over_n_seperate_ms_and_tribs (float A_0, int n_movern, float d_movern, float start_movern, int minimum_segment_length, float sigma, int target_nodes_mainstem, string fname)
	Routine for calculating the best fit m over n values for a channel network, but calculates the mainstem and the tributaries seperately. More...

float	search_for_best_fit_m_over_n_colinearity_test (float A_0, int n_movern, float d_movern, float start_movern, int minimum_segment_length, float sigma, int target_nodes, int n_iterations, vector< float > &m_over_n_values, vector< float > &AICc_mean, vector< float > &AICc_sdtd)
	This function looks for the best fit values of m over n by simply testing for the least variation in the tributaries. More...

float	search_for_best_fit_m_over_n_colinearity_test_with_breaks (float A_0, int n_movern, float d_movern, float start_movern, int minimum_segment_length, float sigma, int target_skip, int target_nodes, int n_iterations, vector< float > &m_over_n_values, vector< float > &AICc_mean, vector< float > &AICc_sdtd, int Monte_Carlo_switch)
	This function calculeates best fit m/n using the collinearity test these channels are ones with breaks. More...

float	search_for_best_fit_m_over_n_individual_channels_with_breaks (float A_0, int n_movern, float d_movern, float start_movern, int minimum_segment_length, float sigma, int target_skip, int target_nodes, int n_iterations, vector< float > &m_over_n_values, vector< vector< float > > &AICc_vals)
	This function calculeates best fit m/n for each channel these channels are ones with breaks. More...

float	search_for_best_fit_m_over_n_individual_channels_with_breaks_monte_carlo (float A_0, int n_movern, float d_movern, float start_movern, int minimum_segment_length, float sigma, int target_skip, int target_nodes, int n_iterations, vector< float > &m_over_n_values, vector< vector< float > > &AICc_means, vector< vector< float > > &AICc_stddev)
	This gets the best fit m over n values of all the individual tributaries. More...

void	monte_carlo_sample_river_network_for_best_fit_dchi (float A_0, float m_over_n, int n_iterations, float fraction_dchi_for_variation, int minimum_segment_length, float sigma, int target_nodes_mainstem)
	This routine uses a monte carlo approach to repeatedly sampling all the data in the channel network. More...

void	monte_carlo_sample_river_network_for_best_fit (float A_0, float m_over_n, int n_iterations, int mean_skip, int skip_range, int minimum_segment_length, float sigma)
	Monte carlo segment fitter. More...

void	monte_carlo_sample_river_network_for_best_fit_after_breaks (float A_0, float m_over_n, int n_iterations, int skip, int minimum_segment_length, float sigma)
	This function samples the river network using monte carlo samplig but after breaking the channels. More...

void	monte_carlo_split_channel (float A_0, float m_over_n, int n_iterations, int target_skip, int target_nodes, int minimum_segment_length, float sigma, int chan, vector< int > &break_nodes)
	Monte carlo segment fitter. More...

void	monte_carlo_split_channel_colinear (float A_0, float m_over_n, int n_iterations, int target_skip, int target_nodes, int minimum_segment_length, float sigma, vector< float > reverse_Chi, vector< float > reverse_Elevation, vector< int > &break_nodes)
	This function uses a monte carlo sampling approach to try and split channels. More...

void	split_all_channels (float A_0, float m_over_n, int n_iterations, int target_skip, int target_nodes, int minimum_segment_length, float sigma)
	This function splits all the channels in one go. More...

float	calculate_AICc_after_breaks (float A_0, float m_over_n, int skip, int minimum_segment_length, float sigma, int chan, vector< int > break_nodes, int &n_total_segments, int &n_total_nodes, float &cumulative_MLE)
	This function gets the AICc after breaking the channel. More...

vector< float >	calculate_AICc_after_breaks_monte_carlo (float A_0, float m_over_n, int target_skip, int minimum_segment_length, float sigma, int chan, vector< int > break_nodes, int &n_total_segments, int &n_total_nodes, float &cumulative_MLE, int n_iterations)
	This function gets the AICc after breaking the channel. More...

float	calculate_AICc_after_breaks_colinear (float A_0, float m_over_n, int skip, int minimum_segment_length, float sigma, vector< float > reverse_chi, vector< float > reverse_elevation, vector< int > break_nodes, int &n_total_segments, int &n_total_nodes, float &cumulative_MLE)
	This function gets the AICc after breaking the channelwith a colinear dataset. More...

vector< float >	calculate_AICc_after_breaks_colinear_monte_carlo (float A_0, float m_over_n, int skip, int minimum_segment_length, float sigma, vector< float > reverse_Chi, vector< float > reverse_Elevation, vector< int > break_nodes, int &n_total_segments, int &n_total_nodes, float &cumulative_MLE, int n_iterations)
	This function gets the AICc after breaking the channelwith a colinear dataset. More...

void	is_channel_long_enough_test (int minimum_segment_length, int N)
	This routine tests to see if channels are long enough to get a decent fitting from the segment finding algorithms. More...

Array2D< float >	calculate_channel_heads (int min_seg_length_for_channel_heads)
	this function fits 2 segments to the chi-elevation data from first order basins and calculates the most likely position of the channel head - added by FC 28/06/13 More...

vector< vector< float > >	get_m_means ()
	This gets the m_means for the channel network. More...

vector< vector< float > >	get_m_standard_deviations ()
	This gets the m_means for the channel network. More...

vector< vector< float > >	get_b_means ()
	This gets the b_means for the channel network. More...

vector< vector< float > >	get_b_standard_deviations ()
	This gets the b_standard deviations for the channel network. More...

vector< vector< int > >	get_node_indices ()
	This gets the node_indices for the channel network. More...

vector< vector< float > >	get_chis ()
	This gets the chi coordinates for the channel network. More...

Protected Attributes
int	NRows
	Number of rows.

int	NCols
	Number of columns.

float	XMinimum
	Minimum X coordinate.

float	YMinimum
	Minimum Y coordinate.

float	DataResolution
	Data resolution.

int	NoDataValue
	No data value.

bool	I_should_calculate_chi
	This boolean lets the routine know if it is to calculate chi.

vector< vector< int > >	node_indices
	Node indices: used in conjunction with other LSD topographic tool objects and not necessary for standalone program.

vector< vector< int > >	row_indices
	Row indices: used in conjunction with other LSD topographic tool objects and not necessary for standalone program.

vector< vector< int > >	col_indices
	Column indices: used in conjunction with other LSD topographic tool objects and not necessary for standalone program.

vector< vector< float > >	elevations
	The elevations along the channels.

vector< vector< float > >	flow_distances
	Flow distances along channels. Used to integrate to arrive at chi.

vector< vector< float > >	drainage_areas
	Drainage areas.

vector< vector< float > >	chis
	The chi values for the channels. This data will be overwritten as m_over_n changes.

vector< int >	node_on_receiver_channel
	This is the node on the reciever channel where the tributary enters the channel. Used to find the downstream chi value of a channel.

vector< int >	receiver_channel
	This is the channel that the tributary enters.

float	m_over_n_for_fitted_data
	This stores the m over n value use to generate the means and standard deviations of the network properties.

float	A_0_for_fitted_data
	This stored the A_0 value.

vector< int >	is_tributary_long_enough
	This vector is the same size as the number of channels and is 1 if the channel analysis n_nodes > 3* minimum_segment_length.

vector< vector< float > >	chi_m_means
	Vector of chi_m means.

vector< vector< float > >	chi_m_standard_deviations
	Vector of chi_m standard deviations.

vector< vector< float > >	chi_m_standard_errors
	Vector of chi_m standard errors.

vector< vector< float > >	chi_b_means
	Vector of chi_b means.

vector< vector< float > >	chi_b_standard_deviations
	Vector of chi_b standard deviations.

vector< vector< float > >	chi_b_standard_errors
	Vector of chi_b standard errors.

vector< vector< float > >	all_fitted_elev_means
	Vector of fitted elevation means.

vector< vector< float > >	all_fitted_elev_standard_deviations
	Vector of fitted elevation standard deviations.

vector< vector< float > >	all_fitted_elev_standard_errors
	Vector of fitted elevation standard errors.

vector< vector< float > >	chi_DW_means
	Vector of Durbin-Watson means.

vector< vector< float > >	chi_DW_standard_deviations
	Vector of Durbin-Watson standard deviations.

vector< vector< float > >	chi_DW_standard_errors
	Vector of Durbin-Watson standard errors.

vector< vector< int > >	n_data_points_used_in_stats
	The parameters are generated using a monte carlo approach and not all nodes will have the same number of data points, so the number of data points is stored.

vector< vector< int > >	break_nodes_vecvec
	This vector holds the vectors containing the node locations of breaks in the segments.

Detailed Description

This object is used to examine a network of channels in chi space.

Constructor & Destructor Documentation

LSDChiNetwork::LSDChiNetwork ( string channel_network_fname )

inline

Crate routine to make a LSDChiNetwork object.

Parameters

channel_network_fname Filename.

Member Function Documentation

float LSDChiNetwork::calculate_AICc_after_breaks	(	float	A_0,
		float	m_over_n,
		int	skip,
		int	minimum_segment_length,
		float	sigma,
		int	chan,
		vector< int >	break_nodes,
		int &	n_total_segments,
		int &	n_total_nodes,
		float &	cumulative_MLE
	)

This function gets the AICc after breaking the channel.

Parameters

A_0
m_over_n
skip
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
chan
break_nodes
n_total_segments
n_total_nodes
cumulative_MLE

Returns: AICc value

Author: SMM

Date: 01/06/13

float LSDChiNetwork::calculate_AICc_after_breaks_colinear	(	float	A_0,
		float	m_over_n,
		int	skip,
		int	minimum_segment_length,
		float	sigma,
		vector< float >	reverse_chi,
		vector< float >	reverse_elevation,
		vector< int >	break_nodes,
		int &	n_total_segments,
		int &	n_total_nodes,
		float &	cumulative_MLE
	)

This function gets the AICc after breaking the channelwith a colinear dataset.

The reverse_chi and reverse_elevation data has to be provided.

Parameters

A_0
m_over_n
skip
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
reverse_chi
reverse_elevation
break_nodes
n_total_segments
n_total_nodes
cumulative_MLE

Returns: AICc value

Author: SMM

Date: 01/06/13

vector< float > LSDChiNetwork::calculate_AICc_after_breaks_colinear_monte_carlo	(	float	A_0,
		float	m_over_n,
		int	skip,
		int	minimum_segment_length,
		float	sigma,
		vector< float >	reverse_Chi,
		vector< float >	reverse_Elevation,
		vector< int >	break_nodes,
		int &	n_total_segments,
		int &	n_total_nodes,
		float &	cumulative_MLE,
		int	n_iterations
	)

This function gets the AICc after breaking the channelwith a colinear dataset.

The reverse_chi and reverse_elevation data has to be provided. It uses a monte carlo scheme and returns a vector with all of the AICc values calcluated from the analyses.

Parameters

A_0
m_over_n
skip
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
reverse_Chi
reverse_Elevation
break_nodes
n_total_segments
n_total_nodes
cumulative_MLE
n_iterations

Returns: AICc value

Author: SMM

Date: 01/06/13

vector< float > LSDChiNetwork::calculate_AICc_after_breaks_monte_carlo	(	float	A_0,
		float	m_over_n,
		int	target_skip,
		int	minimum_segment_length,
		float	sigma,
		int	chan,
		vector< int >	break_nodes,
		int &	n_total_segments,
		int &	n_total_nodes,
		float &	cumulative_MLE,
		int	n_iterations
	)

This function gets the AICc after breaking the channel.

It does this for n_iterations and returns a vector with all of the AICc values for each iteration reported. This can then be used to calculate the statistics of the AICc to tell if the minimum AICc is significantly different from the other AICc values for different values of m/n

Parameters

A_0
m_over_n
target_skip
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
chan
break_nodes
n_total_segments
n_total_nodes
cumulative_MLE
n_iterations

Returns: AICc value

Author: SMM

Date: 01/06/13

Array2D< float > LSDChiNetwork::calculate_channel_heads ( int min_seg_length_for_channel_heads )

this function fits 2 segments to the chi-elevation data from first order basins and calculates the most likely position of the channel head - added by FC 28/06/13

Parameters

min_seg_length_for_channel_heads Minimum number of nodes used for segment fitting

Returns: array with channel head locations

Author: Fiona Clubb

Date: 03/09/2013

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Calculate channel head locations using LSDChiNetwork.

Fitting segments to the chi-elevation data of the main stem. We assume that the profile is made up of 2 segments in chi-space: a linear channel segment and a non-linear hillslope segment. We loop through the possible combinations of segment lengths, performing a linear regression to calculate the r^2 and DW of each segment length. We then calculate a test value: r^2 of the channel segment - ((DW of the hillslope segment - 2)/2). This value will vary between 0 and 1. The maximum test_value will give the best fit channel and hillslope segments. Need to get the best fit m_over_n value and calculate the chi profile of the main stem first. Will output the chi and elevation values of the predicted channel head location. Parameters: min_seg_length_for_channel_heads (length used for fitting segments to the chi- elevation profile, a value of 10 is suggested) Return value: Array with channel head locations FC 03/09/2013

void LSDChiNetwork::calculate_chi	(	float	A_0,
		float	m_over_n
	)

This function calculates the chi values for the channel network using the rectangle rule.

Note: the entire network must be caluculated because the chi values of the tributaries depend on the chi values of the mainstem.

Parameters

A_0	A_0 value.
m_over_n	m over n ratio.

Author: SMM

Date: 01/04/13

float LSDChiNetwork::calculate_optimal_chi_spacing ( int target_nodes )

This function calucaltes the chi spacing of the main stem channel (the longest channel).

The maximum length of the dataset will be in the main stem so this will determine the target spacing of all the tributaries.

Parameters

target_nodes Node index of the target node.

Returns: Optimal chi spacing.

Author: SMM

Date: 01/04/13

int LSDChiNetwork::calculate_skip ( int target_nodes )

This function calucaltes the skip parameter of the main stem (the longest channel).

The maximum length of the dataset will be in the main stem so this will determine the target spacing of all the tributaries.

Parameters

target_nodes Node index of the target node.

Returns: Skip value.

Author: SMM

Date: 01/06/13

int LSDChiNetwork::calculate_skip	(	int	target_nodes,
		vector< float > &	sorted_chis
	)

This function calucaltes the skip parameter based on a vector of chi values.

Parameters

target_nodes	Node index of the target node.
sorted_chis	Vector of chi values

Returns: Skip value.

Author: SMM

Date: 01/06/13

int LSDChiNetwork::calculate_skip	(	int	target_nodes,
		int	channel_number
	)

This function calucaltes the skip parameter of a give channel.

The maximum length of the dataset will be in the main stem so this will determine the target spacing of all the tributaries.

Parameters

target_nodes	Node index of the target node.
channel_number	The channel to be analysed.

Returns: Skip value.

Author: SMM

Date: 01/04/13

Array2D< float > LSDChiNetwork::data_to_array ( int data_member )

Routine for returning calculated data to an array.

It includes a switch that tells the function what data member to write to the array code for data members:

1 elevations
2 chis
3 chi_m_means
4 chi_m_standard_deviations
5 chi_m_standard_errors
6 chi_b_means
7 chi_b_standard_deviations
8 chi_b_standard_errors
9 chi_DW_means
10 chi_DW_standard_deviations
11 chi_DW_standard_errors
12 all_fitted_DW_means
13 all_fitted_DW_standard_deviations
14 all_fitted_DW_standard_errors

Parameters

data_member Switch to select data to be written.

Returns: Array of data.

Author: SMM

Date: 01/04/13

void LSDChiNetwork::extend_tributaries_to_outlet ( )

This extends the tributary channels all the way to the outlet.

In its current version this only works if the tributaries all drain to the mainstem.

Author: SMM

Date: 01/04/13

void LSDChiNetwork::find_most_likeley_segments	(	int	channel,
		int	minimum_segment_length,
		float	sigma,
		int	N,
		vector< float > &	b_vec,
		vector< float > &	m_vec,
		vector< float > &	r2_vec,
		vector< float > &	DW_vec,
		vector< float > &	thinned_chi,
		vector< float > &	thinned_elev,
		vector< float > &	fitted_elev,
		vector< int > &	node_reference,
		vector< int > &	these_segment_lengths,
		float &	this_MLE,
		int &	this_n_segments,
		int &	n_data_nodes,
		float &	this_AIC,
		float &	this_AICc
	)

This function gets the most likely channel segments for a particular channel.

This function replaces the b, m, r2 and DW values of each segment into vectors it also returns the fitted elevation and the index into the original channel (since this is done with thinned data).

Parameters

channel	The index into the channel.
minimum_segment_length	is how many nodes the mimimum segment will have.
sigma	is the standard deviation of error on elevation data
N
b_vec
m_vec
r2_vec
DW_vec
thinned_chi
thinned_elev
fitted_elev
node_reference
these_segment_lengths
this_MLE
this_n_segments
n_data_nodes
this_AIC
this_AICc

Author: SMM

Date: 01/04/13

void LSDChiNetwork::find_most_likeley_segments_dchi	(	int	channel,
		int	minimum_segment_length,
		float	sigma,
		float	dchi,
		vector< float > &	b_vec,
		vector< float > &	m_vec,
		vector< float > &	r2_vec,
		vector< float > &	DW_vec,
		vector< float > &	thinned_chi,
		vector< float > &	thinned_elev,
		vector< float > &	fitted_elev,
		vector< int > &	node_reference,
		vector< int > &	these_segment_lengths,
		float &	this_MLE,
		int &	this_n_segments,
		int &	n_data_nodes,
		float &	this_AIC,
		float &	this_AICc
	)

This function gets the most likely channel segments for a particular channel.

This function replaces the b, m, r2 and DW values of each segment into vectors it also returns the fitted elevation and the index into the original channel (since this is done with thinned data).

Parameters

channel	The index into the channel.
minimum_segment_length	is how many nodes the mimimum segment will have.
sigma	is the standard deviation of error on elevation data
dchi
b_vec
m_vec
r2_vec
DW_vec
thinned_chi
thinned_elev
fitted_elev
node_reference
these_segment_lengths
this_MLE
this_n_segments
n_data_nodes
this_AIC
this_AICc

Author: SMM

Date: 01/04/13

void LSDChiNetwork::find_most_likeley_segments_monte_carlo	(	int	channel,
		int	minimum_segment_length,
		float	sigma,
		int	mean_skip,
		int	skip_range,
		vector< float > &	b_vec,
		vector< float > &	m_vec,
		vector< float > &	r2_vec,
		vector< float > &	DW_vec,
		vector< float > &	thinned_chi,
		vector< float > &	thinned_elev,
		vector< float > &	fitted_elev,
		vector< int > &	node_reference,
		vector< int > &	these_segment_lengths,
		float &	this_MLE,
		int &	this_n_segments,
		int &	n_data_nodes,
		float &	this_AIC,
		float &	this_AICc
	)

This gets the most likely segments but uses the monte carlo data thinning method.

The expectation is that this will be used repeatedly on channels to generate statistics of the best fit segments by individual nodes in the channel network.

Parameters

channel	The index into the channel.
minimum_segment_length	is how many nodes the mimimum segment will have.
sigma	is the standard deviation of error on elevation data
mean_skip
skip_range
b_vec
m_vec
r2_vec
DW_vec
thinned_chi
thinned_elev
fitted_elev
node_reference
these_segment_lengths
this_MLE
this_n_segments
n_data_nodes
this_AIC
this_AICc

Author: SMM

Date: 01/04/13

void LSDChiNetwork::find_most_likeley_segments_monte_carlo_dchi	(	int	channel,
		int	minimum_segment_length,
		float	sigma,
		float	mean_dchi,
		float	variation_dchi,
		vector< float > &	b_vec,
		vector< float > &	m_vec,
		vector< float > &	r2_vec,
		vector< float > &	DW_vec,
		vector< float > &	thinned_chi,
		vector< float > &	thinned_elev,
		vector< float > &	fitted_elev,
		vector< int > &	node_reference,
		vector< int > &	these_segment_lengths,
		float &	this_MLE,
		int &	this_n_segments,
		int &	n_data_nodes,
		float &	this_AIC,
		float &	this_AICc
	)

This gets the most likely segments but uses the monte carlo data thinning method.

The mean_dchi is the mean value of dchi, and the variation is how much the chi chan vary, such that minimum_dchi = dchi-variation_dchi. The expectation is that this will be used repeatedly on channels to generate statistics of the best fit segments by individual nodes in the channel network.

Parameters

channel	The index into the channel.
minimum_segment_length	is how many nodes the mimimum segment will have.
sigma	is the standard deviation of error on elevation data
mean_dchi
variation_dchi
b_vec
m_vec
r2_vec
DW_vec
thinned_chi
thinned_elev
fitted_elev
node_reference
these_segment_lengths
this_MLE
this_n_segments
n_data_nodes
this_AIC
this_AICc

Author: SMM

Date: 01/04/13

vector< vector<float> > LSDChiNetwork::get_b_means ( )

inline

This gets the b_means for the channel network.

Returns: vector of vectors with b means @ author SMM

Date: 24/05/16

vector< vector<float> > LSDChiNetwork::get_b_standard_deviations ( )

inline

This gets the b_standard deviations for the channel network.

Returns: vector of vectors with b tandard deviations @ author SMM

Date: 24/05/16

vector< vector<float> > LSDChiNetwork::get_chis ( )

inline

This gets the chi coordinates for the channel network.

Returns: vector of vectors with m means @ author DTM

Date: 24/03/15

float LSDChiNetwork::get_DataResolution ( ) const

inline

Returns: Data resolution as an integer.

vector< vector< float > > LSDChiNetwork::get_m_means ( )

This gets the m_means for the channel network.

Returns: vector of vectors with m means @ author FJC

Date: 04/08/14

vector< vector<float> > LSDChiNetwork::get_m_standard_deviations ( )

inline

This gets the m_means for the channel network.

Returns: vector of vectors with m means @ author FJC

Date: 04/08/14

int LSDChiNetwork::get_n_channels ( )

inline

Returns: Number of channels.

int LSDChiNetwork::get_NCols ( ) const

inline

Returns: Number of columns as an integer.

int LSDChiNetwork::get_NoDataValue ( ) const

inline

Returns: No Data Value as an integer.

vector< vector<int> > LSDChiNetwork::get_node_indices ( )

inline

This gets the node_indices for the channel network.

Returns: vector of vectors with m means @ author DTM

Date: 24/03/15

int LSDChiNetwork::get_NRows ( ) const

inline

Returns: Number of rows as an integer.

float LSDChiNetwork::get_XMinimum ( ) const

inline

Returns: Minimum X coordinate as an integer.

float LSDChiNetwork::get_YMinimum ( ) const

inline

Returns: Minimum Y coordinate as an integer.

void LSDChiNetwork::is_channel_long_enough_test	(	int	minimum_segment_length,
		int	N
	)

This routine tests to see if channels are long enough to get a decent fitting from the segment finding algorithms.

Writes to the is_tributary_long_enough vector. If this equals 1, the channel is long enough. If it is zero the channel is not long enough.

Parameters

minimum_segment_length	How many nodes the mimimum segment will have.
N

void LSDChiNetwork::monte_carlo_sample_river_network_for_best_fit	(	float	A_0,
		float	m_over_n,
		int	n_iterations,
		int	mean_skip,
		int	skip_range,
		int	minimum_segment_length,
		float	sigma
	)

Monte carlo segment fitter.

This takes a fixed m_over_n value and then samples the indivudal nodes in the full channel profile to repeadetly get the best fit segments on thinned data. the m, b fitted elevation, r^2 and DW statistic are all stored for every iteration on every node. These can then be queried later for mean, standard deviation and standard error information

the fraction_dchi_for_variation is the fration of the optimal dchi that dchi can vary over. So for example if this = 0.4 then the variation of dchi will be 0.4*mean_dchi and the minimum dchi will be min_dchi = (1-0.4)*mean_dchi

Note: this is extremely computationally and data intensive.

Parameters

A_0
m_over_n
n_iterations
mean_skip
skip_range
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data

Author: SMM

Date: 01/04/13

void LSDChiNetwork::monte_carlo_sample_river_network_for_best_fit_after_breaks	(	float	A_0,
		float	m_over_n,
		int	n_iterations,
		int	skip,
		int	minimum_segment_length,
		float	sigma
	)

This function samples the river network using monte carlo samplig but after breaking the channels.

Parameters

A_0
m_over_n
n_iterations
skip
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data

Author: SMM

Date: 01/04/13

void LSDChiNetwork::monte_carlo_sample_river_network_for_best_fit_dchi	(	float	A_0,
		float	m_over_n,
		int	n_iterations,
		float	fraction_dchi_for_variation,
		int	minimum_segment_length,
		float	sigma,
		int	target_nodes_mainstem
	)

This routine uses a monte carlo approach to repeatedly sampling all the data in the channel network.

Uses a reduced number of data elements and then populates each channel node with a distribution of m, b and fitted elevation values. These then can be averaged and details of their variation calculated.
This is based on a fixed value of dchi

Parameters

A_0
m_over_n
n_iterations
fraction_dchi_for_variation
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
target_nodes_mainstem

Author: SMM

Date: 01/04/13

void LSDChiNetwork::monte_carlo_split_channel	(	float	A_0,
		float	m_over_n,
		int	n_iterations,
		int	target_skip,
		int	target_nodes,
		int	minimum_segment_length,
		float	sigma,
		int	chan,
		vector< int > &	break_nodes
	)

Monte carlo segment fitter.

This takes a fixed m_over_n value and then samples the indivudal nodes in the full channel profile to repeadetly get the best fit segments on thinned data. the m, b fitted elevation, r^2 and DW statistic are all stored for every iteration on every node. These can then be queried later for mean, standard deviation and standard error information.

The break nodes vector tells the algorithm where the breaks in the channel occur this function is called repeatedly until the target skip equals the all of the this_skip values.

This function continues to split the channel into segments until the target skip is achieved.

Parameters

A_0
m_over_n
n_iterations
target_skip
target_nodes
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
chan
break_nodes

Author: SMM

Date: 01/04/13

void LSDChiNetwork::monte_carlo_split_channel_colinear	(	float	A_0,
		float	m_over_n,
		int	n_iterations,
		int	target_skip,
		int	target_nodes,
		int	minimum_segment_length,
		float	sigma,
		vector< float >	reverse_Chi,
		vector< float >	reverse_Elevation,
		vector< int > &	break_nodes
	)

This function uses a monte carlo sampling approach to try and split channels.

The channel is sampled at the target skipping interval. It does it with a colinear dataset.

Parameters

A_0
m_over_n
n_iterations
target_skip
target_nodes
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
reverse_Chi
break_nodes
reverse_Elevation
break_nodes

Author: SMM

Date: 01/04/13

void LSDChiNetwork::print_channel_details_to_file	(	string	fname,
		float	A_0,
		float	m_over_n
	)

Print channel details to file for bug checking.

Format of file:

channel_number << " " << receiver_channel[channel_number] << " " << node_on_receiver_channel[channel_number] << " " << node[i] << " " << row[i] << " " << col[i] << " " << flow_distance[i] << " " << chi[i] << " " << elevation[i] << " " << drainage_area[i]

Parameters

fname	Output filename.
A_0	A_0 value.
m_over_n	m over n ratio.

Author: SMM

Date: 01/04/13

void LSDChiNetwork::print_channel_details_to_file_full_fitted ( string fname )

This function prints the details of all channels to a file.

It includes data from monte carlo fitting. Format is:

A_0 m_over_n channel_number node_on_receiver_channel node_index row col flow_distance chi elevation darainage_area.

Parameters

fname Output filename.

Author: SMM

Date: 01/04/13

void LSDChiNetwork::print_channel_details_to_file_full_fitted	(	string	fname,
		int	target_nodes,
		int	minimum_segment_length
	)

This function prints the details of all channels to a file.

It includes data from monte carlo fitting. Format is:

A_0 m_over_n channel_number node_on_receiver_channel node_index row col flow_distance chi elevation darainage_area.

Parameters

fname	Output filename.
target_nodes
minimum_segment_length

Author: SMM

Date: 01/04/13

void LSDChiNetwork::print_channel_details_to_file_full_fitted_for_ArcMap ( string fname )

This function prints the details of all channels to a csv file that can be ingested by ArcMap.

It includes data from monte carlo fitting.

Parameters

fname Output filename, which has _for_Arc automatically appended.

Author: SMM

Date: 28/02/14

void LSDChiNetwork::print_channel_details_to_screen ( int channel_number )

Print channel details to screen for bug checking.

Format of file:

channel_number << " " << receiver_channel[channel_number] << " " << node_on_receiver_channel[channel_number] << " " << node[i] << " " << row[i] << " " << col[i] << " " << flow_distance[i] << " " << chi[i] << " " << elevation[i] << " " << drainage_area[i]

Parameters

channel_number Channel to examine

Author: SMM

Date: 01/04/13

float LSDChiNetwork::search_for_best_fit_m_over_n	(	float	A_0,
		int	n_movern,
		float	d_movern,
		float	start_movern,
		int	minimum_segment_length,
		float	sigma,
		int	target_nodes_mainstem,
		string	fname
	)

The master routine for calculating the best fit m over n values for a channel network.

Parameters

A_0
n_movern
d_movern
start_movern
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
target_nodes_mainstem
fname	Output filename

Returns: Best fit m over n.

Author: SMM

Date: 01/04/13

float LSDChiNetwork::search_for_best_fit_m_over_n_colinearity_test	(	float	A_0,
		int	n_movern,
		float	d_movern,
		float	start_movern,
		int	minimum_segment_length,
		float	sigma,
		int	target_nodes,
		int	n_iterations,
		vector< float > &	m_over_n_values,
		vector< float > &	AICc_mean,
		vector< float > &	AICc_sdtd
	)

This function looks for the best fit values of m over n by simply testing for the least variation in the tributaries.

Parameters

A_0
n_movern
d_movern
start_movern
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
target_nodes
n_iterations
m_over_n_values
AICc_mean
AICc_sdtd

Returns: Best fit m over n.

Author: SMM

Date: 01/04/13

float LSDChiNetwork::search_for_best_fit_m_over_n_colinearity_test_with_breaks	(	float	A_0,
		int	n_movern,
		float	d_movern,
		float	start_movern,
		int	minimum_segment_length,
		float	sigma,
		int	target_skip,
		int	target_nodes,
		int	n_iterations,
		vector< float > &	m_over_n_values,
		vector< float > &	AICc_mean,
		vector< float > &	AICc_sdtd,
		int	Monte_Carlo_switch
	)

This function calculeates best fit m/n using the collinearity test
these channels are ones with breaks.

Parameters

A_0	float the reference area
n_movern	int the number of m over n ratios to iterate through
d_movern	float the change in m/n in each iterations
start_movern	float the starting value of m/n
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
target_skip	int the mean skipping value
target_nodes	int the target number of nodes in a break
n_iterations	int the number of iterations
m_over_n_values	vector<float>& this gets written, it contains the m/n values for the run
AICc_mean	vector<float> gets written, the mean values of the AICc for each m/n
AICc_sdtd	vector<float> gets written, the standard deviation values of the AICc for each m/n
Monte_Carlo_switch	int if 1, run the code using the iterative Monte Carlo scheme

Returns: Best fit m over n.

Author: SMM

Date: 01/07/13

for (int i = 0; i< int(movn_values.size()); i++)

float LSDChiNetwork::search_for_best_fit_m_over_n_dchi	(	float	A_0,
		int	n_movern,
		float	d_movern,
		float	start_movern,
		int	minimum_segment_length,
		float	sigma,
		int	target_nodes_mainstem,
		string	fname
	)

The master routine for calculating the best fit m over n values for a channel network, based on a fixed value of dchi.

Parameters

A_0
n_movern
d_movern
start_movern
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
target_nodes_mainstem
fname	Output filename

Returns: Best fit m over n.

Author: SMM

Date: 01/04/13

float LSDChiNetwork::search_for_best_fit_m_over_n_individual_channels_with_breaks	(	float	A_0,
		int	n_movern,
		float	d_movern,
		float	start_movern,
		int	minimum_segment_length,
		float	sigma,
		int	target_skip,
		int	target_nodes,
		int	n_iterations,
		vector< float > &	m_over_n_values,
		vector< vector< float > > &	AICc_vals
	)

This function calculeates best fit m/n for each channel these channels are ones with breaks.

Parameters

A_0	float the reference area

this does not report variability of the AICc values and so should not be used, instead
use the Monte Carlo version
retained in case you want rapid calculation of best fit m/n

Parameters

n_movern	int the number of m over n ratios to iterate through
d_movern	float the change in m/n in each iterations
start_movern	float the starting value of m/n
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
target_skip	int the mean skipping value
target_nodes	int the target number of nodes in a break
n_iterations	int the number of iterations
m_over_n_values	vector<float>& this gets written, it contains the m/n values for the run
AICc_vals

Returns: Best fit m over n.

Author: SMM

Date: 01/04/13

float LSDChiNetwork::search_for_best_fit_m_over_n_individual_channels_with_breaks_monte_carlo	(	float	A_0,
		int	n_movern,
		float	d_movern,
		float	start_movern,
		int	minimum_segment_length,
		float	sigma,
		int	target_skip,
		int	target_nodes,
		int	n_iterations,
		vector< float > &	m_over_n_values,
		vector< vector< float > > &	AICc_means,
		vector< vector< float > > &	AICc_stddev
	)

This gets the best fit m over n values of all the individual tributaries.

It uses a monte carlo appraoach so all tributaries have both the mean and the variability of the AICc values reported.

Parameters

A_0	float the reference area
n_movern	int the number of m over n ratios to iterate through
d_movern	float the change in m/n in each iterations
start_movern	float the starting value of m/n
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
target_skip	int the mean skipping value
target_nodes	int the target number of nodes in a break
n_iterations	int the number of iterations
m_over_n_values	vector<float>& this gets written, it contains the m/n values for the run
AICc_means	vector<float> gets written, the mean values of the AICc for each m/n
AICc_stddev	vector<float> gets written, the standard deviation values of the AICc for each m/n

Returns: Best fit m over n.

Author: SMM

Date: 01/07/13

float LSDChiNetwork::search_for_best_fit_m_over_n_seperate_ms_and_tribs	(	float	A_0,
		int	n_movern,
		float	d_movern,
		float	start_movern,
		int	minimum_segment_length,
		float	sigma,
		int	target_nodes_mainstem,
		string	fname
	)

Routine for calculating the best fit m over n values for a channel network, but calculates the mainstem and the tributaries seperately.

Parameters

A_0
n_movern
d_movern
start_movern
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data
target_nodes_mainstem
fname	Output filename

Returns: Best fit m over n.

Author: SMM

Date: 01/04/13

void LSDChiNetwork::slope_area_extraction_horizontal_intervals	(	float	interval,
		float	area_thin_fraction,
		string	fname
	)

Extract slope over fixed horizontal intervals.

This one is the horizontal intervals version: it measures slope over fixed flow distance as used by many authors including DiBiasie et al 2010 and Ouimet et al 2009 This function gets slope data and area data for use in making slope area plots. It generates several data elements, which are written to the file with name fname (passed to function). The file format is for each row:

chan << " " << start_row << " " << mp_row << " " << end_row << " " << start_col << " " << mp_col << " " << end_col << " " << start_interval_elevations << " " << mp_interval_elevations << " " << end_interval_elevations << " " << start_interval_flowdistance << " " << mp_interval_flowdistance << " " << end_interval_flowdistance << " " << start_area << " " << mp_area << " " << end_area << " " << slope << " " << log10(mp_area) << " " << log10(slope)

where start, mp and end denote the start of the interval over which slope is measured, the midpoint and the end.

The area thin fraction is used to thin the data so that segments with large changes in drainage area are not used in the regression (because these will affect the mean slope) the fraction is determined by (downslope_area-upslope_area)/midpoint_area. So if the fraction is 1 it means that the change is area is equal to the area at the midpoint a restictive value is 0.05, you will eliminate major tributaries with a 0.2, and 1 will catch almost all of the data.

Parameters

interval
area_thin_fraction
fname	Output filename

Author: SMM

Date: 01/04/13

void LSDChiNetwork::slope_area_extraction_vertical_intervals	(	float	interval,
		float	area_thin_fraction,
		string	fname
	)

Extract slope over fixed vertical intervals.

This one is the vertical intervals version: it measures slope over fixed vertical intervals as reccomended by Wobus et al 2006. This function gets slope data and area data for use in making slope area plots. It generates several data elements, which are written to the file with name fname (passed to function). The file format is for each row:

chan << " " << start_row << " " << mp_row << " " << end_row << " " << start_col << " " << mp_col << " " << end_col << " " << start_interval_elevations << " " << mp_interval_elevations << " " << end_interval_elevations << " " << start_interval_flowdistance << " " << mp_interval_flowdistance << " " << end_interval_flowdistance << " " << start_area << " " << mp_area << " " << end_area << " " << slope << " " << log10(mp_area) << " " << log10(slope)

where start, mp and end denote the start of the interval over which slope is measured, the midpoint and the end.

The area thin fraction is used to thin the data so that segments with large changes in drainage area are not used in the regression (because these will affect the mean slope) the fraction is determined by (downslope_area-upslope_area)/midpoint_area. So if the fraction is 1 it means that the change is area is equal to the area at the midpoint a restictive value is 0.05, you will eliminate major tributaries with a 0.2, and 1 will catch almost all of the data.

Parameters

interval
area_thin_fraction
fname	Output filename

Author: SMM

Date: 01/04/13

void LSDChiNetwork::split_all_channels	(	float	A_0,
		float	m_over_n,
		int	n_iterations,
		int	target_skip,
		int	target_nodes,
		int	minimum_segment_length,
		float	sigma
	)

This function splits all the channels in one go.

Parameters

A_0
m_over_n
n_iterations
target_skip
target_nodes
minimum_segment_length	How many nodes the mimimum segment will have.
sigma	Standard deviation of error on elevation data

Author: SMM

Date: 01/06/13

The documentation for this class was generated from the following files:

LSDChiNetwork.hpp
LSDChiNetwork.cpp

Public Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation