MATLAB - déterminer une trajectoire à l'aide de séries temporelles XY et vecteur de caractéristiques - Stack Overflow

I have a time-series of XY coordinates and attributes. I need to determine a trajectory T, using vector characteristics that match certain attributes of the XY time-series. The dataset (https://www.dropbox.com/s/hnsaaqc3wwq5wpo/summer_disp_model.csv) shows that the T vector corresponding to record #2 has a direction t_phi, and magnitude t_vel and should be summed to the vector of the previous record. Starting point is the lat, lon coordinates of record #1, then at each time-step (delta_ts) the vector is summed to the previous one, resulting in a trajectory that’s different from the track of the XY timeseries.

I want to use a Monte-Carlo simulation to account for the variance (95% upper and lower confidence intervals) in direction and magnitude of the T vectors and hope to obtain something akin to a predicted hurricane path.

A key assumption is that all currents in the area covered by the XY track behave the same way. The trajectory T is independent from XY and is a sum of vectors with corresponding confidence intervals.

The vector characteristics of T, which reflect animal behavior in different current scenarios (direction and velocity of XY, given by b_cat and v_cat, resp.), change when XY vector characteristics change (in real life, in a W-flowing current the angle/direction of dispersal of the animal, which always keeps swimming North –hardwired by evolution- tilts towards positive values as the velocity of the current increases, while its velocity approaches that of the current).

When the direction of XY approaches values that are opposed to the swimming direction of the animal (N), the velocity of T (t_vel) approaches 0 and the XY vectors (given by cur_v and cur_dir) should be summed instead of the T vectors given by t_phi and t_vel (in real life, in a SE-flowing current > 0.2 m.s, the N-swimming animal can’t deviate from the XY track, so the T vector=XY vector). In this case there is no variance.

Here is a summary of the parameters given in the .csv file:

rec: record number;
latXY: latitude;
lonXY: longitude;
cur_v: velocity of XY (in m.s);
cur_dir: direction of XY; 
v_cat: velocity category of XY;
b_cat: direction category of XY;
delta_ts: time (s) elapsed between successive records;
t_phi: angle of T vector for the corresponding v_cat and b_cat;
phi_up: upper 95% confidence Interval;
phi_low: lower 95% confid interval;
t_vel: velocity (in m.s) of T vector for the corresponding v_cat and b_cat;
vel_low: lower 95% confidence interval;
vel_up: upper 95% confidence interval;

Directions/angles are in degrees, on a 180 degree scale (0=N, 90=E, +/- 180=S, -90=W).

Any idea on how to approach this in Matlab?

Thanks! Corto

I have a time-series of XY coordinates and attributes. I need to determine a trajectory T, using vector characteristics that match certain attributes of the XY time-series. The dataset (https://www.dropbox.com/s/hnsaaqc3wwq5wpo/summer_disp_model.csv) shows that the T vector corresponding to record #2 has a direction t_phi, and magnitude t_vel and should be summed to the vector of the previous record. Starting point is the lat, lon coordinates of record #1, then at each time-step (delta_ts) the vector is summed to the previous one, resulting in a trajectory that’s different from the track of the XY timeseries.

I want to use a Monte-Carlo simulation to account for the variance (95% upper and lower confidence intervals) in direction and magnitude of the T vectors and hope to obtain something akin to a predicted hurricane path.

A key assumption is that all currents in the area covered by the XY track behave the same way. The trajectory T is independent from XY and is a sum of vectors with corresponding confidence intervals.

The vector characteristics of T, which reflect animal behavior in different current scenarios (direction and velocity of XY, given by b_cat and v_cat, resp.), change when XY vector characteristics change (in real life, in a W-flowing current the angle/direction of dispersal of the animal, which always keeps swimming North –hardwired by evolution- tilts towards positive values as the velocity of the current increases, while its velocity approaches that of the current).

When the direction of XY approaches values that are opposed to the swimming direction of the animal (N), the velocity of T (t_vel) approaches 0 and the XY vectors (given by cur_v and cur_dir) should be summed instead of the T vectors given by t_phi and t_vel (in real life, in a SE-flowing current > 0.2 m.s, the N-swimming animal can’t deviate from the XY track, so the T vector=XY vector). In this case there is no variance.

Here is a summary of the parameters given in the .csv file:

rec: record number;
latXY: latitude;
lonXY: longitude;
cur_v: velocity of XY (in m.s);
cur_dir: direction of XY; 
v_cat: velocity category of XY;
b_cat: direction category of XY;
delta_ts: time (s) elapsed between successive records;
t_phi: angle of T vector for the corresponding v_cat and b_cat;
phi_up: upper 95% confidence Interval;
phi_low: lower 95% confid interval;
t_vel: velocity (in m.s) of T vector for the corresponding v_cat and b_cat;
vel_low: lower 95% confidence interval;
vel_up: upper 95% confidence interval;

Directions/angles are in degrees, on a 180 degree scale (0=N, 90=E, +/- 180=S, -90=W).

Any idea on how to approach this in Matlab?

Thanks! Corto

Source