## Reading model output

Reading model output is done with the *xb_read_output* function. The function is capable of reading both netCDF as DAT output. Any file ending with *.nc* is supposed to be a XBeach netCDF output file. Any file enign with *.dat* or any directory is supposed to be or contain one or more XBeach DAT output files.

% reading DAT files xbo = xb_read_output('path_to_model/'); xbo = xb_read_output('path_to_model/zb.dat'); % reading netCDF files xbo = xb_read_output('path_to_model/xboutput.nc');

It is also possible to supply the result structure from either the *xb_run* or *xb_run_remote* function that contains information on the location where the model can be found on disk.

xbo = xb_read_output(xbr);

## Narrowing the amount of data read

For large models, the amount of data read from the output files can become very large. In order to cope with the large amounts of data, the amount of data read can be narrowed down to a specific part. Narrowing down the data can be done in two ways: limiting the number of variables read or limiting the amount of data read for each variable.

% limit the number of variables read from directory with DAT files xbo = xb_read_output('path_to_model/', 'vars', {'zb', 'zs', 'H'}); % limit the number of variables read from netCDF file xbo = xb_read_output('path_to_model/xboutput.nc', 'vars', {'zb', 'zs', 'H'}); % use asterix filtering to limit the number of parameters read xbo = xb_read_output('path_to_model/', 'vars', {'z*', 'H', '*_mean'}); % use regular expression filtering to limit the number of parameters read % (don't forget the starting slash) xbo = xb_read_output('path_to_model/', 'vars', '/_(min|max|mean)$'); xbo = xb_read_output('path_to_model/', 'vars', {'/^z', '/g$'});

Limiting the amount of data read per variable is done using starting indices, lengths and strides. This way of filtering is common to netCDF files and also implemented in the DAT read functions in the XBeach Toolbox.

The filtering works using at most three vectors (start, length and stride) with each a length of at most the number of dimensions in the variable to be read. Thus, a 3D variable has vectors of length equal to or smaller than 3. If the length is smaller than the number of dimensions, the filtering is assigned to the first few dimensions. The other dimensions are kept unfiltered.

The start indices are zero-based and indicate which part of the data at the beginning of a particular dimension should be skipped. So, with zero nothing is skipped, with one only the first item, with ten the first ten items, etcetera.

The lengths indicate the number of items to be read counting from the starting index. The strides indicates the read resolution of the data to be read. So, specifying a length 99 and a stride 3 means that 33 items are read.

For example:

% using start, length and stride options to limit the amount of data read % from a netCDF file xbo = xb_read_output('path_to_model/xboutput.nc', ... 'vars', {'z*', 'H'}, ... 'start', [99 10 0], ... 'length', [1 10 -1], ... 'stride', [1 2 5] ... ); % in the XBeach Toolbox limiting the amount of data read from DAT files is % implemented according to the netCDF standards xbo = xb_read_output('path_to_model/', ... 'vars', {'z*', 'H'}, ... 'start', [99 10 0], ... 'length', [1 10 -1], ... 'stride', [1 2 5] ... );

In the two examples above, the 100st item is read from the first dimension (time), the 10th to 20th items using a stride of 2 are read from the second dimension (y) and all data is read from the third dimension (x) using a stride of 5. Consequently, the matrices read will have a size of [1 5 20] assuming the third dimension to have a length of 100.

## Swapping your data dimensions

In the model output DAT files the first two dimensions are x and y. The last dimension is time. In case there are more than three dimensions, they are located in between y and time. Following netCDF conventions, the dimension order in model output netCDF files is different. The first three dimensions are time, y and x respectively. All other dimensions are located thereafter. It is decided that the XBeach Toolbox is designed for the time, y, x dimension order. Consequently, reading DAT files using the XBeach Toolbox will return matrices in this order as well. To facilitate the use of other, older functions, the dimension order of an XBeach output structure can be swapped:

% swap dimension from new to old standard ... xbo = xb_swap(xbo); % ... and back xbo = xb_swap(xbo); % force swapping from new to old standard ... xbo = xb_swap(xbo, 'order', 'tyx', 'force', true); % ... and back xbo = xb_swap(xbo, 'order', 'xyt', 'force', true);