diff --git a/src/data/beach_orientations.m b/src/data/beach_orientations.m
new file mode 100644
index 0000000..d537ad6
--- /dev/null
+++ b/src/data/beach_orientations.m
@@ -0,0 +1,73 @@
+% Calculate orientation the beach profile at each unique site and save to .mat file
+% Needs the following coastal tools:
+% J:\Coastal\Tools\MALT Logspiral Transformation
+% J:\Coastal\Tools\Coordinate Transformations
+
+clear 
+clc
+
+% Load profile data, this is where we want to calculate orientations.
+warning('off','all')
+data = load('C:\Users\z5189959\Desktop\nsw_2016_storm_impact\data\raw\processed_shorelines\profiles.mat');
+data = data.data;
+
+output = [];
+
+for ii = 1:n
+  disp(num2str(ii))
+  lat = data(ii).lat;
+  lon = data(ii).lon;
+  beach = data(ii).site{1};
+  [x,y,utmzone] = deg2utm(lat,lon);
+
+  if strcmp(beach, 'BOOM') == 1 || strcmp(beach, 'HARGn') == 1 || strcmp(beach, 'BILG') == 1 || strcmp(beach, 'HARGs') == 1 || strcmp(beach, 'DEEWHYn') == 1
+    % log spiral transformation file is out of date. Talk to Mitch
+    continue
+  end
+    
+  if strcmp(beach, 'AVOCAs') == 1
+    % negative solution. Talk to Mitch
+    continue
+  end
+  
+  % Get the sp log spiral transformed coordinates
+  xyz.x = x;
+  xyz.y = y;
+  xyz.z = 0;
+  out = xyz2spz(xyz,beach);
+
+  % Define new bounds towards land and towards sea
+  % Distance is the number of m in the section to go towards
+  distance = 200;
+  s = out.s;
+  p_center = out.p;
+
+  % Convert landwards and seawrds point back to lat/lon
+  spz_land.s = out.s;
+  spz_land.p = out.p-distance;
+  spz_land.z = 0;
+  xyz_land = spz2xyz(spz_land,beach);
+  [lat_land,lon_land] = utm2deg(xyz_land.x,xyz_land.y,utmzone);
+
+  spz_sea.s = out.s;
+  spz_sea.p = out.p+distance;
+  spz_sea.z = 0;
+  xyz_sea = spz2xyz(spz_sea,beach);
+  [lat_sea,lon_sea] = utm2deg(xyz_sea.x,xyz_sea.y,utmzone);
+
+  % Orientation in degrees anticlockwise from east, pointing towards land
+  orientation = radtodeg(atan2((xyz_land.y - xyz_sea.y), (xyz_land.x - xyz_sea.x)));
+  
+  row.lat_center = lat;
+  row.lon_center = lon;
+  row.lat_land = lat_land;
+  row.lon_land = lat_land;
+  row.lat_sea = lat_sea;
+  row.lon_sea = lon_sea;
+  row.orientation = orientation;
+  row.beach = beach;
+  output = [output; row];
+end
+warning('on','all')
+
+save('orientations.mat','output','-v7')
diff --git a/src/data/mat_to_csv.py b/src/data/mat_to_csv.py
index 6643d35..2bbf471 100644
--- a/src/data/mat_to_csv.py
+++ b/src/data/mat_to_csv.py
@@ -12,6 +12,50 @@ logging.config.fileConfig('../logging.conf', disable_existing_loggers=False)
 logger = logging.getLogger(__name__)
 
 
+def parse_orientations(orientations_mat):
+    """
+    Parses the raw orientations.mat file and returns a pandas dataframe. Note that orientations are the direction
+    towards land measured in degrees anti-clockwise from east.
+    :param orientations_mat:
+    :return:
+    """
+    logger.info('Parsing %s', orientations_mat)
+    mat_data = loadmat(orientations_mat)['output']
+    rows = []
+    for i in range(0, len(mat_data['beach'])):
+        rows.append({
+            'beach': mat_data['beach'][i],
+            'orientation': mat_data['orientation'][i],
+            'lat_center': mat_data['lat_center'][i],
+            'lon_center': mat_data['lon_center'][i],
+            'lat_land': mat_data['lat_land'][i],
+            'lon_land': mat_data['lon_land'][i],
+            'lat_sea': mat_data['lat_sea'][i],
+            'lon_sea': mat_data['lon_sea'][i],
+        })
+
+    df = pd.DataFrame(rows)
+    return df
+
+def combine_sites_and_orientaions(df_sites, df_orientations):
+    """
+    Replaces beach/lat/lon columns with the unique site_id
+    :param dfs:
+    :param df_sites:
+    :return:
+    """
+
+    df_merged_sites = df_sites.merge(df_orientations[['beach', 'lat_center', 'lon_center', 'orientation']],
+                                      left_on=['beach', 'lat', 'lon'],
+                                       right_on=['beach', 'lat_center', 'lon_center'])
+
+    # Check that all our records have a unique site identifier
+    n_unmatched = len(df_sites) - len(df_merged_sites)
+    if n_unmatched > 0:
+        logger.warning('Not all records (%d of %d) matched with an orientation', n_unmatched, len(df_sites))
+
+    return df_merged_sites
+
 def parse_waves(waves_mat):
     """
     Parses the raw waves.mat file and returns a pandas dataframe
@@ -156,12 +200,16 @@ def main():
     df_tides = parse_tides(tides_mat='../../data/raw/tides.mat')
     df_profiles = parse_profiles(profiles_mat='../../data/raw/profiles.mat')
     df_sites = get_unique_sites(dfs=[df_waves, df_tides, df_profiles])
+    df_orientations = parse_orientations(orientations_mat='./data/raw/processed_shorelines/orientations.mat')
 
     logger.info('Identifying unique sites')
     df_waves = replace_unique_sites(df_waves, df_sites)
     df_tides = replace_unique_sites(df_tides, df_sites)
     df_profiles = replace_unique_sites(df_profiles, df_sites)
 
+    logger.info('Combine orientations into sites')
+    df_sites = combine_sites_and_orientaions(df_sites, df_orientations)
+
     logger.info('Setting pandas index')
     df_profiles.set_index(['site_id', 'profile_type', 'x'], inplace=True)
     df_waves.set_index(['site_id', 'datetime'], inplace=True)