OSM rivers to one main branch Python

Introduction

• 

  The river map of Finland. The main river is plotted using larger linewidth than the tertiarys.

• 

  1. The first plot of all the rivers without modifications.

• 

  2. The main branch and tributaries.

OSM Geofabrik provides huge amount of GIS data. I need to extract one main branch of that all.

The problems are

1. There are many rivers haring the same name
2. The shp data breaks
3. There are cycles (loops) and tributaries (branches)

1. Plot and check

The analysis of different coordinate systems is shown elsewhere.

Plot all Waterways

import geopandas as gpd
import matplotlib.pyplot as plt

finland = gpd.read_file("fi.shp")
rivers = gpd.read_file("gis_osm_waterways_free_1.shp")

#Check coordinate system
print(finland.crs)
print(rivers.crs) #Says None.
rivers = rivers.set_crs(epsg=4326)
rivers = rivers.to_crs(finland.crs)

#Other coordinate system. Looks even better, more familiar
finland = finland.to_crs(epsg=3067)
rivers = rivers.to_crs(epsg=3067)

#rivers = gpd.clip(rivers, finland) # Clip rivers to Finland

# Plotting
fig, ax = plt.subplots(figsize=(6, 10))
finland.plot( ax=ax, color="#f0f0f0", edgecolor="#444444", linewidth=0.8 )
rivers.plot( ax=ax, color="#4aa3df", linewidth=0.3, alpha=0.7 )

ax.set_axis_off() # Remove axes
plt.title("Rivers of Finland", fontsize=14)
plt.show()

The shp datafile

General format

• .shp = shapes
• .dbf = table (like a spreadsheet with headers)
• .prj = map projection

print(rivers.head())
    osm_id  ...                                           geometry
0  4251428  ...  LINESTRING (384623.065 6671261.378, 384613.062...
1  4336744  ...  LINESTRING (424219.382 6758572.023, 424231.281...
2  4418854  ...  LINESTRING (271969.063 6961997.189, 272511.243...
3  4418989  ...  LINESTRING (263355.182 6994559.516, 263029.852...
4  4494437  ...  LINESTRING (235473.685 6716109.64, 235468.154 ...

print(rivers.columns)
Index(['osm_id', 'code', 'fclass', 'width', 'name', 'geometry'], dtype='object')

print( rivers.info() )
<class 'geopandas.geodataframe.GeoDataFrame'>
RangeIndex: 127503 entries, 0 to 127502
Data columns (total 6 columns):
 #   Column    Non-Null Count   Dtype   
---  ------    --------------   -----   
 0   osm_id    127503 non-null  object  
 1   code      127503 non-null  int32   
 2   fclass    127503 non-null  object  
 3   width     127503 non-null  int32   
 4   name      26074 non-null   object  
 5   geometry  127503 non-null  geometry
dtypes: geometry(1), int32(2), object(3)
memory usage: 4.9+ MB
None

print( river( ['fclass'].value_counts() )
fclass
stream    104937
river      12776
drain       9186
canal        604
Name: count, dtype: int64

print(rivers.geometry)
0         LINESTRING (384623.065 6671261.378, 384613.062...
1         LINESTRING (424219.382 6758572.023, 424231.281...
2         LINESTRING (271969.063 6961997.189, 272511.243...
3         LINESTRING (263355.182 6994559.516, 263029.852...
4         LINESTRING (235473.685 6716109.64, 235468.154 ...
                                ...                        
127498    LINESTRING (193860.594 6771306.041, 193862.48 ...
127499    LINESTRING (193500.06 6770531.712, 193481.661 ...
127500    LINESTRING (267357.405 6823421.996, 267348.152...
127501    LINESTRING (266604.152 6824271.814, 266614.528...
127502    LINESTRING (543881.124 6893248.187, 543974.428...
Name: geometry, Length: 127503, dtype: geometry

print(rivers.describe(include='all'))
        osm_id  ...                                           geometry
count    127503  ...                                             127503
unique   127503  ...                                             127502
top     4251428  ...  LINESTRING (388682.74185180914 6718165.8999426...
freq          1  ...                                                  2
mean        NaN  ...                                                NaN
std         NaN  ...                                                NaN
min         NaN  ...                                                NaN
25%         NaN  ...                                                NaN
50%         NaN  ...                                                NaN
75%         NaN  ...                                                NaN
max         NaN  ...                                                NaN

[11 rows x 6 columns]

gdf['name'].dropna().head(20)
0     Ruoholahdenkanava
1           Porvoonjoki
2            Kyrönjoki
3            Kyrönjoki
4           Raisionjoki
5           Riipinluoma
6             Tjäck Ã¥
7            Kyrönjoki
8            Kyrönjoki
9             Kyro Älv
10            Jalasjoki
11          Kupparinoja
12           Harjunjoki
13         Jyräänjoki
15           Hirvenjoki
16             Takajoki
18             Juotjoki
21     Luhtaanmäenjoki
22         Kostianvirta
23        Ahlaistenjoki
Name: name, dtype: object

2. Plot Kemijoki and tributaries

• 

  The effect of buffer

• 

  Larger buffer

• 

  Clipped with 10 km buffer

• 

  Clipped with 20 km buffer

• 

  Clipped with 40 km buffer

• 

  Clipped with 80 km buffer

• 

  The unary union of the main geometry

• 

  Connected, intersections

• 

  Touching 1. Buffer 80 km.

• 

  Touching 2. Buffer 80 km.

• 

  Touching 3. Buffer 80 km.

• 

  Touching 4. Buffer 80 km.

• 

  Touching 5. Buffer 80 km.

• 

  Touching 10. Buffer 80 km.

• 

  Touching 10. Buffer 80 km.

The beginning (loading the river data files and converting) is the same as before.

kemijoki = rivers[rivers["name"] == "Kemijoki"]
buffer = kemijoki.buffer(50000)  # 20 km buffer around the Kemijoko
buffer_gdf = gpd.GeoDataFrame(geometry=buffer, crs=rivers.crs)

kemijoki_rivers = gpd.clip(rivers, buffer_gdf)

#Find the connected parts
main_geom = unary_union(kemijoki.geometry)
connected = kemijoki_rivers[kemijoki_rivers.geometry.intersects(main_geom)]

current = main_geom
for i in range(5):  # increase if needed
    touching = kemijoki_rivers[kemijoki_rivers.geometry.intersects(current)]
    current = unary_union(touching.geometry)

final_rivers = kemijoki_rivers[kemijoki_rivers.geometry.intersects(current)]

# Plot it
fig, ax = plt.subplots(figsize=(6, 10))

finland.plot(ax=ax, color="#f0f0f0", edgecolor="gray")
final_rivers.plot(ax=ax, color="blue", linewidth=0.4)
kemijoki.plot(ax=ax, color="darkblue", linewidth=1.5)
ax.set_axis_off()

plt.show()

This is easy to edit to show all rivers. However, there are many rivers sharing the same name.

3. Fix the gaps (Make a graph)

Kemijoki_mainbranch.parquet. 4e_combinedByHand.py