Skip to content

Commit

Permalink
fix auth/init, use geemap for interactive map, use Google asset
Browse files Browse the repository at this point in the history 
PiperOrigin-RevId: 598977841
  • Loading branch information
@jdbcode @copybara-github
jdbcode authored and copybara-github committed Jan 17, 2024
1 parent 21ce5a4 commit e77cb78
Show file tree
Hide file tree
Showing 3 changed files with 35 additions and 140 deletions.
52 changes: 8 additions & 44 deletions tutorials/imad-tutorial-pt1/index.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -73,58 +73,22 @@ The cells below execute the necessary formalities for accessing Earth Engine fro


```python
# Authenticate
COLAB_AUTH_FLOW_CLOUD_PROJECT_FOR_API_CALLS = None

import ee
import google
import os

if COLAB_AUTH_FLOW_CLOUD_PROJECT_FOR_API_CALLS is None:
print("Authenticating using Notebook auth...")
if os.path.exists(ee.oauth.get_credentials_path()) is False:
ee.Authenticate()
else:
print('\N{check mark} '
'Previously created authentication credentials were found.')
ee.Initialize()
else:
print('Authenticating using Colab auth...')
# Authenticate to populate Application Default Credentials in the Colab VM.
google.colab.auth.authenticate_user()
# Create credentials needed for accessing Earth Engine.
credentials, auth_project_id = google.auth.default()
# Initialize Earth Engine.
ee.Initialize(credentials, project=COLAB_AUTH_FLOW_CLOUD_PROJECT_FOR_API_CALLS)
print('\N{check mark} Successfully initialized!')
```


```python
# Install the ee_jupyter package
import os
try:
import ee_jupyter
except ModuleNotFoundError:
print('ee_jupyter was not found. Installing now...')
result = os.system('pip -q install earthengine-jupyter')
import ee_jupyter
print(f'ee_jupyter (version {ee_jupyter.__version__}) '
f'is installed.')
ee.Authenticate(auth_mode='notebook')
ee.Initialize()
```


```python
# Import other packages used in the tutorial
%matplotlib inline
import geemap
import numpy as np
import random, time
import matplotlib.pyplot as plt
from scipy.stats import norm, chi2

from ee_jupyter.colab import set_colab_output_cell_height
from ee_jupyter.ipyleaflet import Map
from ee_jupyter.ipyleaflet import Inspector
from pprint import pprint # for pretty printing
```

Expand Down Expand Up @@ -181,8 +145,7 @@ For example, here is an area of interest (aoi) covering a heavily forested admin


```python
aois = ee.FeatureCollection('projects/ee-mortcanty/assets/dvg1krs_nw').geometry()
aoi = ee.Geometry(aois.geometries().get(26))
aoi = ee.FeatureCollection('projects/google/imad_tutorial_aoi').geometry()
```

We first collect two Sentinel-2 scenes which bracket some of the recent clear cutting (June, 2021 and June, 2022) and print out their timestamps. For demonstration purposes we choose top of atmosphere reflectance since the MAD transformation does not require absolute surface reflectances.
Expand Down Expand Up @@ -221,8 +184,8 @@ In order to view the images, we instantiate an interactive map located at the ce

```python
# Interactive map
location = aoi.centroid().coordinates().getInfo()[::-1]
M1 = Map(**{'center': location, 'zoom': 11})
M1 = geemap.Map()
M1.centerObject(aoi, 11)

M1
```
Expand Down Expand Up @@ -520,7 +483,8 @@ we display an RGB composite of the first 3 MAD variates, together with the origi


```python
M2 = Map(**{'center': location, 'zoom': 11})
M2 = geemap.Map()
M2.centerObject(aoi, 11)
display_ls(im1.select(visbands), M2, 'Image 1')
display_ls(im2.select(visbands), M2, 'Image 2')
display_ls(diff, M2, 'Difference')
Expand Down
62 changes: 14 additions & 48 deletions tutorials/imad-tutorial-pt2/index.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -47,56 +47,21 @@ EXPORT_PATH = 'projects/YOUR_GEE_PROJECT_NAME/assets/imad/'
```

```python
# Authenticate
COLAB_AUTH_FLOW_CLOUD_PROJECT_FOR_API_CALLS = None

import ee
import google
import os

if COLAB_AUTH_FLOW_CLOUD_PROJECT_FOR_API_CALLS is None:
print("Authenticating using Notebook auth...")
if os.path.exists(ee.oauth.get_credentials_path()) is False:
ee.Authenticate()
else:
print('\N{check mark} '
'Previously created authentication credentials were found.')
ee.Initialize()
else:
print('Authenticating using Colab auth...')
# Authenticate to populate Application Default Credentials in the Colab VM.
google.colab.auth.authenticate_user()
# Create credentials needed for accessing Earth Engine.
credentials, auth_project_id = google.auth.default()
# Initialize Earth Engine.
ee.Initialize(credentials, project=COLAB_AUTH_FLOW_CLOUD_PROJECT_FOR_API_CALLS)
print('\N{check mark} Successfully initialized!')
```

```python
# Install the ee_jupyter package.
import os
try:
import ee_jupyter
except ModuleNotFoundError:
print('ee_jupyter was not found. Installing now...')
result = os.system('pip -q install earthengine-jupyter')
import ee_jupyter
print(f'ee_jupyter (version {ee_jupyter.__version__}) '
f'is installed.')
ee.Authenticate(auth_mode='notebook')
ee.Initialize()
```

```python
# Import other packages used in the tutorial.
%matplotlib inline
import geemap
import numpy as np
import random, time
import matplotlib.pyplot as plt
from scipy.stats import norm, chi2

from ee_jupyter.colab import set_colab_output_cell_height
from ee_jupyter.ipyleaflet import Map
from ee_jupyter.ipyleaflet import Inspector
from pprint import pprint # for pretty printing
```

Expand Down Expand Up @@ -322,9 +287,7 @@ the MAD variates that we have determined so far for the Landkreis Olpe scene.

```python
# Landkreis Olpe.
# (`aois` FeatureCollection is owned and managed by tutorial author)
aois = ee.FeatureCollection('projects/ee-mortcanty/assets/dvg1krs_nw').geometry()
aoi = ee.Geometry(aois.geometries().get(26))
aoi = ee.FeatureCollection('projects/google/imad_tutorial_aoi').geometry()

visirbands = ['B2', 'B3', 'B4', 'B8', 'B11', 'B12']
visbands = ['B2', 'B3', 'B4']
Expand Down Expand Up @@ -549,8 +512,8 @@ Agreement is not perfect, but the plot is certainly closer to the ideal chi-squa
after iteration than after the single MAD transformation. So let's display the *iMAD* image:

```python
location = aoi.centroid().coordinates().getInfo()[::-1]
M1 = Map(**{'center': location, 'zoom': 11})
M1 = geemap.Map()
M1.centerObject(aoi, 11)
display_ls(im1.select(visbands), M1, 'im1')
display_ls(im2.select(visbands), M1, 'im2')
display_ls(im_imad.select('iMAD1', 'iMAD2', 'iMAD3'), M1, 'iMAD123', True)
Expand Down Expand Up @@ -603,7 +566,8 @@ result = im_imadstd.cluster(clusterer)
Here we display the four clusters overlaid onto the two Sentinel 2 images:

```python
M2 = Map(**{'center': location, 'zoom': 13})
M2 = geemap.Map()
M2.centerObject(aoi, 13)
display_ls(im1.select(visbands), M2, 'im1')
display_ls(im2.select(visbands), M2, 'im2')
cluster0 = result.updateMask(result.eq(0))
Expand Down Expand Up @@ -655,7 +619,8 @@ dyn = ee.ImageCollection('GOOGLE/DYNAMICWORLD/V1') \
# 'trees' class = class 1
dw = dyn.clip(aoi).updateMask(dyn.eq(1))

M3 = Map(**{'center': location, 'zoom': 13})
M3 = geemap.Map()
M3.centerObject(aoi, 13)
display_ls(im1.select(visbands), M3, 'im1')
display_ls(im2.select(visbands), M3, 'im2')
M3.addLayer(dw, {'min': 0, 'max': 1, 'palette': ['black', 'green']}, 'dynamic world')
Expand All @@ -672,7 +637,8 @@ and is also only weakly correlated with the other 5 bands. However, close inspec
that there are many more false positives, especially in agricultural fields, as well as in the reservoir.

```python
M4 = Map(**{'center': location, 'zoom': 13})
M4 = geemap.Map()
M4.centerObject(aoi, 13)
diff = im1.subtract(im2).select(visirbands)
training = diff.sample(region=aoi, scale=20, numPixels=50000)
clusterer = ee.Clusterer.wekaKMeans(4).train(training)
Expand Down Expand Up @@ -778,8 +744,8 @@ aois = filtered.geometry()
# There are many Montgomery Counties in USA, we want the 12th in the list.
aoi = ee.Geometry(aois.geometries().get(12))
im1, im2 = collect(aoi, '2021-07-01', '2021-07-30', '2022-06-01', '2022-06-30')
location = aoi.centroid().coordinates().getInfo()[::-1]
M5 = Map(**{'center':location, 'zoom': 10})
M5 = geemap.Map()
M5.centerObject(aoi, 10)
display_ls(im1.select(visbands), M5, 'Image 1')
display_ls(im2.select(visbands), M5, 'Image 2')

Expand Down
61 changes: 13 additions & 48 deletions tutorials/imad-tutorial-pt3/index.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -47,57 +47,21 @@ EXPORT_PATH = 'projects/YOUR_GEE_PROJECT_NAME/assets/imad/'
```

```python
# Authenticate.
COLAB_AUTH_FLOW_CLOUD_PROJECT_FOR_API_CALLS = None

import ee
import google
import os

if COLAB_AUTH_FLOW_CLOUD_PROJECT_FOR_API_CALLS is None:
print("Authenticating using Notebook auth...")
if os.path.exists(ee.oauth.get_credentials_path()) is False:
ee.Authenticate()
else:
print('\N{check mark} '
'Previously created authentication credentials were found.')
#ee.Authenticate()
ee.Initialize()
else:
print('Authenticating using Colab auth...')
# Authenticate to populate Application Default Credentials in the Colab VM.
google.colab.auth.authenticate_user()
# Create credentials needed for accessing Earth Engine.
credentials, auth_project_id = google.auth.default()
# Initialize Earth Engine.
ee.Initialize(credentials, project=COLAB_AUTH_FLOW_CLOUD_PROJECT_FOR_API_CALLS)
print('\N{check mark} Successfully initialized!')
```

```python
# Install the ee_jupyter package.
import os
try:
import ee_jupyter
except ModuleNotFoundError:
print('ee_jupyter was not found. Installing now...')
result = os.system('pip -q install earthengine-jupyter')
import ee_jupyter
print(f'ee_jupyter (version {ee_jupyter.__version__}) '
f'is installed.')
ee.Authenticate(auth_mode='notebook')
ee.Initialize()
```

```python
# Import other packages used in the tutorial.
# Import other packages used in the tutorial
%matplotlib inline
import geemap
import numpy as np
import random, time
import matplotlib.pyplot as plt
from scipy.stats import norm, chi2

from ee_jupyter.colab import set_colab_output_cell_height
from ee_jupyter.ipyleaflet import Map
from ee_jupyter.ipyleaflet import Inspector
from pprint import pprint # for pretty printing
```

Expand Down Expand Up @@ -467,10 +431,7 @@ As a first illustration, we collect two minimally cloudy Sentinel-2 images over

```python
# Landkreis Olpe area of interest.
# (`aois` FeatureCollection is owned and managed by tutorial author)
aois = ee.FeatureCollection('projects/ee-mortcanty/assets/dvg1krs_nw').geometry()
aoi = ee.Geometry(aois.geometries().get(26))
location = aoi.centroid().coordinates().getInfo()[::-1]
aoi = ee.FeatureCollection('projects/google/imad_tutorial_aoi').geometry()

# Spectral band combinations
#Sentinel-2
Expand Down Expand Up @@ -560,7 +521,8 @@ plot_orthoregress( coeffs, im_stack )
The next cell compares the normalized and unnormalized TOA images with the reference SR image. The no-change pixels used in the regression are also displayed:

```python
M1 = Map(**{'center': location, 'zoom': 13})
M1 = geemap.Map()
M1.centerObject(aoi, 13)
M1.addLayer(im1_sr.select(visnirbands), visnirparams, 'Im1_sr')
M1.addLayer(im2_toa.select(visnirbands), visnirparams, 'Im2_toa')
M1.addLayer(im2_toa_norm.select(visnirbands), visnirparams, 'Im2_toa_norm')
Expand All @@ -579,7 +541,8 @@ def ndvi_s2(im):
```

```python
M2 = Map(**{'center': location, 'zoom': 13})
M2 = geemap.Map()
M2.centerObject(aoi, 13)
M2.addLayer(ndvi_s2(im1_sr), {'min': 0, 'max': 1.2, 'palette': ['black', 'yellow']}, 'Im1_ndvi')
M2.addLayer(ndvi_s2(im2_toa_norm), {'min': 0, 'max': 1.2, 'palette': ['black', 'yellow']}, 'Im2_toa_norm_ndvi')
M2.addLayer(ndvi_s2(im2_toa), {'min': 0, 'max': 1.2, 'palette': ['black', 'yellow']}, 'Im2_toa_ndvi')
Expand Down Expand Up @@ -620,7 +583,8 @@ The above plot for the NIR band B8 is interesting as it seems to indicate that t
In any case, the regression plots look reasonable, so here again is the comparison of the normalized and unnormalized TOA images with the reference SR image:

```python
M3 = Map(**{'center': location, 'zoom': 13})
M3 = geemap.Map()
M3.centerObject(aoi, 13)
M3.addLayer(im1_sr.select(visnirbands), visnirparams, 'Im1_sr')
M3.addLayer(im2_toa.select(visnirbands), visnirparams, 'Im1_toa')
M3.addLayer(im2_toa_norm.select(visnirbands), visnirparams, 'Im2_toa_norm')
Expand Down Expand Up @@ -664,7 +628,8 @@ plot_orthoregress(coeffs, im_stack, bandNames=visirbands_ls)
For most of the bands, the intercepts are small and the slopes fairly close to one. The NIR band 'SR_B5' is the exception. This is apparent in the direct comparison of the original and harmonized Landsat 8 images:

```python
M4 = Map(**{'center': location, 'zoom': 11})
M4 = geemap.Map()
M4.centerObject(aoi, 11)
M4.addLayer(im_ls9.select(visnirbands_ls), visnirparams, 'Im_ls9')
M4.addLayer(im_ls8.select(visnirbands_ls), visnirparams, 'Im_ls8')
M4.addLayer(im_ls8_norm.select(visnirbands_ls), visnirparams, 'Im_ls8_norm')
Expand Down

0 comments on commit e77cb78

Please sign in to comment.