',
unsafe_allow_html=True,
)
# Convert the date objects to strings in the format expected by EE
start_date_str = startDate.strftime('%Y-%m-%d')
end_date_str = endDate.strftime('%Y-%m-%d')
sentinel_image_collection = ee.ImageCollection('COPERNICUS/S2') \
.filterBounds(roi) \
.filterDate(start_date_str, end_date_str)
sentinel_image = sentinel_image_collection \
.sort('CLOUDY_PIXEL_PERCENTAGE') \
.first() \
.clip(roi)
# Visualize using RGB
m.addLayer(sentinel_image,
{'min': 0.0, 'max': 2000, 'bands': ['B4', 'B3', 'B2']},
'RGB')
ndwi = sentinel_image.normalizedDifference(['B3', 'B8']).rename('NDWI')
m.addLayer(ndwi,
{'palette': ['red', 'yellow', 'green', 'cyan', 'blue']},
'NDWI')
# Create NDWI mask
ndwi_threshold = ndwi.gte(0.0)
ndwi_mask = ndwi_threshold.updateMask(ndwi_threshold)
m.addLayer(ndwi_threshold,
{'palette': ['black', 'white']},
'NDWI Binary Mask')
m.addLayer(ndwi_mask,
{'palette': ['blue']},
'NDWI Mask')
if "Water Surface Elevation" in output:
# Send a request to the Hydrocron API to get lake data in CSV format
url = (
"https://soto.podaac.earthdatacloud.nasa.gov/hydrocron/v1/timeseries?"
"feature=PriorLake"
f"&feature_id={lake_id}"
f"&start_time={start_date_str}T00:00:00Z"
f"&end_time={end_date_str}T00:00:00Z"
"&output=csv"
"&fields=time_str,wse"
)
# Request the data
hydrocron_response = requests.get(url).json()
# Extract CSV data from the response
csv_str = hydrocron_response['results']['csv']
# Convert the CSV string into a pandas DataFrame
df = pd.read_csv(StringIO(csv_str))
# Prepare to plot water surface elevation (WSE) and area
# Convert 'time_str' column to datetime format
df['time_str'] = pd.to_datetime(df['time_str'], errors='coerce')
# Filter and store dates and elevations
df_filtered = df.dropna(subset=['time_str', 'wse'])
df_filtered['wse'] = pd.to_numeric(df_filtered['wse'], errors='coerce')
df_filtered = df_filtered.dropna(subset=['wse'])
# Plot water surface elevation (WSE) over time
plt.figure(figsize=(10, 5))
plt.plot(df_filtered['time_str'], df_filtered['wse'], marker='o', linestyle='-')
plt.xlabel('Date')
plt.ylabel('Water Surface Elevation (m)')
plt.title(f'Water Surface Elevation for Lake {lake_name}')
plt.xticks(rotation=45)
plt.grid(True)
# Show the plot in Streamlit
st.pyplot(plt)
# Display the filtered DataFrame
st.write(df_filtered[['time_str', 'wse']])
with st.spinner('Retrieving satilite images...'):
#Define a function to calculate NDWI
def calculate_ndwi(image):
ndwi = image.normalizedDifference(["B8", "B3"]) # B8 is NIR and B3 is green
return ndwi
# Filter Sentinel-2 images
sentinelImageCollection = ee.ImageCollection('COPERNICUS/S2') \
.filterBounds(roi) \
.filterDate(start_date_str, end_date_str) \
.filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', threshold)) \
# Check if images are available
num_images = sentinelImageCollection.size().getInfo()
st.write("Number of images:", num_images)
volumes = []
# Alternatively, convert acquisition times to readable format (if needed)
acquisition_times = sentinelImageCollection.aggregate_array('system:time_start').getInfo()
acquisition_dates = [datetime.datetime.utcfromtimestamp(time / 1000).strftime('%Y-%m-%d') for time in acquisition_times]
if num_images == 0:
st.warning("No images available within the specified date range.")
else:
if threshold >= 15:
st.write("CLoudless Algorithm will identify and remove the effects of clouds and shadows")
START_DATE = start_date_str
END_DATE = end_date_str
CLOUD_FILTER = 40
CLD_PRB_THRESH = 70
NIR_DRK_THRESH = 0.15
CLD_PRJ_DIST = 2
BUFFER = 100
# Function to get Sentinel-2 surface reflectance and cloud probability collections
def get_s2_sr_cld_col(aoi, start_date, end_date):
s2_sr_col = (ee.ImageCollection('COPERNICUS/S2_SR')
.filterBounds(aoi)
.filterDate(start_date, end_date)
.filter(ee.Filter.lte('CLOUDY_PIXEL_PERCENTAGE', CLOUD_FILTER)))
s2_cloudless_col = (ee.ImageCollection('COPERNICUS/S2_CLOUD_PROBABILITY')
.filterBounds(aoi)
.filterDate(start_date, end_date))
return ee.ImageCollection(ee.Join.saveFirst('s2cloudless').apply(**{
'primary': s2_sr_col,
'secondary': s2_cloudless_col,
'condition': ee.Filter.equals(**{
'leftField': 'system:index',
'rightField': 'system:index'
})
}))
# Apply the function to build the collection
s2_sr_cld_col = get_s2_sr_cld_col(roi, START_DATE, END_DATE)
# Function to get cloud cover percentage for an image
def get_cloud_cover_percentage(image):
cloud_cover = ee.Image(image).get('CLOUDY_PIXEL_PERCENTAGE')
return ee.Feature(None, {'cloud_cover': cloud_cover, 'image_id': image.id()})
# Apply the function to the collection
image_list = s2_sr_cld_col.map(get_cloud_cover_percentage).getInfo()
# Debug: Print the properties of the first image to inspect the available properties
print("Inspecting the first image's properties:")
print(image_list['features'][0]['properties'])
# Extract the image ids, cloud covers, and dates (if available)
image_info = []
for f in image_list['features']:
image_id = f['properties'].get('image_id', 'Unknown')
cloud_cover = f['properties'].get('cloud_cover', 'Unknown')
timestamp = f['properties'].get('system:time_start', None)
# If timestamp is None, we'll set it to 'Unknown'
if timestamp:
date = datetime.utcfromtimestamp(timestamp / 1000).strftime('%Y-%m-%d')
else:
date = 'Unknown'
image_info.append((image_id, cloud_cover, date))
print("Available images and their cloud cover percentages:")
for idx, (image_id, cloud_cover, date) in enumerate(image_info):
print(f"{idx}: Image ID: {image_id}, Date: {date}, Cloud Cover: {cloud_cover}%")
water_area_info = []
# Count the number of images in the collection
num_images = s2_sr_cld_col.size().getInfo()
print(f"Total number of images in the collection: {num_images}")
# Loop through each image in the collection and print its cloud cover
for i in range(num_images):
selected_idx = i
selected_image_id = image_info[selected_idx][0]
cloud_cover = image_info[selected_idx][1] # Get the cloud cover for the selected image
selected_image = ee.Image(s2_sr_cld_col.filter(ee.Filter.eq('system:index', selected_image_id)).first())
print(f"Image ID: {selected_image_id}, Cloud Cover: {cloud_cover}%")
if cloud_cover >= 15:
# Define functions to add cloud and shadow bands
def add_cloud_bands(img):
cld_prb = ee.Image(img.get('s2cloudless')).select('probability')
is_cloud = cld_prb.gt(CLD_PRB_THRESH).rename('clouds')
return img.addBands(ee.Image([cld_prb, is_cloud]))
def add_shadow_bands(img):
not_water = img.select('SCL').neq(6)
SR_BAND_SCALE = 1e4
dark_pixels = img.select('B8').lt(NIR_DRK_THRESH * SR_BAND_SCALE).multiply(not_water).rename('dark_pixels')
shadow_azimuth = ee.Number(90).subtract(ee.Number(img.get('MEAN_SOLAR_AZIMUTH_ANGLE')))
cld_proj = (img.select('clouds').directionalDistanceTransform(shadow_azimuth, CLD_PRJ_DIST * 10)
.reproject(crs=img.select(0).projection(), scale=100)
.select('distance').mask().rename('cloud_transform'))
shadows = cld_proj.multiply(dark_pixels).rename('shadows')
return img.addBands(ee.Image([dark_pixels, cld_proj, shadows]))
def add_cld_shdw_mask(img):
img_cloud = add_cloud_bands(img)
img_cloud_shadow = add_shadow_bands(img_cloud)
is_cld_shdw = img_cloud_shadow.select('clouds').add(img_cloud_shadow.select('shadows')).gt(0)
is_cld_shdw = (is_cld_shdw.focalMin(2).focalMax(BUFFER * 2 / 20)
.reproject(crs=img.select([0]).projection(), scale=20)
.rename('cloudmask'))
return img_cloud_shadow.addBands(is_cld_shdw)
# Define the function to apply the cloud and shadow mask
def apply_cld_shdw_mask(img):
not_cld_shdw = img.select('cloudmask').Not()
return img.select('B.*').updateMask(not_cld_shdw)
# Add cloud and shadow bands, apply the mask
selected_image_with_mask = add_cld_shdw_mask(selected_image)
cloud_free_image = apply_cld_shdw_mask(selected_image_with_mask)
# Define a function to calculate NDWI and mask
def calculate_ndwi_and_mask(image):
ndwi = image.normalizedDifference(['B3', 'B8']).rename('NDWI')
ndwi_threshold = ndwi.gte(0.0)
ndwi_mask = ndwi_threshold.updateMask(ndwi_threshold)
return ndwi_mask
# Apply the function to the latest image to calculate NDWI mask
ndwi_mask = calculate_ndwi_and_mask(selected_image)
# Define a function to calculate water area
def calculate_water_area(image):
water_area = image.multiply(ee.Image.pixelArea()).reduceRegion(
reducer=ee.Reducer.sum(),
geometry=roi,
scale=5
).get('NDWI')
return image.set('water_area', water_area)
# Calculate water area for the NDWI mask
ndwi_mask_with_area = calculate_water_area(ndwi_mask)
waterarea = ndwi_mask_with_area.get('water_area').getInfo()
w = waterarea
#print(f"This is the water area of the NDWI image:{w}")
# Load the bathymetry dataset from Earth Engine
globathy = ee.Image("projects/sat-io/open-datasets/GLOBathy/GLOBathy_bathymetry")
# Export the data as an image
out_dir = "/Users/joaopimenta/Desktop/GEE_test" # Specify the output directory
if not os.path.exists(out_dir):
os.makedirs(out_dir)
out_image_path = os.path.join(out_dir, "globathy_bathymetry.tif") # Specify the output image path
# Export the image
geemap.ee_export_image(globathy, filename=out_image_path, scale=10, region=roi)
# Load Bathymetry image
bathymetry_path = out_image_path
bathymetry_dataset = rasterio.open(bathymetry_path)
#print(f"This is the ndwi area of the lake {ndwi_masked_area}")
# Export the binary water mask to a GeoTIFF file
folder_name = datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S") + "_dam_volume_images_tif"
directory = "/Users/joaopimenta/Desktop"
folder_path = os.path.join(directory, folder_name)
os.makedirs(folder_path)
geemap.ee_export_image(
ndwi_mask,
filename=os.path.join(folder_path, "binary_NDWI.tif"),
region=roi,
scale=10
)
file_name = "binary_NDWI.tif"
file_path = os.path.join(folder_path, file_name)
# Load NDWI image
ndwi_path = file_path # Update this path
ndwi_dataset = rasterio.open(ndwi_path)
ndwi = ndwi_dataset.read(1)
with rasterio.open(file_path) as src:
ndwi_data = src.read(1) # Read the first band
transform = src.transform
# Convert NDWI to binary format for visualization
binary_ndwi = np.where(ndwi_data == 1, 255, 0).astype(np.uint8)
# Calculate the area of the detected water bodies from binary mask
def calculate_area(image, transform):
# Mask the image to include only water
water_mask = image == 0
# Compute the area in square meters
pixel_area = abs(transform[0] * transform[4]) # pixel size (in square meters)
water_area_pixels = np.sum(water_mask)
total_area_m2 = water_area_pixels * pixel_area
return total_area_m2
# Calculate the area using the converted binary mask
total_area_m2 = calculate_area(binary_ndwi, transform)/ 1e3
#print(f"Total area calculated from binary mask: {total_area_m2 :.2f} km²")
# Plot the results
plt.figure(figsize=(15, 10))
# Binary NDWI (K-means method)
plt.subplot(1, 2, 1)
plt.imshow(binary_ndwi, cmap='gray')
plt.title('Binary NDWI (K-means)')
# Identified contour (K-means method)
plt.subplot(1, 2, 2)
contour_image = np.zeros_like(binary_ndwi)
contours, _ = cv2.findContours(binary_ndwi, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
if contours:
cv2.drawContours(contour_image, [max(contours, key=cv2.contourArea)], -1, (255), 2)
plt.imshow(contour_image, cmap='gray')
plt.title('Identified Dam Contour (K-means)')
plt.show()
# Check for cloud pixels within the dam (ROI)
cloud_pixels_in_roi = selected_image_with_mask.select('cloudmask').reduceRegion(
reducer=ee.Reducer.sum(),
geometry=roi,
scale=10
).get('cloudmask').getInfo()
print(f"This is the cloud pixels in the ROI:{cloud_pixels_in_roi}")
# Export the cloud mask
geemap.ee_export_image(
selected_image_with_mask.select('cloudmask'),
filename=os.path.join(folder_path, "cloud_mask.tif"),
region=roi,
scale=10
)
cloud_mask_path = os.path.join(folder_path, "cloud_mask.tif")
cloud_mask_dataset = rasterio.open(cloud_mask_path)
cloud_mask = cloud_mask_dataset.read(1)
# Reproject the cloud mask to match NDWI resolution
resampled_cloud_mask = np.empty_like(ndwi)
reproject(
source=cloud_mask,
destination=resampled_cloud_mask,
src_transform=cloud_mask_dataset.transform,
src_crs=cloud_mask_dataset.crs,
dst_transform=ndwi_dataset.transform,
dst_crs=ndwi_dataset.crs,
resampling=Resampling.nearest)
# Mask the NDWI image by removing cloud pixels
ndwi_masked = np.where(resampled_cloud_mask == 0, ndwi, np.nan)
# Load Bathymetry image
bathymetry_dataset = rasterio.open(bathymetry_path)
# Reproject Bathymetry to the NDWI CRS
dst_crs = ndwi_dataset.crs
transform, width, height = calculate_default_transform(
bathymetry_dataset.crs, dst_crs, bathymetry_dataset.width,
bathymetry_dataset.height, *bathymetry_dataset.bounds)
kwargs = bathymetry_dataset.meta.copy()
kwargs.update({
'crs': dst_crs,
'transform': transform,
'width': width,
'height': height
})
reprojected_bathymetry = np.empty((height, width), dtype=np.float32)
reproject(
source=rasterio.band(bathymetry_dataset, 1),
destination=reprojected_bathymetry,
src_transform=bathymetry_dataset.transform,
src_crs=bathymetry_dataset.crs,
dst_transform=transform,
dst_crs=dst_crs,
resampling=Resampling.nearest)
# Resample Bathymetry to match NDWI resolution
resampled_bathymetry = np.empty_like(ndwi)
reproject(
source=reprojected_bathymetry,
destination=resampled_bathymetry,
src_transform=transform,
src_crs=dst_crs,
dst_transform=ndwi_dataset.transform,
dst_crs=dst_crs,
resampling=Resampling.bilinear)
# Plot the images
fig, ax = plt.subplots(figsize=(10, 10))
# Plot the NDWI image
ndwi_extent = (ndwi_dataset.bounds.left, ndwi_dataset.bounds.right,
ndwi_dataset.bounds.bottom, ndwi_dataset.bounds.top)
cax_ndwi = ax.imshow(ndwi_masked, cmap='Blues', extent=ndwi_extent,
alpha=0.6)
# Overlay the Bathymetry image
bathy_extent = (ndwi_dataset.bounds.left, ndwi_dataset.bounds.right,
ndwi_dataset.bounds.bottom, ndwi_dataset.bounds.top)
cax_bathy = ax.imshow(resampled_bathymetry, cmap='viridis',
extent=bathy_extent, alpha=0.4)
fig.colorbar(cax_bathy, ax=ax, fraction=0.046, pad=0.04,
label='Bathymetry')
# Plot the NDWI cloud-removed image
fig, ax = plt.subplots(figsize=(10, 10))
# Plot the NDWI image
ndwi_extent = (ndwi_dataset.bounds.left, ndwi_dataset.bounds.right, ndwi_dataset.bounds.bottom, ndwi_dataset.bounds.top)
cax_ndwi = ax.imshow(ndwi_masked, cmap='Blues', extent=ndwi_extent)
fig.colorbar(cax_ndwi, ax=ax, fraction=0.046, pad=0.04, label='NDWI')
# Get the cloud mask from the selected image
cloud_mask = selected_image_with_mask.select('cloudmask')
# Apply cloud mask to the NDWI mask
ndwi_cloud_removed_mask = ndwi_mask.updateMask(cloud_mask.Not())
# Calculate the pixel area for the masked NDWI image
pixel_area = ndwi_cloud_removed_mask.multiply(ee.Image.pixelArea())
# Reduce the region to calculate the total water area
water_area = pixel_area.reduceRegion(
reducer=ee.Reducer.sum(),
geometry=roi,
scale=10, # Adjust the scale as needed
maxPixels=1e10
)
# Assuming water_area is the result from reduceRegion
water = water_area.getInfo().get('NDWI')
print(water)
# Get the total water area in square meters
total_water_area_m2 = total_area_m2
# Convert the area to square kilometers
total_water_area_km2 = total_water_area_m2 / 1e6# Convert the area to square kilometers
total_water_area_adjusted = total_water_area_m2
area_cloud_aftected = w - total_water_area_adjusted
cloud_affect_percentage = area_cloud_aftected/ cloud_pixels_in_roi
print(f"The total NDWI water area is:{w}")
print(f"The adjusted water area is: {total_water_area_adjusted}")
print(f"The total amount of pixels covering the reservoir is:{area_cloud_aftected}")
print(f"This is the area cloud pixels in the ROI:{cloud_pixels_in_roi*10}")
print(f"The percentage of pixels which affect the reservoir's area are :{cloud_affect_percentage}")
if cloud_pixels_in_roi > 0:
import rasterio
import numpy as np
import matplotlib.pyplot as plt
from skimage import measure
from shapely.geometry import Polygon
from pyproj import Transformer
import rasterio.transform
from scipy.ndimage import binary_fill_holes # To fill inside polygons
from rasterio.warp import reproject, Resampling, calculate_default_transform
# Path to bathymetry raster file
path_bathymetry = "/Users/joaopimenta/Desktop/GEE_test/globathy_bathymetry.tif"
# Path to NDWI raster file (the one with the projection you want)
path_ndwi = ndwi_path
path_cloud_mask = cloud_mask_path
# Load Bathymetry image
bathymetry_dataset = rasterio.open(path_bathymetry)
cloud_mask_dataset = rasterio.open(path_cloud_mask)
ndwi_dataset = rasterio.open(path_ndwi)
# Reproject Bathymetry to the NDWI CRS if necessary
dst_crs = ndwi_dataset.crs
transform, width, height = calculate_default_transform(
bathymetry_dataset.crs, dst_crs, bathymetry_dataset.width, bathymetry_dataset.height, *bathymetry_dataset.bounds)
kwargs = bathymetry_dataset.meta.copy()
kwargs.update({
'crs': dst_crs,
'transform': transform,
'width': width,
'height': height
})
reprojected_bathymetry = np.empty((height, width), dtype=np.float32)
reproject(
source=rasterio.band(bathymetry_dataset, 1),
destination=reprojected_bathymetry,
src_transform=bathymetry_dataset.transform,
src_crs=bathymetry_dataset.crs,
dst_transform=transform,
dst_crs=dst_crs,
resampling=Resampling.nearest)
# Reproject cloud mask to the bathymetry CRS if necessary
if bathymetry_dataset.crs != cloud_mask_dataset.crs:
print("CRS misalignment detected. Reprojecting cloud mask to bathymetry CRS.")
reprojected_cloud_mask = np.empty_like(reprojected_bathymetry)
reproject(
source=rasterio.band(cloud_mask_dataset, 1),
destination=reprojected_cloud_mask,
src_transform=cloud_mask_dataset.transform,
src_crs=cloud_mask_dataset.crs,
dst_transform=transform,
dst_crs=dst_crs,
resampling=Resampling.nearest
)
else:
reprojected_cloud_mask = cloud_mask_dataset.read(1)
# Resample Bathymetry and cloud mask to match NDWI resolution if necessary
resampled_bathymetry = np.empty_like(ndwi_dataset.read(1))
resampled_cloud_mask = np.empty_like(ndwi_dataset.read(1))
reproject(
source=reprojected_bathymetry,
destination=resampled_bathymetry,
src_transform=transform,
src_crs=dst_crs,
dst_transform=ndwi_dataset.transform,
dst_crs=dst_crs,
resampling=Resampling.bilinear)
reproject(
source=reprojected_cloud_mask,
destination=resampled_cloud_mask,
src_transform=transform,
src_crs=dst_crs,
dst_transform=ndwi_dataset.transform,
dst_crs=dst_crs,
resampling=Resampling.nearest)
# Load bathymetry raster data
lakeBottom = resampled_bathymetry
resolution = bathymetry_dataset.res
lake_crs = bathymetry_dataset.crs
lake_transform = bathymetry_dataset.transform
# Load NDWI raster data (to get CRS and extent)
ndwi_extent = (ndwi_dataset.bounds.left, ndwi_dataset.bounds.right, ndwi_dataset.bounds.bottom, ndwi_dataset.bounds.top)
# Replace no-data value with np.nan for the bathymetry raster
noDataValue = lakeBottom[0, 0]
lakeBottom = lakeBottom.astype(float)
lakeBottom[lakeBottom == noDataValue] = np.nan
# Calculate minimum and maximum elevation
minElev = np.nanmin(lakeBottom)
maxElev = np.nanmax(lakeBottom)
# Define number of steps for calculation
nSteps = 50
elevSteps = np.round(np.linspace(minElev, maxElev, nSteps), 2)
# Define function to create a mask for a specific elevation
def createMaskForElevation(elevation, elevDem, cloud_mask):
# Create a mask based on the elevation
mask = np.where(elevDem <= elevation, 1, 0)
# Fill holes inside the polygon
filled_mask = binary_fill_holes(mask) * mask # Ensures it's a binary mask
waterarea = np.sum(mask) * (resolution[0] * resolution[1])
area_Array.append(waterarea)
# Apply the cloud mask, setting cloud-covered pixels to 0
filled_mask[cloud_mask == 1] = 0
return filled_mask
# Set up transformation to match NDWI CRS
transformer = Transformer.from_crs(lake_crs, ndwi_dataset.crs, always_xy=True)
# Arrays to store the areas for each elevation step
areaArray = []
area_Array = []
# Plot setup
fig, ax = plt.subplots(figsize=(12, 10))
colors = plt.cm.viridis(np.linspace(0, 1, len(elevSteps)))
for i, elev in enumerate(elevSteps):
# Create a mask for the current elevation and apply cloud mask
mask = createMaskForElevation(elev, lakeBottom, resampled_cloud_mask)
# Calculate water area by summing valid pixels (non-cloud, non-zero)
water_area = np.sum(mask) * (resolution[0] * resolution[1]) # Pixel resolution area
areaArray.append(water_area)
# Find contours (polygons) from the mask
contours = measure.find_contours(mask, 0.5)
# Reproject and plot each contour as a polygon
for contour in contours:
lon_lat_coords = rasterio.transform.xy(lake_transform, contour[:, 0], contour[:, 1])
x_coords, y_coords = np.array(lon_lat_coords[0]), np.array(lon_lat_coords[1])
# Reproject coordinates to NDWI CRS
x_proj, y_proj = transformer.transform(x_coords, y_coords)
# Plot the reprojected contour
ax.plot(x_proj, y_proj, color=colors[i], label=f'Elevation {elev} m' if i == 0 else "")
# Set the same extent as the NDWI image
ax.set_xlim(ndwi_extent[0], ndwi_extent[1])
ax.set_ylim(ndwi_extent[2], ndwi_extent[3])
# Plot the elevation vs area
areaArraySqM = [area * 1e8 for area in areaArray] # Convert to square meters
area_ArraySqM = [area * 1e8 for area in area_Array]
# Paths to the raster file
path = "/Users/joaopimenta/Desktop/GEE_test/globathy_bathymetry.tif"
# Load the raster data
lakeRst = rasterio.open(path)
lakeBottom = lakeRst.read(1)
# Raster resolution (in meters, assuming UTM projection)
resolution = lakeRst.res
print("Resolution:", resolution)
# Replace no-data value with np.nan
noDataValue = np.copy(lakeBottom[0, 0])
lakeBottom = lakeBottom.astype(float)
lakeBottom[lakeBottom == noDataValue] = np.nan
# Get the pixel size from raster resolution (in meters)
pixelArea = lakeRst.res[0] * lakeRst.res[1] # in square meters
# Calculate the area of the detected water bodies from binary mask
def calculate_area(image, transform):
# Mask the image to include only water
water_mask = image == 0
# Compute the area in square meters
pixel_area = abs(transform[0] * transform[4]) # pixel size (in square meters)
water_area_pixels = np.sum(water_mask)
total_area_m2 = water_area_pixels * pixel_area
return total_area_m2
# Define function to create mask for a specific elevation
def createMaskForElevation(elevation, elevDem):
mask = np.where(elevDem <= elevation, 1, 0) # White pixels for inundated area
return mask
# Arrays to store the areas for each elevation step
area_normal_Array = []
# Plot all polygons representing water area for each elevation step
fig, ax = plt.subplots(figsize=(12, 10))
# Colors for different elevation levels
colors = plt.cm.viridis(np.linspace(0, 1, len(elevSteps)))
# Loop over each elevation step, calculate area, and plot polygons
for i, elev in enumerate(elevSteps):
# Create a mask for the current elevation step
mask = createMaskForElevation(elev, lakeBottom)
# Calculate the water area at this elevation
waterArea = np.sum(mask) * pixelArea # sum of all '1' pixels * pixel area
area_normal_Array.append(waterArea) # Store the area for this elevation
# Find contours (polygons) from the mask
contours = measure.find_contours(mask, 0.5)
# Plot each contour as a polygon
for contour in contours:
# Transform contour coordinates to UTM coordinates using the raster transform
utm_coords = rasterio.transform.xy(lakeRst.transform, contour[:, 0], contour[:, 1])
x_coords, y_coords = np.array(utm_coords[0]), np.array(utm_coords[1])
# Plot the polygon for the current elevation step
ax.plot(x_coords, y_coords, color=colors[i], label=f'Elevation {elev} m' if i == 0 else "")
# Plot the elevation vs area
# Multiply the area by 1,000,000 to convert from km² to m² if necessary
area_normal_ArraySqM = [area * 1e8 for area in area_normal_Array] # Convert to square meters
# Function to create a binary mask for the chosen elevation
def createMaskForElevation(elevation, elevDem, resolution):
# Step 1: Generate the initial binary mask (1 for water, 0 for no water)
mask = np.where(elevDem <= elevation, 1, 0)
# Step 2: Fill the holes inside the lake region
filled_mask = binary_fill_holes(mask) * mask # Fill holes only inside the mask
# Step 3: Create a mask for the lake region (anything inside the boundary is considered lake)
lake_mask = np.where(np.isnan(elevDem), 1, 0) # NaN represents outside the lake
# Step 4: Assign a value of 1 to everything outside the lake region
result_mask = np.where(lake_mask == 1, 1, filled_mask)
# Step 5: Calculate the inundated area for the white pixels inside the lake
area = np.sum(filled_mask) * resolution[0] * resolution[1]
return result_mask, area
# Ensure differences, areaArraySqM, and elev are arrays or lists
differences = []
for area in areaArraySqM:
dif = abs(water - area) # Absolute difference
differences.append(dif)
print(f"The water area without the cloud pixels is: {water}")
# Find the index of the smallest difference
best_match_index = differences.index(min(differences))
best_match = area_ArraySqM[best_match_index]
# Reverse the elevation steps and convert to a list to allow indexing
step_elevation_reversed = list(reversed(elevSteps))
# Allow user to input a specific elevation based on the best match index
specificElevation = step_elevation_reversed[best_match_index]
# Generate the binary mask and calculate the area for the selected elevation
maskForSpecificElevation, specificArea = createMaskForElevation(specificElevation, lakeBottom, resolution)
# Load NDWI image and bathymetry mask
ndwi_dataset = rasterio.open(path_ndwi) # Path to NDWI image
ndwi_crs = ndwi_dataset.crs
ndwi_transform = ndwi_dataset.transform
ndwi_res = ndwi_dataset.res
# Ensure the mask for the specific elevation is reprojected to the NDWI's CRS, extent, and resolution
mask_for_elevation_reprojected = np.empty_like(ndwi_dataset.read(1))
reproject(
source=maskForSpecificElevation, # Mask to reproject
destination=mask_for_elevation_reprojected,
src_transform=lake_transform, # Transform from the bathymetry mask
src_crs=lake_crs, # CRS of the mask
dst_transform=ndwi_transform, # NDWI transform
dst_crs=ndwi_crs, # NDWI CRS
resampling=Resampling.nearest # Nearest neighbor interpolation for binary masks
)
# Now overlay the NDWI image with the mask
ndwi_image = ndwi_dataset.read(1) # Read the NDWI image (band 1)
# Assign value 1 to NDWI where mask is 1
ndwi_image[mask_for_elevation_reprojected == 0] = 1
# Save the NDWI image locally as a GeoTIFF
output_path = '/Users/joaopimenta/Desktop/GEE_test/reconstructed_plygon.tif' # Define the output path for the saved image
# Retrieve the metadata from the NDWI dataset to use it for saving the file
meta = ndwi_dataset.meta.copy()
# Update metadata for a single band output
meta.update({
'dtype': 'float32', # or 'uint8' depending on the NDWI data type
'count': 1, # Number of bands
'driver': 'GTiff', # Save as a GeoTIFF file
'crs': ndwi_crs, # Coordinate reference system
'transform': ndwi_transform # Affine transform for georeferencing
})
# Save the NDWI image with the mask applied as a GeoTIFF
with rasterio.open(output_path, 'w', **meta) as dst:
dst.write(ndwi_image.astype('float32'), 1) # Write the NDWI data to band 1
print(f'Saved NDWI image as {output_path}')
# Create a figure with 2 subplots side by side
fig, axes = plt.subplots(1, 2, figsize=(12, 6)) # 1 row, 2 columns
# Plot the first contour: Identified contour (K-means method) on binary_ndwi
contour_image_binary = np.zeros_like(binary_ndwi)
contours_binary, _ = cv2.findContours(binary_ndwi, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
if contours_binary:
cv2.drawContours(contour_image_binary, [max(contours_binary, key=cv2.contourArea)], -1, (255), 2)
axes[0].imshow(contour_image_binary, cmap='gray')
axes[0].set_title('Polygon of the NDWI affected by clouds')
# Plot the second contour: Identified contour (K-means method) on ndwi_image
contour_image_ndwi = np.zeros_like(ndwi_image)
contours_ndwi, _ = cv2.findContours(ndwi_image, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
if contours_ndwi:
cv2.drawContours(contour_image_ndwi, [max(contours_ndwi, key=cv2.contourArea)], -1, (255), 2)
axes[1].imshow(contour_image_ndwi, cmap='gray')
axes[1].set_title('Polygon of the reconstructed Image')
# Show the plots
plt.tight_layout()
plt.show()
# Create contour image for binary NDWI
contour_image_binary = np.zeros_like(binary_ndwi)
contours_binary, _ = cv2.findContours(binary_ndwi, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
if contours_binary:
cv2.drawContours(contour_image_binary, [max(contours_binary, key=cv2.contourArea)], -1, (255), 2)
# Create contour image for NDWI image
contour_image_ndwi = np.zeros_like(ndwi_image)
contours_ndwi, _ = cv2.findContours(ndwi_image, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
if contours_ndwi:
# Draw only the largest contour for the NDWI image
largest_contour = max(contours_ndwi, key=cv2.contourArea)
cv2.drawContours(contour_image_ndwi, [max(contours_ndwi, key=cv2.contourArea)], -1, (255), 2)
# Stack the binary contour and NDWI contour into a 3-channel image for overlay (Red, Green, Blue)
overlay_image = np.zeros((contour_image_binary.shape[0], contour_image_binary.shape[1], 3), dtype=np.uint8)
overlay_image[..., 0] = contour_image_binary # Red channel for binary NDWI contour
overlay_image[..., 1] = contour_image_ndwi # Green channel for NDWI image contour
# Create a mask to fill the inside of the largest NDWI contour
mask = np.zeros_like(ndwi_image, dtype=np.uint8)
if contours_ndwi:
cv2.drawContours(mask, [largest_contour], -1, (255), thickness=cv2.FILLED)
# Create a new output image, initialized to zeros (0 for a black image)
output_image = np.zeros_like(ndwi_image, dtype=np.uint8)
# Set the inside of the largest contour to one (255)
output_image[mask == 255] = 1 # Change to fill with pixel value 1
# Optionally convert to uint8 range for visualization
output_image *= 255 # If you need the output image to be in the 0-255 range
# Save the NDWI image locally as a GeoTIFF
output_path = '/Users/joaopimenta/Desktop/GEE_test/reconstructed_water_mask.tif' # Define the output path for the saved image
# Retrieve the metadata from the NDWI dataset to use it for saving the file
meta = ndwi_dataset.meta.copy()
# Update metadata for a single band output
meta.update({
'dtype': 'float32', # or 'uint8' depending on the NDWI data type
'count': 1, # Number of bands
'driver': 'GTiff', # Save as a GeoTIFF file
'crs': ndwi_crs, # Coordinate reference system
'transform': ndwi_transform # Affine transform for georeferencing
})
# Save the NDWI image with the mask applied as a GeoTIFF
with rasterio.open(output_path, 'w', **meta) as dst:
dst.write(output_image.astype('float32'), 1) # Write the NDWI data to band 1
print(f'Saved NDWI image as {output_path}')
# If resolution is a tuple (x_resolution, y_resolution)
x_resolution, y_resolution = resolution
pixel_area = x_resolution * y_resolution # Area of one pixel in square meters
# Count water pixels
water_pixels = ndwi_image > 0
water_pixel_count = np.sum(water_pixels)
# Calculate total water area
total_water_area = water_pixel_count * pixel_area*1e8
# Print the result
print(f'Total water area: {total_water_area} square meters')
water_area_info.append(total_water_area)
# Optionally, save the modified NDWI image as a new file
out_meta = ndwi_dataset.meta.copy()
with rasterio.open('ndwi_with_elevation_mask.tif', 'w', **out_meta) as dst:
dst.write(ndwi_image, 1) # Write the new image to disk
# Close datasets
ndwi_dataset.close()
else:
print("There are no cloud pixels inside the reservoir's area.")
water_area_info.append(waterarea)
st.write(water_area_info)
else:
# Options for confirming the reservoir selection
water_method = ['Fixed thershold', 'Dynamic']
# Select box to confirm selection
water = st.selectbox("Choose this method of identifying water pixels", water_method)
def extract_bbox_from_aoi(aoi):
# Get the bounding box of the AOI
bounds = aoi.bounds().getInfo()
# Extract the coordinates based on the observed structure
try:
lon_min = bounds['coordinates'][0][0][0] # First point's longitude
lat_min = bounds['coordinates'][0][0][1] # First point's latitude
lon_max = bounds['coordinates'][0][2][0] # Third point's longitude
lat_max = bounds['coordinates'][0][2][1] # Third point's latitude
return lat_min, lon_min, lat_max, lon_max
except (IndexError, KeyError, TypeError):
print("Unexpected bounds structure:", bounds)
raise ValueError("Unable to extract bounding box; check structure of bounds data.")
lat_min, lon_min, lat_max, lon_max = extract_bbox_from_aoi(roi)
# Function to query bridges
def check_bridge_in_area(lat_min, lon_min, lat_max, lon_max):
overpass_url = "http://overpass-api.de/api/interpreter"
overpass_query = f"""
[out:json];
(
way["man_made"="bridge"]({lat_min},{lon_min},{lat_max},{lon_max});
node(w);
);
out body qt;
"""
print(f"Querying Overpass API with:\n{overpass_query}")
response = requests.get(overpass_url, params={'data': overpass_query})
if response.status_code == 200:
print("Received response from Overpass API.")
return response.json()
else:
print(f"Error: {response.status_code}")
return None
# Query the Overpass API for bridges in the bounding box
bridge_data = check_bridge_in_area(lat_min, lon_min, lat_max, lon_max)
# Initialize lists for GeoDataFrame
bridge_names = []
elem = None
# Parse and display bridges with their shapes and names
if bridge_data and 'elements' in bridge_data:
node_coords = { # Store node coordinates for reference
int(node['id']): (node['lat'], node['lon'])
for node in bridge_data['elements']
if node['type'] == 'node'
}
if len(node_coords) > 0:
print(f"Node coordinates found: {node_coords}")
else:
print("No node coordinates found.")
for element in bridge_data['elements']:
if element['type'] == 'way':
elem = element['type']
bridge_name = element.get('tags', {}).get('name')
if bridge_name: # Check if bridge has a name
st.write(f"It was detected the bridge {bridge_name} inside the ROI")
coords = [node_coords.get(node_id) for node_id in element.get('nodes', [])]
coords = [coord for coord in coords if coord is not None] # Remove invalid coordinates
else:
st.write(f"Bridge ID: {element['id']} has no name and will be excluded.")
else:
st.write("No bridge data or 'elements' not in the response.")
if water == 'Fixed thershold':
# Define a function to calculate NDWI and mask for each image
def calculate_ndwi_and_mask(image):
ndwi = image.normalizedDifference(['B3', 'B8']).rename('NDWI')
ndwi_threshold = ndwi.gte(0.0)
ndwi_mask = ndwi_threshold.updateMask(ndwi_threshold)
return ndwi_mask
# Map the function over the image collection to get NDWI masks for each image
ndwi_masks = sentinelImageCollection.map(calculate_ndwi_and_mask)
# Perform erosion (shrinking the mask slightly to remove small gaps and noise)
eroded_ndwi = ndwi_masks.map(lambda img: img.focal_min(radius=1, kernelType='circle', iterations=1))
# Perform dilation after erosion (expanding the mask back to restore shape)
closed_ndwi = eroded_ndwi.map(lambda img: img.focal_max(radius=1, kernelType='circle', iterations=1))
# Now, closed_ndwi contains the NDWI masks that have been eroded and then dilated for each image in the collection.
# Define a function to calculate water area
def calculate_water_area(image):
water_area = image.multiply(ee.Image.pixelArea()).reduceRegion(
reducer=ee.Reducer.sum(),
geometry=roi,
bestEffort=True,
scale=5
).get('NDWI')
return image.set('water_area', water_area)
if elem is not None:
# Map the function over the NDWI masks to calculate water area for each image
ndwi_masks_with_area = closed_ndwi.map(calculate_water_area)
else:
# Map the function over the NDWI masks to calculate water area for each image
ndwi_masks_with_area = ndwi_masks.map(calculate_water_area)
# Get the water area information
water_area_info = ndwi_masks_with_area.aggregate_array('water_area').getInfo()
# Display the list of water areas
#st.write(water_area_info)
# Get acquisition dates in human-readable format
dates = ndwi_masks_with_area.aggregate_array('system:time_start') \
.map(lambda d: ee.Date(d).format('YYYY-MM-dd')).getInfo()
# Display the dates
#st.write("Acquisition dates for each image:", dates)
# Alternatively, convert acquisition times to readable format (if needed)
acquisition_times = sentinelImageCollection.aggregate_array('system:time_start').getInfo()
acquisition_dates = [datetime.datetime.utcfromtimestamp(time / 1000).strftime('%Y-%m-%d') for time in acquisition_times]
#st.write("Alternative acquisition dates:", acquisition_dates)
else:
# Define a function to calculate NDWI
def calculate_ndwi(image):
ndwi = image.normalizedDifference(['B3', 'B8']).rename('NDWI')
return image.addBands(ndwi)
# Define a function to sample NDWI values for clustering
def sample_ndwi(image):
ndwi = image.select('NDWI')
sampled_ndwi = ndwi.sample(
region=roi_geometry,
scale=10,
numPixels=10000,
seed=0
).select('NDWI')
return sampled_ndwi
# Define a function to perform K-means clustering
def cluster_ndwi(sampled_ndwi):
clusterer = ee.Clusterer.wekaKMeans(2).train(sampled_ndwi)
return clusterer
# Define a function to determine the water cluster
def get_water_cluster(clustered_image):
mean_ndwi_per_cluster = clustered_image.reduceRegion(
reducer=ee.Reducer.mean(),
geometry=roi_geometry,
scale=10
)
mean_values = ee.List(mean_ndwi_per_cluster.values())
water_cluster = mean_values.indexOf(mean_values.reduce(ee.Reducer.max()))
return water_cluster
# Define a function to create a binary water mask based on the cluster
def create_water_mask(clustered_image, water_cluster):
water_mask = clustered_image.eq(water_cluster).rename('water_mask')
return water_mask
# Define a function to compute the area of water bodies in square meters
def compute_water_area(water_mask):
water_area = water_mask.reduceRegion(
reducer=ee.Reducer.sum(),
geometry=roi_geometry,
scale=10
).get('water_mask')
water_area = ee.Number(water_area).multiply(100).divide(1e4) # Convert to square kilometers
return water_area
if elem is not None:
# Combine all functions into one for mapping
def process_image(image):
# Calculate NDWI
image = calculate_ndwi(image)
# Sample and cluster NDWI for water detection
sampled_ndwi = sample_ndwi(image)
clusterer = cluster_ndwi(sampled_ndwi)
clustered_image = image.select('NDWI').cluster(clusterer).rename('cluster')
# Determine which cluster represents water
water_cluster = get_water_cluster(clustered_image)
water_mask = create_water_mask(clustered_image, water_cluster)
# Perform morphological operations (closing)
eroded_ndwi = water_mask.focal_min(radius=1, kernelType='circle', iterations=1)
closed_ndwi = eroded_ndwi.focal_max(radius=1, kernelType='circle', iterations=1)
water_area = compute_water_area(closed_ndwi)
return image.set('water_area_km2', water_area)
else:
# Combine all functions into one for mapping
def process_image(image):
image = calculate_ndwi(image)
sampled_ndwi = sample_ndwi(image)
clusterer = cluster_ndwi(sampled_ndwi)
clustered_image = image.select('NDWI').cluster(clusterer).rename('cluster')
water_cluster = get_water_cluster(clustered_image)
water_mask = create_water_mask(clustered_image, water_cluster)
water_area = compute_water_area(water_mask)
return image.set('water_area_km2', water_area)
# Apply the processing function to each image in the collection
processed_images = sentinelImageCollection.map(process_image)
# Extract the water area and date information
water_area_info = processed_images.aggregate_array('water_area_km2').getInfo()
dates = processed_images.aggregate_array('system:time_start').map(lambda d: ee.Date(d).format('YYYY-MM-dd')).getInfo()
# Get acquisition times of the images
acquisition_times = sentinelImageCollection.aggregate_array('system:time_start').getInfo()
# Convert acquisition times to human-readable dates
acquisition_dates = [datetime.datetime.utcfromtimestamp(time / 1000).strftime('%Y-%m-%d') for time in acquisition_times]
if method == ("Write the A-V function of your reservoir"):
for area in water_area_info:
volume = None
# Calculate the volume using the area-storage equation
volume = (area / a) ** (1 / b)
if volume is not None:
volumes.append(volume)
else:
st.write("Error with the coefficients")
st.write(f"The list of the volumes in cubic meters for the chosen dates is: {volumes}")
elif method == ("upload excel sheet"):
if dictionary:
for area in water_area_info:
volume = None
keys = sorted(dictionary.keys())
for i in range(len(keys)):
key = keys[i]
if key >= area:
if i == 0:
volume = dictionary[key]
st.write(f"This is the volume {volume/10**6}km³")
volumes.append(volume)
else:
prev_key = keys[i - 1]
delta_volume = dictionary[key] - dictionary[prev_key]
delta_key = key - prev_key
delta_area = area - prev_key
interpolated_volume = dictionary[prev_key] + (delta_volume * delta_area / delta_key)
volume = (interpolated_volume/10**6)
st.write(f"This is the volume {volume}km³")
volumes.append(volume)
break
else:
# This else block belongs to the for loop, not the if condition
st.write("Dam value not found in the dictionary ")
elif method == ("upload the DEM"):
if dictionary:
for area in water_area_info:
volume = None
keys = sorted(dictionary.keys())
for i in range(len(keys)):
key = keys[i]
if key >= area:
if i == 0:
volume = dictionary[key]
st.write(f"This is the volume {volume/10**6}km³")
volumes.append(volume)
else:
prev_key = keys[i - 1]
delta_volume = dictionary[key] - dictionary[prev_key]
delta_key = key - prev_key
delta_area = area - prev_key
interpolated_volume = dictionary[prev_key] + (delta_volume * delta_area / delta_key)
volume = (interpolated_volume/10**6)
st.write(f"This is the volume {volume}km³")
volumes.append(volume)
break
else:
# This else block belongs to the for loop, not the if condition
st.write("Dam value not found in the dictionary ")
elif method == "Don't have that info":
import netCDF4 as nc
import numpy as np
volumes =[]
# Open the NetCDF file
nc_file = nc.Dataset('/Users/joaopimenta/Downloads/Master thesis/GLOBathy_hAV_relationships.nc')
# Specify the lake ID you want to search for
target_lake_id = hydrolakes_id # Replace this with the actual lake ID you're interested in
# Find the index of the lake based on the lake ID
lake_ids = nc_file.variables['lake_id'][:]
# Check if the target lake ID exists in the lake_id variable
lake_index = np.where(lake_ids == target_lake_id)[0]
if len(lake_index) == 0:
st.write("Lake not found in the dataset.")
else:
lake_index = lake_index[0] # Use the first match if found
# Extract coefficients of the area-storage equation for the identified lake
area_storage_coeffs = nc_file.variables['f_hA'][lake_index, :]
lon_lat = nc_file.variables['lon_lat'][lake_index, :]
# Print the lake's ID, coordinates, and area-storage equation coefficients
st.write("Coordinates (Lon, Lat):", lon_lat)
# Print the coefficients
st.write("Area-Storage equation coefficients:")
st.write("a:", area_storage_coeffs[0])
st.write("b:", area_storage_coeffs[1])
st.write("R^2:", area_storage_coeffs[2])
import numpy as np
for area in water_area_info:
# Coefficients obtained from the NetCDF dataset
a = area_storage_coeffs[0]
b = area_storage_coeffs[1]
# Calculate the volume using the area-storage equation
volume = ((area/1e6) / a) ** (1 / b)
volumes.append(volume)
from io import BytesIO
# Function to generate the sample data DataFrame
def generate_sample_data():
date = acquisition_dates
area = water_area_info
vol = volumes
return pd.DataFrame({'Date': date, 'Volume (10⁸m³)': vol, 'Area (km²)': area })
# Generate the sample data DataFrame
df = generate_sample_data()
# Save the DataFrame to an Excel file in memory
excel_buffer = BytesIO()
import pandas as pd
import io
# Assuming excel_buffer and output, area_storage_coeffs are defined elsewhere in your code
excel_buffer = io.BytesIO()
# Use a single `ExcelWriter` for writing all sheets
with pd.ExcelWriter(excel_buffer, engine='xlsxwriter') as writer:
# Check if 'Water Surface Elevation' is in output and write relevant data
if 'Water Surface Elevation' in output:
# Convert date to timezone-unaware if necessary
elevation_dates = pd.to_datetime(df_filtered['time_str']).dt.tz_localize(None)
elevations = df_filtered['wse']
df_2 = pd.DataFrame({'Date': elevation_dates, 'Water Surface Elevations': elevations})
df_2.to_excel(writer, sheet_name='Elevations Data', index=False)
# Check if 'Storage-Capacity curve' is in output and write relevant data
if 'Storage-Capacity curve' in output:
# Assume `area_storage_coeffs` contains appropriate data in tuple or list format
df_3 = pd.DataFrame({
'a': [area_storage_coeffs[0]],
'b': [area_storage_coeffs[1]],
'R^2': [area_storage_coeffs[2]]
})
df_3.to_excel(writer, sheet_name='Storage_capacity_curve', index=False)
# Assuming `df` is a base DataFrame you want to write to a default sheet
df.to_excel(writer, sheet_name='Reservoir Data', index=False)
# Reset buffer position to the start for reading/download
excel_buffer.seek(0)
# Create a download button for the Excel file
st.download_button(
label="Download Excel file",
data=excel_buffer,
file_name="reservoir_data.xlsx",
mime="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
)
st.write("Click the button above to download the data as an Excel file.")
import tempfile
# Load the bathymetry dataset from Earth Engine
globathy = ee.Image("projects/sat-io/open-datasets/GLOBathy/GLOBathy_bathymetry")
# Define the function to export the image and return the path
def export_image_for_download(image, roi, scale=10):
# Use a temporary directory for saving the file
with tempfile.NamedTemporaryFile(delete=False, suffix=".tif") as temp_file:
out_image_path = temp_file.name
geemap.ee_export_image(image, filename=out_image_path, scale=scale, region=roi)
return out_image_path
if 'Bathymetry file' in output:
# Call the function and set up the download button
if st.button("Download Image"):
# Assuming `globathy` is your Earth Engine image and `roi` is the region of interest
image_path = export_image_for_download(globathy, roi)
# Read the file as bytes for download
with open(image_path, "rb") as file:
file_bytes = file.read()
st.download_button(
label="Click here to download the image",
data=file_bytes,
file_name="exported_image.tif",
mime="image/tiff"
)
# Display the bar charts
col1, col2= st.columns([7,3])
# Combine acquisition dates and volumes into a list of tuples
with col1:
import pandas as pd
import altair as alt
# Function to generate sample data
def generate_sample_data():
date = acquisition_dates
vol= volumes
return pd.DataFrame({'Date': date, 'Volume': vol})
# Sample data
df = generate_sample_data()
if 'Water Surface Elevation' in output:
st.subheader("WSE over the chosen date range")
# Convert 'time_str' to timezone-unaware and set up DataFrame for plotting
elevation_dates = pd.to_datetime(df_filtered['time_str']).dt.tz_localize(None)
elevations = df_filtered['wse']
df_wse = pd.DataFrame({'Date': elevation_dates, 'Water Surface Elevations': elevations})
# Create the line chart for water surface elevations
wse_chart = alt.Chart(df_wse).mark_line(
color='#00FFFF'
).encode(
x=alt.X('Date:T', title='Date'),
y=alt.Y('Water Surface Elevations:Q', title='Water Surface Elevation (m)')
)
# Display the WSE chart in Streamlit
st.altair_chart(wse_chart, use_container_width=True)
st.subheader("Volume over the chosen date range")
# Ensure that 'Date' and 'Volume' columns are available in df
volume_chart = alt.Chart(df).mark_line(
color='#00FFFF'
).encode(
x=alt.X('Date:T', title='Date'),
y=alt.Y('Volume:Q', title='Volume (10⁶ m³)',scale=alt.Scale(zero=False))
)
# Display the volume chart in Streamlit
st.altair_chart(volume_chart, use_container_width=True)
with col2:
import pandas as pd
# Donut chart function
def make_donut(input_response, input_text, input_color):
if input_color == 'green':
chart_color = ['#27AE60', '#12783D']
elif input_color == 'red':
chart_color = ['#E74C3C', '#781F16']
elif input_color == 'yellow':
chart_color = ['#FFFF00', '#FFD700'] # Yellow colors
elif input_color == 'orange':
chart_color = ['#FFA500', '#FF4500'] # Orange colors
elif input_color == 'light green':
chart_color = ['#90EE90', '#006400'] # Light green colors
else:
raise ValueError("Invalid color. Please choose either 'green' or 'red'.")
source = pd.DataFrame({
"Topic": ['', input_text],
"% value": [100-input_response, input_response]
})
source_bg = pd.DataFrame({
"Topic": ['', input_text],
"% value": [100, 0]
})
plot = alt.Chart(source).mark_arc(innerRadius=45, cornerRadius=25).encode(
theta="% value",
color=alt.Color("Topic:N",
scale=alt.Scale(
domain=[input_text, ''],
range=chart_color),
legend=None),
).properties(width=130, height=130)
text = plot.mark_text(align='center', color=chart_color[0], font="sans-serif", fontSize=20, fontWeight=500, fontStyle="italic").encode(text=alt.value(f'{input_response} %'))
plot_bg = alt.Chart(source_bg).mark_arc(innerRadius=45, cornerRadius=20).encode(
theta="% value",
color=alt.Color("Topic:N",
scale=alt.Scale(
domain=[input_text, ''],
range=chart_color), # 31333F
legend=None),
).properties(width=130, height=130)
return plot_bg + plot + text
def get_color(value):
"""Helper function to determine the color based on percentage."""
if value < 25:
return 'red'
elif 25 <= value < 50:
return 'orange'
elif value == 50:
return 'yellow'
elif 50 < value < 75:
return 'light green'
else:
return 'green'
# Check if storage is not None, not an empty string, and can be converted to a float
if Vol_res is not None:
try:
storage_float = Vol_res/10
if storage_float > 0:
total_volume = storage_float
worst = (min(volumes) / total_volume) * 100
best = (max(volumes) / total_volume) * 100
# Colors for worst and best day
wrst_color = get_color(worst)
bst_color = get_color(best)
# Display donut charts
st.subheader("Lower storage")
st.altair_chart(make_donut(round(worst, 2), 'Worst day', wrst_color), use_container_width=True)
st.subheader("Higher storage")
st.altair_chart(make_donut(round(best, 2), 'Best day', bst_color), use_container_width=True)
except ValueError:
st.write("Invalid storage value; cannot convert to float.")
# Fallback if storage is invalid or not provided, and ref_area is available
elif properties and ref_area is not None:
ref_area_float = (float(ref_area)*1e6)
worst = (min(water_area_info) / ref_area_float) * 100
best = (max(water_area_info) / ref_area_float) * 100
# Colors for worst and best day
wrst_color = get_color(worst)
bst_color = get_color(best)
# Display fallback donut charts
st.subheader("Lower storage")
st.altair_chart(make_donut(round(worst, 2), 'Worst day', wrst_color), use_container_width=True)
st.subheader(" Higher storage")
st.altair_chart(make_donut(round(best, 2), 'Best day', bst_color), use_container_width=True)
def calculate_max_percentage_variation(volumes, acquisition_dates):
max_variation = 0
max_variation_index = None
for i in range(1, len(volumes)):
# Calculate percentage variation
percentage_variation = abs((volumes[i] - volumes[i - 1]) / volumes[i - 1]) * 100
# Update max variation and index if current variation is greater
if percentage_variation > max_variation:
max_variation = percentage_variation
max_variation_index = i - 1 # Store index of the first date in the pair
# Get the dates corresponding to the max variation
date1 = acquisition_dates[max_variation_index]
date2 = acquisition_dates[max_variation_index + 1]
return max_variation, date1, date2
max_variation, date1, date2 = calculate_max_percentage_variation(volumes, acquisition_dates)
import statistics
std_dev = round(statistics.stdev(volumes),2)
mean_vol = round(statistics.mean(volumes),2)
mean_area = round(statistics.mean(water_area_info),2)
max_variation_area, date1, date2 = calculate_max_percentage_variation(water_area_info, acquisition_dates)
#st.write(" The greates variation occured between:", date1, "and", date2)
if max_variation >=0:
delta = 1
else:
delta = -1
max_area = max(water_area_info)
max_volume = max(volumes)
current_volume = round(volumes[-1],2)
current_area = round(water_area_info[-1],2)
#create column span
col1, col2, col3 = st.columns(3)
#Customize metric style to have white text color
metric_style = "color: black;"
col1.metric(label="Max variation", value="%" + " " + f"{max_variation:,.2f}", delta=delta)
col2.metric(label="Mean area", value="km²" + " " + f"{mean_area/1e6:,.2f}", delta=round(max_variation_area,2))
col3.metric(label="standard deviation", value = "km³ " + " " + f"{std_dev:,.2f}")
st.write(" The greates variation occured between:", date1, "and", date2)
# Create a select box for asking ChatGPT about this reservoir
opt = ["No","Yes"]
info = st.selectbox(
f"Ask ChatGPT more information about this lake {lake_name}",
opt,
key="info")
if info == "Yes":
import streamlit as st
from openai import OpenAI
# Define the reservoir information retrieval function
def get_reservoir_info(lake_name, country, lng, lat):
client = OpenAI(api_key="sk-proj-uK1IbwMXiNImV7RFDxr3T3BlbkFJXjBzHIIJYRoQbiJE6Kc5")
prompt = (
f"Provide detailed information about the dam/reservoir {lake_name}, "
f"located in {country}, with coordinates (longitude: {lng}, latitude: {lat}). "
f"Use short sentences and split the response into concise, clear lines."
f"Specify the type of use of that reservoir or lake and when it wast built. "
)
stream = client.chat.completions.create(
model="gpt-3.5-turbo",
messages=[{"role": "user", "content": prompt}],
stream=True,
)
response_content = ""
for chunk in stream:
if chunk.choices[0].delta.content is not None:
response_content += chunk.choices[0].delta.content
return response_content
st.subheader(f"Information about the reservoir {lake_name}")
info_text = get_reservoir_info(lake_name, Country, lng, lat)
# Displaying the information with line breaks
for line in info_text.split(". "): # Split into sentences
st.text(line.strip() + ".")
else:
st.write("Please select the date range and cloud coverage thershold for the analysis.")
else:
st.write("Please search on the map the lake you want to analyse and click on it to select it")
elif page =="About":
from PIL import Image
# For handling images
# Load images
satellite_image = Image.open("/Users/joaopimenta/Desktop/images thesis and figures/Captura de ecrã 2024-08-29, às 18.17.19.png")
workflow_diagram = Image.open("/Users/joaopimenta/Downloads/image-Photoroom.png")
reservoir_map = Image.open("/Users/joaopimenta/Desktop/images thesis and figures/Captura de ecrã 2024-08-30, às 00.43.24.png")
worflow_image = Image.open("/Users/joaopimenta/Desktop/SCR-20241218-bmyx.png")
#comparison_graph = Image.open("images/comparison_graph.jpg")
#future_advancements = Image.open("images/future_advancements.jpg")
# Path to your video file
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(" ")
st.markdown(
"""
""",
unsafe_allow_html=True
)
# Content with the "about-text" class for styling
st.markdown(
"""
Efficient management of water reservoirs is essential for water security, flood control, and hydroelectric power generation. Traditional methods of evaluating reservoir volumes depend on in-situ measurements and physical surveys, which are often time-consuming, resource-intensive, and impractical for many regions due to financial and logistical limitations.
To address these challenges, this research introduces a novel remote sensing tool designed to provide an accurate, scalable, and globally accessible method for reservoir volume evaluation. The tool integrates high-resolution Sentinel-2 satellite imagery with geospatial analysis techniques and machine learning algorithms to automatically calculate inundated areas and reservoir water storage.
This problem is both significant and complex. Reservoir volume measurements are critical for effective water resource management, yet current methodologies struggle to scale for large or remote areas due to high costs. Furthermore, environmental factors such as cloud cover and human-made structures, like bridges, can interfere with satellite imagery, presenting additional challenges for large-scale remote sensing. By employing the algorithms developed in this study, the proposed tool overcomes these barriers, delivering flexible and reliable estimates of reservoir volumes.
The solution was tested on reservoirs in Portugal and California, USA, achieving high accuracy. Results showed an average mean absolute percentage error of 5.35% and an average correlation coefficient (R²) of 0.90 when compared to published data obtained through traditional methods.
This software includes a free demo version designed to allow users to test the tool, with continuous improvements planned. Currently, it utilizes a sample of polygons from the SWOT lakes database, covering 93% of all lakes across Europe. Over time, the database will be expanded, incorporating additional data, while also optimizing the website's RAM usage.
Beyond the current promising results, this research opens pathways for future enhancements. These include:
The Reservoir Volume Monitoring Application is a cutting-edge tool designed to address global challenges in water management, flood prevention, and hydroelectric power optimization. By leveraging high-resolution satellite imagery from Sentinel-2 and advanced geospatial analysis, the app automates the estimation of reservoir volumes, delivering accurate, efficient, and globally scalable solutions.This software includes a free demo version designed to allow users to test the tool, with continuous improvements planned. Currently, it utilizes a sample of polygons from the SWOT lakes database, covering 93% of all lakes across Europe. Over time, the database will be expanded, incorporating additional data, while also optimizing the website's RAM usage.
This application simplifies complex workflows into an intuitive and automated process. Users start by selecting a Region of Interest (ROI) using an interactive map or by uploading geojson files. Satellite imagery for the specified area and date range is automatically retrieved and preprocessed. Advanced algorithms remove interferences such as clouds, shadows, or structural obstacles like bridges. The app applies indices such as NDWI (Normalized Difference Water Index) to classify water pixels and calculate inundated areas. For reservoirs affected by cloud cover, bathymetric data from global databases like GLOBathy is used to reconstruct accurate water surfaces. Finally, volumes are calculated using area-volume relationships derived from either existing databases or user-provided data. The results are visualized through dynamic charts, maps, and downloadable reports, providing users with actionable insights.
Reservoirs are crucial for ensuring water security, supporting agriculture, producing hydroelectric power, and maintaining ecological balance. However, many regions lack efficient monitoring tools, leading to water mismanagement and heightened risks of droughts, floods, and ecological disruption. This application bridges the gap by offering a globally accessible solution that requires no physical infrastructure. Its scalability enables it to monitor reservoirs of all sizes, from local irrigation ponds to massive hydroelectric reservoirs. With an accuracy of 94.65%, tested on reservoirs in Portugal and California, the app provides reliable data to support decision-making in water management, flood risk mitigation, and ecosystem preservation.
The application offers a robust yet user-friendly platform that integrates cutting-edge technology into a seamless user experience. Built with Python and the Streamlit framework, it provides: • Automated Image Processing: Reduces manual effort by leveraging advanced algorithms for water surface detection and volume calculation. • Accurate Data Insights: Achieves a mean absolute percentage error of just 5.35%. • Global Scalability: Access anywhere, monitor reservoirs of all sizes, and ensure reliable data even in remote areas. • Environmental Sustainability: Supports efficient water use and better management of natural resources.
The application is poised for future enhancements, including integration with new satellite missions like SWOT (Surface Water and Ocean Topography), predictive modeling using LSTM algorithms, and expanded compatibility with emerging geospatial technologies. With its scalable and adaptable design, this tool is set to become an indispensable resource for water resource management professionals, researchers, and environmental advocates.