hydroanomaly 1.2.7

A Python package for hydro anomaly detection with simple USGS data retrieval

HydroAnomaly

A Python package for water bodies anomaly detection. It retrieves USGS water data and Sentinel-2 bands to use in ML models for checking the quality of the data collected by USGS gages.

Installation

Python script:

pip install hydroanomaly

For Jupyter

!pip install hydroanomaly

For updating the package:

!pip install hydroanomaly --upgrade

---

USGS Data Retrieval

Easily retrieve real-time and historical turbidity of water from USGS Water Services:

import ee
import geemap
import hydroanomaly

# ------------------------
# User-defined settings: Example USGS site and date range (change to a site with turbidity data)
# ------------------------
site_number = "294643095035200"  # USGS site number
start_date = "2020-01-01"
end_date = "2024-12-30"

# ------------------------
# Data Extraction from USGS
# ------------------------
USGSdata, (lat, lon) = get_turbidity(site_number, start_date, end_date)
print("=" * 70)
print("Latitude:", lat)
print("Longitude:", lon)
print("=" * 70)
print(USGSdata.head())

---

Sentinel-2 Data Retrieval

Retrieve Sentinel data from Google Earth Engine

Defining the Google Earth Engine API

ee.Authenticate()
ee.Initialize(project='XXXXXX-XXXX-XXXXXX-XX') # Replace with your own project ID number

Defining settings, coordinates, masks, etc:

Defining the area from which you want to retrieve data:

latitude = 29.7785861
longitude = -95.0644278
bands = ['B2','B3','B4','B5','B6','B7','B8','B8A','B9','B11','B12', 'SCL']
buffer_meters = 20
cloudy_pixel_percentage = 20
masks_to_apply = [
    "water",
    "no_cloud_shadow",
    "no_clouds",
    "no_snow_ice",
    "no_saturated"
]

Sentinel-2 data retrieval:

from hydroanomaly import get_sentinel_bands

df = get_sentinel_bands(
    latitude=latitude,
    longitude=longitude,
    start_date=start_date,
    end_date=end_date,
    bands=bands,
    buffer_meters=buffer_meters,
    cloudy_pixel_percentage=cloudy_pixel_percentage,
    masks_to_apply=masks_to_apply
)

print(df.head())
print("=" * 70)
print(f"Retrieved {len(df)} rows")

Visualizing the map:

from hydroanomaly import show_sentinel_ndwi_map

Map = show_sentinel_ndwi_map(
    latitude, longitude, start_date, end_date,
    buffer_meters=buffer_meters, cloudy_pixel_percentage=cloudy_pixel_percentage, zoom=15)
Map

---

Time Series Plotting

Create visualizations of your water data:

from hydroanomaly.visualize import plot_timeseries
# For USGS data
plot_timeseries(USGSdata)
# For Sentinel data
plot_timeseries(df)

from hydroanomaly.visualize import plot_turbidity
# For USGS data
plot_turbidity(USGSdata)

from hydroanomaly.visualize import plot_sentinel
# For Sentinel data
plot_sentinel(df)

from hydroanomaly.visualize import plot_comparison
plot_comparison(USGSdata, df[['B6']], label1="Turbidity", label2="Sentinel-2 B6", title="Comparison: Turbidity vs Band 6")

from hydroanomaly import plot
# For Sentinel data
plot(df)

from hydroanomaly import visualize
# For USGS data
visualize(USGSdata)

NDVI:

import matplotlib.pyplot as plt
# Check available columns
print(df.columns)
print("=" * 70)
# Calculate NDVI if bands are available
if {'B4', 'B8'}.issubset(df.columns):
    df['NDVI'] = (df['B8'] - df['B4']) / (df['B8'] + df['B4'])
    df['NDVI'].plot(marker='o')
    plt.title("NDVI Time Series")
    plt.xlabel("Date")
    plt.ylabel("NDVI")
    plt.grid()
    plt.show()
else:
    print("NDVI bands (B4, B8) not found. Try plotting individual bands:")
    df[['B2', 'B3', 'B4', 'B8']].plot()
    plt.title("Sentinel-2 Reflectance (selected bands)")
    plt.ylabel("Reflectance")
    plt.xlabel("Date")
    plt.grid()
    plt.show()

---

Machine Learning for Anomaly Detection of USGS Data

print(df.columns)
print("=" * 70)
display(df.head(2))
print("=" * 70)
print(USGSdata.columns)
print("=" * 70)
display(USGSdata.head(2))

USGSdata = USGSdata[~USGSdata.index.duplicated(keep='first')]
print(df.index.duplicated().sum())                     # Number of duplicate datetimes in df
print(USGSdata.index.duplicated().sum())               # Number of duplicate datetimes in USGSdata

OneClassSVM

from hydroanomaly.ml import run_oneclass_svm
df_out, params, f1 = run_oneclass_svm(df, USGSdata)

# F1 Score
print(f"F1: {f1:.3f}")

Isolation Forest

from hydroanomaly.ml import run_isolation_forest
df_out_if, params_if, f1_if = run_isolation_forest(df, USGSdata)

print(f"F1: {f1_if:.3f}")

---

Features

• USGS & Sentinel-2 Data Retrieval

  • Download real-time and historical water data from USGS Water Services (any site, any parameter)
  • Retrieve Sentinel-2 satellite bands using Google Earth Engine for any location and time range
  • Automatic data cleaning, validation, and alignment between ground (USGS) and satellite (Sentinel) data
  • Synthetic data generation fallback for testing
  • Convenient CSV export functionality

• Time Series & Satellite Visualization

  • Quick plotting for single or multiple water quality parameters
  • Multi-parameter and multi-site comparison plots
  • Satellite band and index visualization (NDVI, NDWI, etc.)
  • Statistical analysis plots (histograms, box plots, trendlines)
  • High-quality plot export (PNG, PDF, SVG) with auto legends and formatting

• Machine Learning & Anomaly Detection

  • Built-in anomaly detection using One-Class SVM and Isolation Forest models
  • Visual comparison of predicted vs. true anomalies in time series data
  • Feature engineering for satellite and in-situ sensor data
  • Easy integration with Pandas workflows

• Powerful Data Analysis Tools

  • Mathematical operations and filtering for hydrologic data
  • Statistical summaries, validation, and automated quality checks
  • Utilities for matching, joining, and synchronizing time series

• Easy to Use

  • Simple, Pythonic API for rapid data exploration and analysis
  • One-liner data retrieval and plotting functions
  • Comprehensive error handling
  • Well-documented with step-by-step examples and tutorials

Find USGS site numbers at: https://waterdata.usgs.gov/nwis

---

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

HydroAnomaly - Making water data analysis simple and beautiful!