Irrigation Detection

This repository contains scripts for downloading, preprocessing, visualization, model creation and analysis for irrigation detection purpose. Main part will be implementing of irrigation detection method from Irrigation Events Detection over Intensively Irrigated Grassland Plots Using Sentinel-1 Data in V1. and from Near Real-Time Irrigation Detection at Plot Scale Using Sentinel-1 Data in V2.

This repository is under construction and functions may not work or be whack explained.

Most functions used inside jupyter_notebooks/... are stored in irrigation_detection.py and tools.py .

To use some functions additional Accounts are requiered:

Google Earth Engine mainly for processing and downloading Satellite data. You can apply here

International Soil Moisture Network to get in-situ soil moisture time series data. You can apply here

API's used:

earthengine-api | Python and JavaScript client libraries for calling the Google Earth Engine API

keras | Deep learning API running on top of the machine learning platform TensorFlow

Notebooks inside jupyter_notebooks folder:

01_SSM_1km_V1 Product Source: World

Load daily .nc files in nested directories as xarray. Subset and mask out unrelevant regions. Apply conversation from digital to physical values.

02_RADOLAN_1km Product Source: Germany

Download Radolan SF product from FTP server. Exctract files. Load extracted files to xarray. Subset data. Apply conversation from digital to physical values.

03_NDVI_10Dmax_1km Product Source | Europe

Download data from ftp server after place an free order. Extract archieves. Load extracted files to xarray. Subset data. Apply conversation from digital to physical values. Find irrigation period.

04_LUCAS_TOPSOIL Product Source: Europe

Preprocess LUCAS TOPSOIL 2015 and 2009 Datasets. Merge to Geopandas GeoDataFrame. Subset and Export data.

05_BDFL5 Product Source: Hessen - Product Source | RLP

Preprocess Soil Information dataset. Spatial join with LUCAS TOPSOIL data. Subset and Export data.

06_Evapo_rel_pot Product Source: Germany

Download potentiel and real Evaporation. Transform projection with Gdal. Load extracted files to xarray. Subset data. Apply conversation from digital to physical values

07_Sentinel_1_grd Product Source: World

Use Earth Engine Cloudcomputing. Get Sentinel - 1 GRD VV Mean for geometry while masking high NDVI values. Construct Geopandas GeoDataFrame and add all Information from filename.SWI map expors. Extract data for plot and grid size

Export_ImageCollection_fromGee

Query data with GEE and geemap to get 10m plot time series and 50m grid(10x10km) time series data for different Obversation modes of Sentinel 1. Get NDVI, SWI, VV, Angle. Download as tif.

10_surface_soil_moisture Not Working!!

PYSMM python module to extract point or grid soil moisture information with GEE.

11_Irrigation_Detection_V1

Irrigation detection with data from 07_Sentinel_1_grd notebook

12_query_data

Acces all exported datasets from previous notebooks. Actual: LTS, BDFL5, NDVI10Dmax, RADOLAN, SSM, EVAPO_R, EVAPO_P, ... . Subset data. Preprocess and merge

13_Visualize

ISMN Dashboard for interactive visualization of Sentinel 1&2 and soil moisture data. Load data from Irrigation_Detection_V1 notebook. Split into equal satellite groups. Matplotlib and hvplot plotting.

Ground_Trouth_data

Extract TimeSeries data from ismn stations. Find corresponding Sentinel 1, Sentinel 2 and ERA5 pixels and extract all information to geojson file. Merge data and add in-situ soil moisture values closest to satellite derived data.

CNN & CNN_MLP_Regression_soil_moisture

Use Keras and Tensorflow to train a MLP regression model with Sentinel 1, Sentinel 2 and ERA 5 data. Visualize and Analyse trained model with test data.

Irrigation_map_whr Not working!!

Extract features from pdf file

old_notebook_01

GEE stuff from first project. Visualizations

old_notebook_02

Statistical and machine learning Stuff from first project. Seasonal Mean, Standart Deviation, Histogram, Correlation, Covariance, Spearman, Pearson, Cross correlation, Windowed time lagged cross correlation, rolling window, Dynamic Time Warping, Clustering, precipitation analysis, Seasonal Patterns, show measurements on map, univariate description, bivariate description, Preprocessing, geographical fescription