Overview / Usage

We usually notice problems like poor crop yield forecasting, ineffective investment analysis, the requirement of established agriculture and food security and inappropriate field maintenance due to unaware ecological parameters in the agricultural domain. Our project helps to overcome those issues by implementing techniques like Vegetation Monitoring, predicting the date of sowing and harvest for intellectual farming, analysing beneficial crop rotations and observing harvesting periods for profitable agricultural business.

We are aiming at developing a high-performance algorithm to analyze multi-temporal data at each pixel to extracting parameters such as date of sowing, date of harvesting and number of harvests based on temporal profile. The basic roadmap of the project is as follows,

1. Start
2. Original Satellite Imagery
3. Conversion into NDVI files
4. Matrix Generation
5. Pixel Wise classification
6. Crop parameters extraction
7. Behavioural Analysis
8. Crop Cycle Predictions
9. Stop

Methodology / Approach

The first step begins with analysing the bands in the provided imagery,

The AWiFS operates in three spectral bands in VNIR and one band in SWIR with 56 meters spatial resolution and a combined swath of 730 km achieved through two AWiFS cameras.

• The catalogue contains the orthorectified multi-spectral AWiFS data.
• The ortho-rectification process is carried out for correcting Terrain relief errors, Scale variation, Sensor attitude/ orientation and Internal errors.

The Normalized Difference Vegetation Index (NDVI) is a standardized index allowing you to generate an image displaying greenness (relative biomass).

• This index takes advantage of the contrast of the characteristics of two bands from a multispectral raster dataset—the chlorophyll pigment absorptions in the red band and the high reflectivity of plant materials in the near-infrared (NIR) band.
• NDVI algorithm calculates the ratio between these two reflectances. This index outputs values between -1.0 and 1.0, representing greenness.
• Negative values (-1 to -0.2) are mainly generated from clouds, water, and snow, and Very low values (-0.1 to 0.1) of NDVI correspond to barren areas of rock, sand and Moderate values (0.2 to 0.4) represent shrub and grassland, while high values (0.5 to 1) indicate temperate and tropical rainforests.
• The NDVI process creates a single-band dataset that mainly represents greenery.

QGIS software, on the other hand, is used to make an analysis of the spatial data and represent the data in a comprehensive format that can be consumed by the end-user.

After the images are converted into the NDVI files, the steps to be executed with the programming are Matrix generation and machine learning technique are used for pixel-wise classification of various NDVI Values. Finally, the algorithm works towards extracting crop cycle parameters and observing the behaviour analysis of the crop. Thus the date of sowing, date of harvesting and number of harvests are predicted.

Technologies Used

• Softwares like QGIS or ArcGIS are used that allows users to analyze and edit spatial information, in addition to composing and exporting graphical maps.
• For the algorithm, libraries in python are used for NDVI data visualisation and prediction. Jupyter notebook is used for implementing the algorithm where the analysis and data preprocessing is done.
• Raster Vision framework is used for built-in support for chip classification, object detection, and semantic segmentation using PyTorch and Tensorflow.
• Intel Devcloud is used for reference in implementations and pre-trained models to help explore real-world workloads on satellite imagery.

Documents and Presentations

Repository

https://vedas.sac.gov.in/vcms/static/SIH-2020/Clipped_NDVI.zip