Crop Growth Monitoring
Satellite Data : Sentinel 2 (Multispectral)
Model : Crop growth monitoring model identifies various stages of a crop cycle. The key phenometrics include the Start of Season (SOS), End of Season (EOS), Peak of Season (POS), the peak Normalized Difference Vegetation index (NDVI) value during the season, and the overall duration of each crop cycle on a specific farm. This model helps in effective agricultural management, enabling farmers and agribusinesses to optimize crop production, resource allocation, and yield predictions.
Input :
The model required following three inputs:
| Input Name | Description | Example |
|---|---|---|
| Farm Boundaries | Boundaries Extracted from the Farm Boundary Model on which phenometrics are to be estimated. | .GeoJSON format |
| Start Date | Start Date of the Season | dd-mm-yyyy format |
| End Date | End Date of the Season | dd-mm-yyyy format |
User Tips: Please ensure that the selected start date and end date are roughly in line with the crop growing season in the designated area of interest. If not, then the model will not be able to estimate crop phenometrics.
Outputs :
The model would use the above inputs to provide growth metrics of sown crops in the given date range. As crop growth metrics, the model would provide the crop sown date, harvest date, duration of crop, peak growth date, and peak NDVI of the crop. The model provides these estimated phenometrics for each farm in both GeoJSON and JSON formats.
| Output | Description |
|---|---|
| Farm Statistics Over the Seasons | It provide weekly, monthly, and yearly statistics for SOS, POS, and EOS metrics. This data aids in understanding sowing, growth, and harvesting patterns, enabling informed decision-making. |
| Crop Phenological Metrics for Farms | Temporal NDVI data and phenometrics corresponding to each farm boundary (Type : GeoJSON/ JSON). This is available for download. |
Model output attributes: [downloadable as .geojson (Crop Phenological Metrics)]
| Attribute | Description |
|---|---|
| Farm ID | ID of each selected farm |
| DateTime | Dates time series of selected sentinel-2 images |
| Mean NDVI | NDVI time series generated from selected sentinel-2 images |
| Cycle n_SOS | Sowing date of crop n from the given selected date |
| Cycle n_Peak Date | Peak growth date of crop n from the given selected date |
| Cycle n_EOS | Harvest date of crop n from the given selected date |
| Cycle n_Peak NDVI | Peak NDVI of crop n from the given selected date |
NOTE:
n depends on a number of crops sown in a given date range. For each ‘n’th crop, there will be a separate column added to the attributes in the geojson.
The output can be visualized on the farm boundary map in the platform. While interacting with the boundary map, the phenometrics for each farm can be seen in the form of interactive graphs and tables.
The output, “Crop Phenological Metrics” for the farms can be downloaded as .geojson
Additional Details
Minimum size of Area of Interest required (in Sq KMs.): 5
Maximum size of Area of Interest supported (in Sq KMs.): 5000
Geographies supported: All geographies
Sensors Supported: Sentinel-2
Model Accuracy: Although the Start of Season (SOS) and End of Season (EOS) are not exactly the same as the sowing and harvest dates, we compared the SOS with the sowing date and the EOS with the harvest date. The total duration matched within 15 days for 75% of the cases.
Limitations:
- Model might not be able to provide accurate phenometrics for farms prone to the high frequency of cloud coverage.
- The accuracy of phenometrics is reduced for those crops that are grown in seasons of precipitation like rainfall and snowfall.