Green Biomass and Yield estimation | ||||||||
Product Category | ||||||||
|
| |||||||
Financial Domains | ||||||||
| ||||||||
User requirements | ||||||||
UN9: Understanding stock levels and monitoring supply chains. UN18: Need to monitor crop productivity. UN29: Need to accurately measure the planted area for crops. UN38: Need for trustworthy time series of reliable data on assets. | ||||||||
Description | ||||||||
Green biomass is a crucial parameter for various applications and a key input for yield estimation. Multiple methods exist to calculate green biomass using satellite imagery. One approach relies on LSP metrics, which discussed briefly earlier. Green biomass is determined at the peak and end of the season using the Leaf Area Index (LAI) rather than NDVI due to LAI’s correlation with the leaf's life stage. Yield estimation is a complex indicator and can be achieved using machine learning algorithms that are trained with EO data (such as vegetation indices like FAPAR), climate data (temperature and precipitation), crop type and acreage maps, LSP metrics, biomass, and ground truth yield samples. By integrating these data sources, accurate yield predictions can be obtained, aiding in effective agricultural planning and management. | ||||||||
Spatial Coverage Target | ||||||||
Individual farm level | ||||||||
Data Throughput | ||||||||
Rapid tasking Data availability |
|
| ||||||
PRODUCT SPECIFICATIONS | ||||||||
Main processing steps | Green biomass can be computed using optical and/or SAR imagery through various algorithms, which may involve the use of vegetation and soil indicators. Alternatively, commercially available products like Planet Biomass Proxy can be utilized to estimate green biomass. Subsequently, the green biomass data is combined with other EO data, such as vegetation indices like FAPAR, climate data (temperature and precipitation), crop type and acreage maps, LSP metrics, and ground truth yield samples. These combined datasets are then used to train machine learning models for accurate yield estimation. | |||||||
Input data sources | Optical: Sentinel-2, VHR based on the availability like Pleiades 1A/1B & NEO, WorldView2&3, and SPOT6/7 Radar: Sentinel-1 Reanalysis products: ERA5 land Supporting data: crop type and acreage maps, LSP metrics, and ground truth yield samples. | |||||||
Accessibility | Sentinel-1&2: freely and publicly available from ESA. Optical VHR imagery: commercially available on demand from EO service providers. ERA5 land: freely and publicly available from ECMWF | |||||||
Spatial resolution | Sentinel-2: 10 m Optical VHR: ≤ 1 m Sentinel-1: 20 m ERA5 land: 0.1° | |||||||
Frequency (Temporal resolution) | Sentinel-2: 6 days Optical VHR: Sub-daily to Daily Sentinel-1: 6 days ERA5 land: Hourly | |||||||
Latency | < 1 Day | |||||||
Geographical scale coverage | Globally | |||||||
Delivery/ output format | Data type: Raster File format: GeoTIFF. NetCDF | |||||||
Accuracies | Thematic accuracy: 80-90% Spatial accuracy: 1.5-2 pixels of input data | |||||||
Constraints and limitations | n The lack of local in-situ data n Cloud presence n The accuracy of Biomass and Yield estimation relies on the accuracies of their inputs like crop type and acreage maps, LSP metrics, and climate data. n Machine learning model uncertainty | |||||||
User's level of knowledge and skills to extract information and perform further analysis on the EO products. | Skills: Essential Knowledge: Essential | |||||||
Similar Products | Planet Biomass Proxy (link) Spatial resolution: 10 m Frequency (Temporal resolution): Daily Latency: 1 Day Geographical scale coverage: Globally with gaps over some major agricultural areas of the world, due to the discontinuity of Sentinel-1B in December 2021 Delivery / output format: GeoTIFF, NetCDF (Raster), CSV (Time series) Accuracies: 80-90% Accessibility: Commercially available from Planet |
P05: Green biomass and yield estimation | |
Maturity score | |
Mean: 2.2 | STD: 0.58 |
Constraints and limitations · Cloud presence · The lack of local in-situ data to train the models. · Machine learning model uncertainty. | |
Relevant user needs UN9: Understanding stock levels and monitoring supply chains. UN18: Need to monitor crop productivity. UN29: Need to accurately measure the planted area for crops. UN38: Need for trustworthy time series of reliable data on assets. | |
R&D gaps · Limited training data. · Models will often be very specific to particular species and cash crops. Also, region-specific. · The accurate and frequent estimation of stock levels will need further information. | |
Potential improvements drivers · Work may be required to review what models are available and then to Identify which are transferable to different regions and potentially different crops and which need to be improved or built from scratch. · Effort in truthing of predicted values against actual accepted values is likely to need to be undertaken to confidently use the data operationally. | |
Utilisation level review | |
Utilisation score | |
Mean: 2.4 | STD: 0.80 |
No utilisation: Users’ lack of EO knowledge and skills to utilize the EO product. Low utilisation · The product is already satisfying the technical and usability requirements. Medium utilisation The product is already satisfying the technical and usability requirements. High utilisation | |
Critical gaps related to relevant user needs | |
Guideline gap UN18: Need to monitor crop productivity |
This page has no comments.