Crop Type and Acreage Mapping | ||||||||
1-year Crop Type map in Kenya based on monthly products from Sentinel1&2 (Source: GMV). | ||||||||
Product Category | ||||||||
|
| |||||||
Financial Domains | ||||||||
| ||||||||
User requirements | ||||||||
UN18: Need to monitor crop productivity. UN19: Identifying types of crops being grown is essential. UN28: Need to classify the types of crops being grown in order to assess the sustainability and environmental impact of agricultural investments. UN29: Need to accurately measure the planted area for crops. | ||||||||
Description | ||||||||
Crop type and acreage mapping play a crucial role in monitoring agricultural land use and making estimations of crop production. These maps provide detailed information about the agricultural species present in a specific area, including their extent, and growth stage at a particular point in time. Satellite images capture detailed data about agricultural areas, allowing for the identification and classification of different crops based on their spectral characteristics. By leveraging advanced image processing algorithms and machine learning models, crop types can be accurately determined. | ||||||||
Spatial Coverage Target | ||||||||
Individual farm level | ||||||||
Data Throughput | ||||||||
Rapid tasking Data availability |
|
| ||||||
PRODUCT SPECIFICATIONS | ||||||||
Main processing steps | Before creating crop type and acreage maps, the initial step involves mapping the location of crops. This process utilizes machine learning-based classification models, incorporating inputs from various Earth Observation (EO) data sources such as vegetation and backscatter indices. In addition to EO data, non-EO data like local in-situ data and land use land cover maps are also incorporated. The resulting crop location maps are then combined with vegetation and backscatter indices, Digital Surface Models, existing crop type maps like ESA WorldCereal, and ground truth data. These combined inputs are then fed into machine learning models for the classification of different crop types. | |||||||
Input data sources | Optical: Sentinel-2, VHR based on the availability like Pleiades 1A/1B & NEO, WorldView2&3, and SPOT6/7 Radar: Sentinel-1 Supporting data: In-situ crop type data, ESA's WorldCover layer, ESA WorldCereal, ALOS Global Digital Surface Model | |||||||
Accessibility | Sentinel-1&2: freely and publicly available from ESA. Optical VHR imagery: commercially available on demand from EO service providers. | |||||||
Spatial resolution | Sentinel-2: 10 m Optical VHR: ≤ 1 m Sentinel-1: 20 m | |||||||
Frequency (Temporal resolution) | Sentinel-2: 6 days Optical VHR: Sub-daily to Daily Sentinel-1: 6 days | |||||||
Latency | < 1 Day | |||||||
Geographical scale coverage | Globally | |||||||
Delivery/ output format | Data type: Raster, Vector File format: GeoTIFF, Shapefile | |||||||
Accuracies | Thematic accuracy: 80-90% Spatial accuracy: 1.5-2 pixels of input data | |||||||
Constraints and limitations |
| |||||||
User's level of knowledge and skills to extract information and perform further analysis on the EO products. | Skills: Essential Knowledge: Essential |
P02: Crop type and acreage mapping | |
Maturity score | |
Mean: 2.5 | STD: 0.5 |
Constraints and limitations · Cloud presence · The lack of local in-situ data to train the machine learning models. · Machine learning model uncertainty | |
Relevant user needs UN18: Need to monitor crop productivity. UN19: Identifying types of crops being grown is essential. UN28: Need to classify the types of crops being grown to assess the sustainability and environmental impact of agricultural investments. UN29: Need to accurately measure the planted area for crops. | |
R&D gaps · Limitations in discrimination of crop types with similar spectral signatures. As the crop type maps are often group classifications where crops with similar spectral signatures are grouped together. · Smallholder farming remains an issue because of the small size of farms where intercropping happens very often. (This comment may not apply in the case of large commercial farms). | |
Potential improvements drivers The main limitations are due to the input data rather than the methodology, so the improvements include: · More field data worldwide. · Increased spatial and temporal resolution of the input EO data. · Increasing spectral resolution by using hyperspectral data to better discriminate between crop types. · Crop-type predictions using multiple datasets may allow you to differentiate between those crops that are similar spectrally. | |
Utilisation level review | |
Utilisation score | |
Mean: 2.6 | STD: 1.02 |
No utilisation: Low utilisation Medium utilisation Unawareness of the existence of the best available commercial EO product with better specifications. High utilisation Only this product satisfies the technical and usability requirements. | |
Critical gaps related to relevant user needs | |
Guideline gap UN18: Need to monitor crop productivity |
This page has no comments.