The field of Earth Observation (EO) from satellites has grown steadily and is expected to rise from its present value of $2.9 billion to $25.273 billion by 2040, according to estimates by Morgan Stanley. The use of such imagery, along with ground-based and aerial observations, has been in practice for a long. In parallel, the evolution of technologies like Geographic Information Systems (GIS), Image Processing, and PNT (Position, Navigation, and Timing) has enhanced the quality of imagery analysis to provide value-added digital services to diverse fields of applications. Analytics alone adds $42 billion value to EO data.[Earth Observation: The Growth Story]
The sheer amount of low-earth objects orbiting the earth is very telling.
Many countries including China, India, Russia, and Japan have invested heavily in Earth Observation dominated by big players like the European Space Agency (Copernicus), and NASA (Earth Observing System Data and Information System). While ESA and NASA offer global-coverage low- and medium-resolution imagery and analytics products for open access, there are companies such as Maxar Technologies, and Airbus that have proprietary business models providing access to very-high-resolution. Recently, there is also an influx of interest from innovative private players in this space who are focusing on providing medium- to high-resolution imagery with almost daily revisits. Some of these companies are Planet Labs, Capella Space, BlackSky, Spire, and HawkEye360.
Even though most revenue generated in the sector still comes from defense and intelligence agencies, the interest in the space economy is rising rapidly in applications for the public sector. According to an OECD Report on Space Economy for People, Planet, and Prosperity, space technologies will play a pivotal role in furthering social well-being and sustainable growth in the post-COVID-19 pandemic recovery. In addition, a recent report by the Group on Earth Observations GEO and the Committee on Earth Observation Satellites CEOS reports that EO data can be used to monitor the progress of many of the 169 targets and 230 indicators of the 17 SDGs.
More than 10 years ago the FutureFarm project reported that the agriculture sector is under a strong influence on several different external drivers, including but not limited to climate change, Demographics (growing population, urbanization, and land abandonment), energy cost, new demands on the quality of food (food quality and safety, aging population, and health problems, ethical and cultural changes), innovative drivers (knowledge-based bio-economy, research, and development, information and communication, education, investment), policies (subsidies, standardization, and regulation, national strategies for rural development), economy and financing (economical and financial instruments, partnerships, cooperation, and integration and voluntary agreements), sustainability and environmental issues (valuation of ecological performances, development of sustainable agriculture), public opinion (press, international organization, politicians). Due to the complexity of this problem we need to better understand all processes involved and build for each agriculture sector a new knowledge management system. Earth observation in particular Remote Sensing and GIS datasets, tools, and techniques are becoming an integral part of this knowledge management in the Agriculture Sector.
This blog assesses how satellite imagery datasets and their derived products are being used to make high-quality research outcomes by public sector organizations with a goal to support policy- and decision-making in the Agriculture sector. The analysis is based on a systematic literature review of 159 Scopus indexed journal publications on this topic.
A systematic literature review of Scopus indexed publications is done to understand the scientific outputs from the public sector and multilateral organizations on applying satellite imagery analytics in Agriculture and Food use cases.
A key-word based query is implemented on the Scopus database to retrieve open access journal publications in English from 2020 with the keywords Agriculture or Food in conjunction with words such as Remote Sensing, GIS, Satellite, or Geospatial.
The following query is used -
ABS ( ( “agriculture” OR “food” ) AND ( “remote sensing” OR “GIS” OR “satellite” OR “geospatial” ) ) AND ( LIMIT-TO ( OA , “all” ) ) AND ( LIMIT-TO ( PUBSTAGE , “final” ) ) AND ( LIMIT-TO ( DOCTYPE , “ar” ) OR LIMIT-TO ( DOCTYPE , “cp” ) ) AND ( LIMIT-TO ( PUBYEAR , 2022 ) OR LIMIT-TO ( PUBYEAR , 2021 ) OR LIMIT-TO ( PUBYEAR , 2020 ) ) AND ( LIMIT-TO ( LANGUAGE , “English” ) ) AND ( LIMIT-TO ( SRCTYPE , “j” ) )
This returned 1,761 results including information such as Title, Year of publication, Abstract, Affiliations, etc.
We then checked the contributors to the publication to identify public sector organizations such as FAO, World Bank, CGIAR, to filter these publications for further analysis. This resulted in a set of 159 publications. The abstracts of these projects were manually verified to ensure relevance to the objective of the study. This process led to the selection of 73 relevant publications which were further reviewed in detail (complete paper) to identify recurring use cases, datasets used, and key trends in research.
It is important to note that the insights outlined in this blog are based on a small sample under a short time frame, Jan 2020 until Feb 2022. Also, the shortlisting of relevant publications is based on the affiliations of the authors involved in the research study. This does not in any way mean that the filtered-out samples are not relevant to decision- or policy-making applications in this sector.
The following section summarizes the key trends identified and the use cases where remote sensing data and GIS tools and techniques have resulted in impactful research outcomes and can be used in data-driven and data-informed decision- and policy-making in different country contexts.
(1) Satellite imagery datasets such as those from Landsat or Sentinel are not always directly or independently employed in the different use case implementations. Analytic output from another research outcome is used in most of the publications to calculate new indicators or for driving key decisions. Similarly, satellite imagery datasets are rarely used solely but as one part of a multi-dataset stack for specific use cases.
For example, the Soil erodibility maps in Europe are used in conjunction with datasets such as Digital Terrain Model, Rainfall, NDVI, and Soil to compare the impacts of conventional agricultural management and conservation agriculture on soil erosion and land degradation.
In another use case in India, Sentinel data, gridded weather data, and Mobile-app based field data were analyzed to generate paddy crop map and crop health indicators, namely NDVI, LSWI, Backscatter, and FAPAR for the current (2020) and past years (2016–2019). Using the metrics derived from these indices and entropy techniques, a composite index of crop performance called Crop Health Factor (CHF), ranging from 0–1, was generated.
Land use-land cover map that was constructed by the classification of the Landsat-8 satellite imagery, STMR digital elevation model, soil map acquired from the Food and Agriculture Organization, and climatic data sourced from the NASA-funded prediction of Worldwide Energy Resource were used in this study to create a Soil and Water Assessment Tool model to assess water balance in Iraq.
When openly accessible satellite imagery datasets like Landsat and Sentinel are used directly, Image Enhancement and Preprocessing are critical steps for atmospheric correction, cloud removal, masking, and visibility improvement. Assessment of land-use change in the Thuma forest reserve region of Malawi, Africa explains the preprocessing methodology followed for Landsat imagery, and A New Method for Winter Wheat Mapping Based on Spectral Reconstruction Technology for Sentinel data
(2) Field data collection through surveys and mobile-based tools is critical to continuously improving the quality of analytic products and datasets and to evaluate the accuracy and other evaluation metrics of machine learning or statistical models. This forms a key step in most of the research work where Machine Learning and AI techniques have been implemented for specific use cases with satellite imagery. For example, Mobile-based field data were collected to augment the calculation of indicators such as NDVI, in this paper. These field data points were linked to satellite indices for crop classification and crop condition assessment. Field data points were also used to check the classification performance. This is why household and farm surveys across the developing world, such as those supported by the World Bank Living Standards Measurement Study — Integrated Surveys on Agriculture (LSMS-ISA) initiative, are increasingly important to obtain precise, within-farm measures of crop production and productivity.
(3) Satellite imagery sources coupled with Artificial Intelligence, Spatio-temporal reconstruction, fusion, and ground-truthing data can be used for the creation of Global Analytics-ready Datasets that can be used across use cases. Longer time high-resolution, high-frequency, consistent, and more detailed land cover data is one such dataset that has relevance across use cases and data products. In this study, a 36-year long, 30 m resolution global land cover map data was produced with an average overall accuracy of annual land cover maps over multiple periods is 80% for level 1 classification and over 73% for level 2 classification.
ESRI also in this paper describes the methodology followed for developing and deploying a deep learning segmentation model on Sentinel-2 data to create a global LULC map at 10m resolution that achieves state-of-the-art accuracy and enables automated LULC mapping from time-series observations.
(4) Many publications were reiterating the importance of considering food and agriculture, not as a standalone sector, but being in a nexus with Energy and Water. According to Wageningen University and Research, the water-food-energy nexus is an approach to consider the interactions between water, food, and energy, while taking into account the synergies and trade-offs that arise from the management of these three resources, and potential areas of conflict.
In this study from Uganda, the correlation of agricultural land-use changes with water quality parameters nutrients like Total Nitrogen (TN) and Total Phosphorus (TP), nitrates and turbidity, Total Suspended Solids (TSS), Biological Oxygen Demand (BOD), and temperature is studied along with its impact on aquatic ecosystems.
In another GIS-Based study in Uganda, to understand the demand for irrigation and other energy services, small-scale ground-water irrigation requirements are studied.
(5) Satellite imagery datasets are also used as inputs to develop tools and software that allow viewing and interpretation. These have a wide range of applications including Land monitoring, etc. For example, Collect Earth is a free, user-friendly, and open-source software for land monitoring developed by the Food and Agriculture Organization of the United Nations (FAO).
Another such tool is developed by the use of remote sensing and ecosystem service modeling to examine nature’s contribution to SDG 6 and trade-offs with other key SDGs and with agriculture, spatially. It is available within the Co$tingNature platform.
Access to Information for Agricultural Production and Management
- Crop Productivity and Yield form the foundation for agricultural production and management for smallholder farmers. Spatial distribution of crops, Yield monitoring, … are some of the key parameters to derive important information about crop productivity and food security. There are numerous studies proposing methods to integrate satellite imagery sources for calculating these parameters. For example, this study from China proposes a method based on discrete grids with machine learning to integrate GaoFen-1 and Sentinel-2 imagery for crop classification.
Another study from Indonesia used the Dynamic Time Wrapping method to map the extent and location of vegetable fields from the time series of Sentinel-1A imagery.
- Crop Planning — Predicting crop maturity dates is important for improving crop harvest planning and grain quality. In a study from China, a data assimilation framework incorporating the leaf area index (LAI) product from Moderate Resolution Imaging Spectroradiometer (MODIS) into a World Food Studies (WOFOST) model was proposed to predict the maturity dates of winter wheat in Henan province, China.
Sustainability and Climate Resilience
- Data-Powered Positive Deviance — The Positive Deviance approach is focused on the identification and scaling of strategies undertaken by positive deviants (PDs), which refer to individuals or communities that use uncommon practices that enable them to achieve better outcomes than their peers, despite having similar conditions and resources. Although the approach has had success, the scaling of successes achieved across diverse geographies and large populations has presented numerous challenges, part of which can be attributed to the conventional qualitative and quantitative approaches employed, namely interviews and surveys. Recent developments in technologies and the availability of big data have also presented new possibilities for the PD approach, whereby big data can be harnessed to fill those Spatio-temporal information gaps. In this publication, Identifying Potential Positive Deviants (PDs) Across Rice Producing Areas in Indonesia, an independent time series-based validation approach was developed entirely using EO data to confirm the detection of significant differences in terms of productivity proxies between positive deviants and the rest of the samples.
- Water Management — This work aims to evaluate the sustainability of water management for agriculture in a specific territory through the creation of a synthetic index resulting from the aggregation of multiple indices (environmental, economic, and social). The resulting synthetic index can be used to set sustainability standards and to guide the choices mandated by the Common Agricultural Policy 2023–2027. In this work, a Multiple Criteria Decision Analysis (MCDA) method facilitates a complex process such as establishing a degree of sustainability in a certain area and, therefore, provides support to national or regional policies and communities. The integration of MCDA and GIS increases the efficiency of the support activity. A case study is presented evaluating the level of sustainability in the Irrigation and Reclamation Consortium of Piacenza and Emilia Centrale
- Monitoring Soil Parameters — Prediction of soil moisture is important for determining when to irrigate and how much water to apply, to avoid problems associated with over-and under-watering. Satellite imagery sources such as Sentinel-1 can be used to predict soil moisture content, through the use of architectures such as the convolutional neural network (CNN).
- Early-warning systems — Drought prediction models have been effective in estimating future agricultural drought by using meteorological data. When coupled with satellite-derived indicators such as the Vegetation Condition Index, it has been shown that the accuracy of these models increases significantly. These models can help establish early-warning systems that help reduce disaster risk and prevent environmental and socio-economic damage.
Another study stresses the importance of monitoring other meteorological disasters such as frost which causes a decline in the quality of grapes and reduction in yield using satellite imagery. These early warning systems can not only allow farmers to take proactive measures but also provide objective evaluation damages for insurance claims. In another study from the FAO, a next-generation agricultural stress index is proposed, which is a fused time series of Advanced Very-High-Resolution Radiometer (AVHRR) data from Meteorological Operational satellite (METOP) and National Oceanic and Atmospheric Administration (NOAA) to produce a consistent time series of a vegetation health index (VHI) at 1 km spatial resolution from 1984 to present for Agricultural Drought Monitoring.
- Cropping Intensity Dynamics — Cropping intensity (CI) in cultivated land, helps in understanding the intensified utilization of cultivated land which is critical information for ensuring food security and realizing agricultural sustainability. This study from China demonstrates how to monitor cropping intensity dynamics from 1982 to 2018 using Global Land Surface Satellite Products at large scales.
Access to Finance, and Markets
- Agricultural finance institutions provide the agri-food industry with credit and insurance services, as well as related services such as re-insurance and decision support services to commodities and derivatives traders. They face challenges due to the lack of availability of standardized and up-to-date datasets. For example, insurance contracts such as area-yield insurance (guaranteeing a certain percentage of normal yield over an insured area) require highly reliable, near real-time yield estimates. Satellite-derived indicators play a key role in bridging this data gap in the financing process. For example, this paper introduces a “satellite-derived crop health index” as an alternative to yield data in such an insurance model.
- Research is also able to identify trends and insights from market-oriented practices in urban and peri-urban agricultural (UPA) practices as demonstrated by the paper from Kenya, where spatiotemporal dynamics of UPA and market-oriented farmer’s responses to changing socio-spatial circumstances in two rapidly growing Kenyan cities: Nyeri and Nakuru were analyzed. Another work from Argentina proposes the calculation of a crop proximity index using satellite information to assess and monitor peri-urban agro-industrial activity.
- A greenhouse is an important land-use type, which can effectively improve agricultural production conditions and increase crop yields. The human population is increasing as well as the demand for food and agricultural products, which is expected to grow by 70–100% of the total production of 2010 by 2050 (Godfray et al., 2010). In the context of globalization, many countries, such as Mexico, are reorienting their agriculture towards exports. To achieve these goals, global annual agricultural expansion and improvements in production technology are expected (Tilman, Balzer, Hill, & Befort, 2011). One of the most common technologies is Protected Agriculture (PA) which consists of infrastructure that modifies the natural environment, to protect the crops from external factors and enhance growth (Jensen & Malter, 1995). These include mulching, controlled environment agriculture, hydroponics, greenhouses, tunnels, etc. The publication Rapid Mapping of Large-Scale Greenhouse Based on Integrated Learning Algorithm and Google Earth Engine demonstrates how this infrastructure type can be classified from Landsat images.
In this blog, we systematically reviewed Scopus indexed publication outcomes from development sector organizations to identify key trends and use cases for employing satellite imagery-based datasets in the agricultural sector. We tagged the use cases into 3 main categories and have included sub-categories within them-
- Access to Information for Agricultural Production and Management
- Sustainability and Climate Resilience
- Access to Finance and Markets
The goal of this work was to identify use cases that could be useful for other projects and to demonstrate the immense potential for using imagery datasets in this growing space.
The complete list of 1,761 publications is available here.