The 2-Minute Summary
Remote sensing in agriculture refers to airborne or space-born measurements of sunlight reflected by crops and soil. Remote sensing has many advantages over traditional monitoring methods. Applications include:
Monitoring landscape change over time
Identifying crops from satellite imagery
Predicting in-season and historical crop yield
Monitoring crop health throughout the growing season
Identifying agricultural management practices
Assessing crop damage remotely
Monitoring soil moisture and soil carbon content
Water usage and quality monitoring
Remote sensing refers to any kind of information collected from a distance (as opposed to taken in contact with the object being measured). In agriculture, remote sensing refers to airborne (manned or unmanned) or space-born measurements of sunlight reflected by the crops and soil across the entire electromagnetic spectrum.
Within the visible range of the electromagnetic spectrum, most plants reflect the same color: green light, making it hard to distinguish different plants or any health factors. Still, there are many useful insights you can draw by comparing these optical measurements over time, including identifying management practices and evaluating some types of crop damage from extreme weather events.
Beyond the visible range (e.g. UV, thermal, microwave wavelengths), vegetation species have different reflectance characteristics, making it possible to gain more insights about crop identification, crop health, and soil condition by using advanced technologies like multispectral and hyperspectral imaging.
For example, leaf cell structure affects reflectance patterns in the near infra-red region. Using correlations established through rigorous agronomic research, these reflectance patterns can be used to predict plant health.
Remote sensing offers a lot of advantages over on-site measurement and monitoring — it’s much cheaper and more scalable. You don’t have to worry about on-the-ground issues that could prevent access like political conflicts or geographical challenges. And you can affordably collect time-series data — which is where the most useful insights come from.
Here’s an overview of 10 distinct agricultural applications of remote sensing.
Landscape Change Monitoring
Optical satellite imagery can provide insights into land use changes over time, to answer questions like: Is this piece of land agriculture? If it’s agriculture, is it currently being cultivated? Was it being cultivated in the last three years?
These insights are useful for things like proving eligibility for prevented planting insurance claims, and proving deforestation-free agricultural practices (which is now a legal requirement for all coffee grown in Europe and the UK).
Within the tropical areas where deforestation is the biggest concern, there are often no reliable records available to show how a piece of land was used in the past. Remote sensing provides objective evidence of land use trends anywhere in the world.
Other applications of landscape change monitoring, according to the US Forest Service, include:
forest planning
updating existing vegetation maps,
assessing landscape conditions, and
supporting post-fire recovery.
Crop Identification
Different crops have different reflectance characteristics, especially in the non-visible wavelengths, which combine to create unique spectral signatures that can be used for crop identification from satellite imagery.
Accurate crop identification depends on access to robust ground-truth data, used for training machine learning models. In many cases, one image alone isn’t enough to identify a crop, but a series of images taken during the growing cycle can provide enough differentiation for accurate identification.
Row crops are easier to identify than tree crops, which require images with much higher resolution.
Using publicly available satellite data, Agrograph can confidently identify major row crops including, but not limited to:
corn
soybean
wheat
cotton
rice
sunflower
canola
sorghum
Private satellite companies offer higher resolution images for a fee that can be used for identification of other types of crops.
Crop Yield Prediction
Crop yield prediction uses machine learning to analyze factors such as plant biomass values derived from satellite imagery, weather data, management practices, and historical yield data to produce in-season yield forecasts.
At Agrograph, this is our main focus. We can provide predictions down to the individual field level. We constantly update these predictions throughout the growing season to capture the impact of that season’s weather, any outbreaks of pests, weeds, or disease, and management practices.
Our most accurate reports are end-of-season yield predictions, which we can use to produce yield reports at harvest that are available six months before the USDA NASS County Yields reports.
We also provide in-season yield prediction reports, which provide valuable directional insights for many agribusiness stakeholders. For example, in July, we might be able to predict that Iowa corn yields will be down by anywhere from 15 to 35 percent. While that’s a wide range, having advance knowledge that yields will be down by any amount is extremely useful for anyone doing large scale grain marketing, such as large farms and co-ops.
Crop Health Monitoring
When crops experience stress, they emit predictable spectral signatures. Using a combination of many types of satellite-derived data, Agrograph can identify when a crop is experiencing stress, whether the cause is drought, pest, or soil conditions, and predict how that stress will affect yield. The separation of those confounding factors is quite difficult, but we don’t need to separate the confounding factors to predict yield.
However, in the realm of precision agriculture, which evaluates crop health and offers prescriptive advice about sub-field level farming practices and optimizing agronomic inputs such as pesticides and fertilizers, differentiating between these different factors is critical.
Precision agriculture services use thermal imaging and multispectral and hyperspectral imaging to identify specific spectral signals highly correlated with issues that can affect crop health, include:
Crop water stress: When crops are experiencing water stress, the temperature in the canopy can increase, which can be detected with thermal imaging from space. Other signs observable from space are things like leaf flipping in soybeans, where leaves turn to expose their silvery-green undersides to reflect sunlight and reduce heat absorption
Weeds, pests, and crop diseases: Traditionally, farmers have relied on manual crop scouting to monitor weed, pest, and disease outbreaks. That’s still the most reliable method, but some of these stressors can be identified from space with hyperspectral imaging before they would be visible with the naked eye, giving an earlier warning than traditional crop scouts would be able to.
Land Management Practice Identification
Satellite imagery is very helpful for identifying agricultural management practices, starting with answering the most basic question: Was this piece of agricultural land cultivated this year?
From there we can examine the data in more detail to answer other questions, such as:
Was cover cropping used?
Was this field irrigated?
Was this field tilled?
Being able to remotely gather these insights for any agricultural field anywhere in the world allows for accurate surveying of adoption of these practices, which can help evaluate their effectiveness or further fine-tune climate change interventions for different geographic regions.
Crop Damage Assessment
Some types of crop damage can be accurately assessed from space. Floods are easy to distinguish with satellite imagery because water has a very different spectral signature from vegetation.
If high winds cause crops like cotton or corn to lean over, this structural damage can be detected with satellite imagery by comparing crop height differences from one week to the next.
Similarly, if a hailstorm causes enough structural damage, that can be detected via remote sensing. However, with some crops, hail damage is harder to assess via satellite imagery. Imagine a soybean field where the hail knocks the pods off of the plant, but the crop itself remains standing. In that scenario, the change in biomass within that field might be less than 5%, which would not be a reliable signal.
Some weather damage takes longer to show up on satellites. In cases where browning occurs because of the damage, it might take a few weeks for that damage to show up in satellite imagery, especially if cloud cover blocks any of the images in the weeks following the storm.
Agrograph’s satellite-based crop damage assessment can be especially helpful for expediting the insurance claims adjustment process. If an insurance company has 1,000 claims following a hail storm, they could use Agrograph data to triage those claims. They could prioritize manual adjustment visits to the areas where damage can’t be detected from space. For the fields where our models predict damage with 95% certainty, they could skip the adjustment process completely.
Soil Condition Monitoring
Remote sensing can be used to evaluate in-season factors affecting soil health, including soil moisture and soil carbon content.
Soil Moisture
As soil moisture increases, reflectance in both the visible and near-infrared wavelengths decreases. This makes it possible to measure soil moisture content using multispectral imaging.
At Agrograph, since our focus is on yield forecasting, we use precipitation levels as a proxy for soil moisture content. However, services like PlanetLabs, which operate satellites equipped with microwave radiation tools, offer daily field-level moisture data.
Soil Carbon
Soil carbon storage or sequestration is a big focus for climate change mitigation. If farmers pursue agricultural practices that increase soil carbon content, they can generate additional revenue by selling carbon credits. But to sell them, you have to reliably be able to measure soil carbon content.
Soil sampling to measure carbon content is expensive — and no specific method is agreed on by the government, agriculture industry, or climate change scientists.
There are some projects and companies (such as carbon credit marketplace Boomitra) that are exploring using remote sensors to track soil carbon content.
Agrograph can provide insights about soil carbon, but like with soil moisture, we use a proxy. To estimate the soil carbon content of an individual field, we look for agricultural practices such as tilling that have research-backed correlations with increased soil carbon content.
Soil Quality Evaluation
Soil quality refers to its relatively static properties — such as structure, texture, density, and water-holding capacity — that affect its productivity. Soil quality is hard to measure from space, but you can measure crop yield over multiple growing seasons, which can function as a good proxy for soil quality.
For example, if you were considering buying a piece of land for farming, looking at satellite-derived historical yield data would tell you a lot about the soil quality across the entire plot of land. That might help corroborate other on-site sampling measurements of soil quality, giving you more confidence in the data.
Water Usage Monitoring
In agricultural settings, water usage is typically measured with propeller meters connected to the water pumping locations, but that’s not the only way it can be measured.
You can use satellite imagery to estimate the amount of water an irrigated field uses per growing season. This can be extremely helpful when metering isn’t reliable or accessible. A few applications of satellite water usage monitoring include:
Water usage comparisons: If you want to benchmark your water usage against neighboring farms growing the same crop on irrigated land, you may not have access to their water usage data. Remote sensing can provide reliable estimates of the water usage of individual fields in any given area, highlighting fields that are more and less efficient with water. If you know which fields are the most efficient, you can investigate why and try to replicate the conditions on other fields.
Aggregate water usage: If you’re a multinational company and you want to evaluate your water usage across a portfolio of multiple subsidiaries, it might be difficult and time-consuming to collect all of that data internally. Estimating water usage across the portfolio via satellite imagery could be a faster and more cost-effective solution.
Water conservation monitoring: Most Regional Water Conservation Districts rely on reported water usage data from farms in the region, but the frequency and timing of that data can vary, impacting its accuracy for comparisons. Satellite-based water usage monitoring could provide a more objective and more timely data source.
Water Quality Monitoring
The reflectance properties of water vary with its depth, suspended sediments, and aquatic vegetation. That makes it possible to detect factors that influence water quality with remote sensing, such as dissolved minerals, suspended material, bacteria levels, oxygen levels, salinity, etc.
NASA provides satellite imagery at 20 to 30 meters spatial resolution that can be used to directly monitor water quality and identify and track the progression of things like harmful algal blooms in lakes.
Since our focus is on crop yield prediction, we don’t directly monitor water quality at Agrograph. However, our focus on crop monitoring gives us the ability to assist with identifying possible contributors to nonpoint source pollution within a particular watershed. Once again, we rely on research-backed correlations between pollution and management practices like tilling and cover cropping. This can help organizations concerned with agricultural runoff prioritize their efforts and measure progress throughout a watershed.
Get Field-Level Actionable Intelligence from Remote Sensing Data
The combination of machine learning, remote sensing data, and ground truth data are unlocking new insights for farmers and agribusinesses. As research continues to uncover more correlations between yields and management practices, spectral signals, weather conditions, and more, the applications of remote sensing in agriculture will continue to grow.
Agrograph’s focus on crop yield prediction helps maximize model accuracy, offering reliable insights both about current season yields and historical yields. We work with clients to customize their data subscription based on the geographies they care about and the actionable insights they need. Contact us to learn more about our data subscriptions.
