Skip to content

Posts from the ‘Agriculture’ Category

Why mesonets make weather prediction more accurate

The staggering cost of Montana’s “flash drought”

Some people figured it was climate change. One statistician said it was a part of a cyclical trend for poor crop years. Whatever the cause, the 2017 flash drought that parched the entire state of Montana and most of South Dakota, severely impacted the profitability of ranchers and farmers. In western Montana, fires burned some of the largest acreages in recent history. It resulted in one of the biggest wildfire incident reports (over one-million acres) and caused virtually 100% crop loss in northeastern Montana. The U.S. Dept. of Agriculture estimated the crop loss to be in the hundreds of millions of dollars, and one question was on everybody’s mind—why did no one see it coming?

Figure 1. Montana drought conditions August 2017 (Source: Montana State Library website:

Getting the right weather data

The 2017 Montana Dept. of Natural Resources and Conservation spring drought report indicated plenty of water: “By the end of the month, almost all drought concern was removed from the state, with the exception of Wibaux and Fallon Counties….As of May 9, 2017, Montana was 98.45% drought free.” But in late May, an abrupt shift in weather conditions led to one of the hottest, driest summers on record.

The problem, says Kevin Hyde, Montana State Mesonet Coordinator, lies not only in the need for more weather data but in obtaining the right kind of data. He says, “One of the reasons drought was missed was because we’re still thinking you measure drought by snowpack and how much water is in the river, which is really great if you’ve got water rights. But we’ve got a lot of dryland out there.”

In addition to weather monitoring, Hyde is a big proponent of adding soil moisture and NDVI measurements to each of the Montana Mesonet stations he oversees. He says, “The conventional weather station only measures atmospheric conditions. But ultimately, to make any decisions, we’ve got to know not just how much water comes into the system, but how much goes into the soil. And even that’s not enough…because what we really need to know is how the water situation is going to affect plants.”

Hyde says more data are needed to warn growers and ranchers about upcoming weather risks. He points to the fact that increasing evapotranspiration got missed leading up to the summer of 2017. “We realized that if we were looking carefully at reference ET, we might have seen it about a month earlier. What would people have done? They would have changed their calf purchases. They would have figured out what kind of forage they needed to buy. These are the types of decisions people can make if they know the information sooner.”

Was the drought over? Soil moisture illuminates the bigger picture

Heavy rains came mid-September of 2017, which led some people to believe the drought was over. However, changes in soil moisture told a different story. Very little of the rain made it into the soil. “At the Havre, MT station you can see we had some heavy precipitation events. Then we had early October snows. So people expected good soil water recharge. But at the end of the day, we didn’t get it. On Sept.15th, we had a big soil moisture response at the surface but only a marginal response at 8 inches.” The melt of early October snows onto the soil, still damp from the September rain, drained to 20 inches or more. But as the snowmelt dissipated, there was minimal net gain going into the winter.

Figure 2. Soil moisture traces at the Havre, MT weather station

Predictive models need more coverage to be effective

Typically in the U.S., the National Weather Service (a division of NOAA) puts out a network of weather monitoring stations spaced out across the country, and that data gets fed into forward-looking models that help predict the weather. Dr. Doug Cobos, research scientist at METER says, “What people are finding out is that putting in a sparse network of very expensive systems has done really well. It’s been a good thing. But the spatial gaps in those networks are a problem, especially for agriculture producers and ranchers. They need to know what’s happening where they are.”

Hyde agrees, adding that we need better predictive tools that help growers and ranchers make practical decisions based on data rather than guessing. “January 1st is when the decision has to be made—do I buy cows? Do I sell cows? Do I need more pasture? But many predictions start on April 1st. As one rancher puts it, ‘We don’t bother with Las Vegas. We sit around the dining room table at the beginning of the year and put a million dollars on one shot.’”

Mesonets improve spatial distribution

Mesonets present a practical solution for the need to fill in data gaps between large, complex weather stations. The Montana Mesonet currently has 57 stations interspersed throughout the state, and through partnerships with both the public and private sector, they’re adding more stations every year.

Figure 3. Map of MT Mesonet weather stations (source:

At each location, the Montana Mesonet team installs METER all-in-one weather stations, soil moisture sensors, NDVI sensors and data loggers that integrate with ZENTRA Cloud: an easy-to-use web software that seamlessly integrates into third-party applications through an API. He says the system enables better spatial distribution and reliability. “When we were deciding on equipment we asked ourselves: What kind of technology should we use? It had to provide high data integrity. It had to be easy to deploy and maintain. And it had to be cost effective. There’s not a lot of people in that sector. METER systems are low profile, they’re affordable, and the reliability is there. I look at some other mesonets, and they cannot afford to build out further because they are relying on large, complex, expensive systems. That’s where the METER system comes into play.”

Figure 4. Montana Mesonet station setup (Photo credit: Kevin Hyde)

Betting on the future

The Mesonet team and its partners are excited to see how their data will mesh with the available predictive tools to be the most useful and practical for growers and ranchers throughout the state, and they realize that there is still much work to do. “It’s not enough just to get the instrumentation out there. The overall crux is: how do we build the information network, and how do we build a relationship with the producers so that we can have an iterative and interactive conversation?” says Hyde. “We know there needs to be an education in how to use and interpret the data. For example: what is NDVI, and what can we learn from it? A lot of what we need to do is translate science into practical terms.” But he adds that it doesn’t need to be perfect. “What the farmers have said to us is, ‘We don’t need exact numbers. We’re gamblers. Give us probability. Teach us what it means, and we’ll make the decision.’”

Find more information on the Montana Mesonet here.

Just released: ATMOS 41 comparison testing data

Climate parameters such as precipitation, air temperature, and wind speed can change considerably across short distances in the natural environment. However, most weather observations either sacrifice spatial resolution for scientific accuracy or research-grade accuracy for spatial resolution.

weather station

ATMOS 41 all-in-one weather station

The ATMOS 41 represents an optimization of both. It was carefully engineered to maximize accuracy at a price point that allows for spatially distributed observations. Additionally, because many researchers need to avoid frequent maintenance and long setup times, the ATMOS 41 was designed to reduce complexity and withstand long-term deployment in harsh environments. To eliminate breakage, it contains no moving parts, and it only requires recalibration every two years. Since all 14 measurements are combined in a single unit, it can be deployed quickly and with almost no effort. Its only requirement is to be mounted and leveled on top of a pole with an unobstructed view of the sky.

Comparison testing and sensor-to-sensor variability data

METER released the ATMOS 41 in January 2017 after extensive development and testing with partnerships across the world, in Africa, Europe, and the US. We performed comparison testing with high-quality, research-grade non-METER sensors and conducted time-series testing for sensor-to-sensor variability.

See the results



Why does my soil moisture sensor read negative?

How is a negative number possible?

METER soil moisture sensors measure the dielectric constant of the substrate in which they are installed. They are designed to measure soil, which has a dielectric constant of around 5.

METER TEROS 12 soil moisture sensor

Water has a dielectric of approximately 80, so if we assume that a dry soil has a dielectric of 5 (VWC = 0.00 m3/m3), then changes to the bulk dielectric read by the sensor will be attributable to changes in water content. If you read a METER sensor in air, which has a dielectric constant of 1, you will quite naturally get a negative number.   

Improving accuracy of dielectric soil moisture sensors

There are two common causes for negative readings on a METER soil moisture sensor:  

1) Poor contact with the soil resulting from improper installation or disturbance

Air gaps next to a sensor will contribute the lower dielectric of air to the measurement resulting in an underestimation of VWC. Air gaps can arise if enough care is not taken to pack soil around the sensor body to approximate native bulk density. Sensors that have been disturbed, such as having a cable tripped over, can also develop air gaps that can result in negative results in dry soils. (To reduce the possibility of air gaps when installing METER sensors, use the new TEROS borehole installation tool

2) A calibration that is inappropriate for the soil in which the sensor is installed

If the standard mineral calibration is used, an error of ~ 3-4% can be expected in METER sensor readings. Negative numbers can be observed in oven-dry soils (by definition a VWC of 0.0 m3/m3) down to ~ – 0.02 m3/m3 with no malfunction of the sensor. The dielectric constant of the soil is assumed to be 5 and this is a valid assumption in the majority of soils of primarily mineral composition. If your soil has a different dielectric constant, such as can occur in soils with high organic matter content, then the uncertainty in your measurements will increase. This is not a large problem because METER sensors can be calibrated to match a given soil with very little investment in resources.

Want more details?  

Watch our webinar titled Why Does My Sensor Read Negative below. This webinar is designed for those who use electromagnetic sensors (capacitance/TDR/FDR) to measure soil water content. Learn about the theory behind the measurements. Dr. Doug Cobos discusses:

  • What is volumetric water content?
  • Dielectric measurement theory basics
  • Dielectric mixing models
  • Why might a sensor read a negative VWC?
  • Can a sensor really have 2% VWC accuracy for all soils?
  • Sources of error in dielectric measurement methods
  • Improving accuracy of dielectric measurements



What’s next for Fukushima?

Shortly after the Fukushima disaster, we donated environmental sensors to Dr. Masaru Mizoguchi, a scientist colleague at the University of Tokyo, to help him contrive a more environmentally friendly method to rid rice fields in the villages near Fukushima of the radioactive isotope cesium 137.

Scientists continue to search for ways to prevent the recontamination of the rice paddies.

Since then, his efforts, along with the efforts of a team of scientists and citizens, have made the rice grown in the paddies near the disaster site safe for human consumption. But questions and challenges remain. For instance, what will happen to the contaminated soil surrounding the decontaminated area? Will it settle in nearby stream beds, eventually contaminating the rice paddies? And what kind of erosion will come from the nearby tree-covered and clearcut hillslopes?

Recently, our scientists and videographers visited the villages near Fukushima to film some of the progress being made. Watch the video, and read the full story here.

Data collection: 8 best practices to avoid costly surprises

Every researcher’s goal is to obtain usable field data for the entire duration of a study. A good data set is one a scientist can use to draw conclusions or learn something about the behavior of environmental factors in a particular application. However, as many researchers have painfully discovered, getting good data is not as simple as installing sensors, leaving them in the field, and returning to find an accurate record. Those who don’t plan ahead, check the data often, and troubleshoot regularly often come back to find unpleasant surprises such as unplugged data logger cables, sensor cables damaged by rodents, or worse: that they don’t have enough data to interpret their results. Fortunately, most data collection mishaps are avoidable with quality equipment, some careful forethought, and a small amount of preparation.

Before selecting a site, scientists should clearly define their goals for gathering data.

Make no mistake, it will cost you

Below are some common mistakes people make when designing a study that cost them time and money and may prevent their data from being usable.

  • Site characterization: Not enough is known about the site, its variability, or other influential environmental factors that guide data interpretation
  • Sensor location: Sensors are installed in a location that doesn’t address the goals of the study (i.e., in soils, both the geographic location of the sensors and the location in the soil profile must be applicable to the research question)
  • Sensor installation: Sensors are not installed correctly, causing inaccurate readings
  • Data collection: Sensors and logger are not protected, and data are not checked regularly to maintain a continuous and accurate data record
  • Data dissemination: Data cannot be understood or replicated by other scientists

When designing a study, use the following best practices to simplify data collection and avoid oversights that keep data from being usable and ultimately, publishable.

Read more

5 ways site disturbance impacts your data—and what to do about it

Lies we tell ourselves about site disturbance

When it comes to measuring soil moisture, site disturbance is inevitable. We may placate ourselves with the idea that soil sensors will tell us something about soil water even if a large amount of soil at the site has been disturbed. Or we might think it doesn’t matter if soil properties are changed around the sensor because the needles are inserted into undisturbed soil.

site disturbance

The key to reducing the impact of site disturbance on soil moisture data is to control the scale of the disturbance.

The fact is that site disturbance does matter, and there are ways to reduce its impact on soil moisture data. Below is an exploration of site disturbance and how researchers can adjust their installation techniques to fight uncertainty in their data.

Non-disturbance methods don’t measure up—yet

During a soil moisture sensor installation, it’s important to generate the least amount of soil disturbance possible in order to obtain a representative measurement. Non-disturbance methods do exist, such as satellite, ground-penetrating radar, and COSMOS. However, these methods face challenges that make them impractical as a single approach to water content. Satellite has a large footprint, but generally measures the top 5-10 cm of the soil, and the resolution and measurement frequency is low. Ground-penetrating radar has great resolution, but it’s expensive, and data interpretation is difficult when a lower boundary depth is unknown. COSMOS is a ground-based, non-invasive neutron method which measures continuously and reaches deeper than a satellite over an area up to 800 meters in diameter. But it is cost prohibitive in many applications and sensitive to both vegetation and soil, so researchers have to separate the two signals. These methods aren’t yet ready to displace soil moisture sensors, but they work well when used in tandem with the ground truth data that soil moisture sensors can provide.

Read more

Get more info on applied environmental research in our


Hydrology 301: What a Hydraulic Conductivity Curve Tells You & More

Hydraulic conductivity is the ability of a porous medium (soil for instance) to transmit water in saturated or nearly saturated conditions. It’s dependent on several factors: size distribution, roughness, tortuosity, shape, and degree of interconnection of water-conducting pores. A hydraulic conductivity curve tells you, at a given water potential, the ability of the soil to conduct water.

One factor that affects hydraulic conductivity is how strong the structure is in the soil you’re measuring.

For example, as the soil dries, what is the ability of water to go from the top of a sample [or soil layer in the field] to the bottom. These curves are used in modeling to illustrate or predict what will happen to water moving in a soil system during fluctuating moisture conditions. Researchers can combine hydraulic conductivity data from two laboratory instruments, the KSAT and the HYPROP, to produce a full hydraulic conductivity curve (Figure 1).

Hydraulic conductivity curve

Figure 1. Example of hydraulic conductivity curves for three different soil types. The curves go from field saturation on the right to unsaturated hydraulic conductivity on the left.  They illustrate the difference between a well-structured clayey soil to a poorly structured clayey soil and the importance of structure to hydraulic conductivity especially at, or near, saturation.

In Hydrology 301, Leo Rivera, Research Scientist at METER, discusses hydraulic conductivity and the advantages and disadvantages of methods used to measure it.

Watch the webinar below.



Get more info on applied environmental research in our

Lab versus in situ soil water characteristic curves—a comparison

The HYPROP and WP4C provide the ability to make fast, accurate soil moisture release curves (soil water characteristic curves-SWCCs), but lab measurements have some limitations: sample throughput limits the number of curves that can be produced, and curves generated in a laboratory do not represent their in situ behavior. Lab-produced soil water retention curves can be paired with information from in situ moisture release curves for deeper insight into real-world variability.

soil water characteristic curves

Soil water characteristic curves help determine soil type, soil hydraulic properties, and mechanical performance and stability

Moisture release curves in the field? Yes, it’s possible.

Colocating matric potential sensors and water content sensors in situ add many more moisture release curves to a researcher’s knowledge base. And, since it is primarily the in-place performance of unsaturated soils that is the chief concern to geotechnical engineers and irrigation scientists, adding in situ measurements to lab-produced curves would be ideal.

In this brief 20-minute webinar, Dr. Colin Campbell, METER research scientist, summarizes a recent paper given at the Pan American Conference of Unsaturated Soils. The paper, “Comparing in situ soil water characteristic curves to those generated in the lab” by Campbell et al. (2018), illustrates how well in situ generated SWCCs using the TEROS 21 calibrated matric potential sensor and METER’s GS3 water content sensor compare to those created in the lab.

Watch the webinar below:



IoT Technologies for Irrigation Water Management (Part 2)

Dr. Yossi Osroosh, Precision Ag Engineer in the Department of Biological Systems Engineering at Washington State University, continues (see part 1) to discuss the strengths and limitations of  IoT technologies for irrigation water management.

Informed irrigation decisions require real-time data from networks of soil and weather sensors at desired resolution and a reasonable cost.

LoRaWAN (a vendor-managed solution see part 1) is ideal for monitoring applications where sensors need to send data only a couple of times per day with very high battery life at very low cost. Cellular IoT, on the other hand, works best for agricultural applications where sensors are required to send data more frequently and irrigation valves need to be turned on/off. Low-Power Wide-Area Networking (LPWAN) technologies need gateways or base stations for functioning. The gateway uploads data to a cloud server through traditional cellular networks like 4G. Symphony Link has an architecture very similar to LoRaWAN with higher degree of reliability appropriate for industrial applications. The power budget of LTE Cat-M1 9 (a network operator LPWAN) is 30% higher per bit than technologies like SigFox or LoRaWAN, which means more expensive batteries are required. Some IoT technologies like LoRa and SigFox only support uplink suited for monitoring while cellular IoT allows for both monitoring and control. LTE-M is a better option for agricultural sensor applications where more data usage is expected.

NB-IoT is more popular in EU and China and LTE Cat-M1 in the U.S. and Japan. T-Mobile is planning to deploy NB-IoT network in the U.S. by mid-2018 following a pilot project in Las Vegas. Verizon and AT&T launched LTE Cat-M1 networks last year and their IoT-specific data plans are available for purchase. Verizon and AT&T IoT networks cover a much greater area than LoRa or Sigfox. An IoT device can be connected to AT&T’s network for close to $1.00 per month, and to Verizon’s for as low as $2 per month for 1MB of data. A typical sensor message generally falls into 10-200 bytes range. With the overhead associated with protocols to send the data to the cloud, this may reach to 1KB. This can be used as a general guide to determine how much data to buy from a network operator.

Studies show there is a potential for over 50% water savings using sensor-based irrigation scheduling methods.

What the future holds

Many startup companies are currently focused on the software aspect of IoT, and their products lack the sensor technology. The main problem they have is that developing good sensors is hard. Most of these companies will fail before batteries of their sensors die. Few will survive or prevail in the very competitive IoT market. Larger companies who own sensor technologies are more concerned with the compatibility and interoperability of these IoT technologies and will be hesitant to adopt them until they have a clear picture. It is going to take time to see both IoT and accurate soil/plant sensors in one package in the market.  

With the rapid growth of IoT in other areas, there will be an opportunity to evaluate different IoT technologies before adopting them in agriculture. As a company, you may be forced to choose specific IoT technology. Growers and consultants should not worry about what solution is employed to transfer data from their field to the cloud and to their computer or smart phones, as long as quality data is collected and costs and services are reasonable. Currently, some companies are using traditional cellular networks. It is highly likely that they will finally switch to cellular IoT like LTE Cat-M1. This, however, may potentially increase the costs in some designs due to the higher cost of cellular IoT data plans.

Get more info on applied environmental research in our

IoT Technologies for Irrigation Water Management

Dr. Yossi Osroosh, Precision Ag Engineer in the Department of Biological Systems Engineering at Washington State University, discusses where and why IoT fits into irrigation water management. In addition, he explores possible price, range, power, and infrastructure road blocks.

Wireless sensor networks collect detailed data on plants in areas of the field that behave differently.

Studies show there is a potential for water savings of over 50% with sensor-based irrigation scheduling methods. Informed irrigation decisions require real-time data from networks of soil and weather sensors at desired resolution and a reasonable cost. Wireless sensor networks can collect data on plants in a lot of detail in areas of the field that behave differently. The need for wireless sensors and actuators has led to the development of IoT (Internet of Things) solutions referred to as Low-Power Wide-Area Networking or LPWAN. IoT simply means wireless communication and connecting to some data management system for further analysis. LPWAN technologies are intended to connect low-cost, low-power sensors to cloud-based services. Today, there are a wide range of wireless and IoT connectivity solutions available raising the question of which LPWAN technology best suits the application?

IoT Irrigation Management Scenarios

The following are scenarios for implementing IoT:

  1. buying a sensor that is going to connect to a wireless network that you own (i.e., customer supplied like Wi-Fi, Bluetooth),
  2. buying the infrastructure or at least pieces of it to install onsite (i.e., vendor managed LPWAN such as LoRaWAN, Symphony Link), and
  3. relying on the infrastructure from a network operator LPWAN (e.g., LTE Cat-M1, NB-IOT, Sigfox, Ingenu, LoRWAN).

This is how cellular network operators or cellular IoT works. LPWAN technology fits well into agricultural settings where sensors need to send small data over a wide area while relying on batteries for many years. This distinguishes LPWAN from Bluetooth, ZigBee, or traditional cellular networks with limited range and higher power requirements. However, like any emerging technology, certain limitations still exist with LPWAN.

Individual sensor subscription fees in cellular IoT may add up and make it very expensive where many sensors are needed.

IoT Strengths and Limitations

The average data rate in cellular IoT can be 20 times faster than LoRa or Symphony Link, making it ideal for applications that require higher data rates. LTE Cat-M1 (aka LTE-M), for example, is like a Ferrari in terms of speed compared to other IoT technologies. At the same time, sensor data usage is the most important driver of the cost in using cellular IoT. Individual sensor subscription fee in cellular IoT may add up and make it very expensive where many sensors are needed. This means using existing wireless technologies like traditional cellular or ZigBee to complement LPWAN. One-to-many architecture is a common approach with respect to wireless communication and can help save the most money. Existing wireless technologies like Bluetooth LE, WiFi or ZigBee can be exploited to collect in-field data. In this case, data could be transmitted in-and-out of the field through existing communication infrastructure like a traditional cellular network (e.g., 3G, 4G) or LAN. Alternatively, private or public LPWAN solutions such as LoRaWAN gateways or cellular IoT can be used to push data to the cloud. Combination of Bluetooth, radio or WiFi with cellular IoT means you will have fewer bills to pay. It is anticipated that, with more integrations, the IoT market will mature, and costs will drop further.

Many of LPWAN technologies currently have a very limited network coverage in the U.S. LTE Cat-M1 by far has the largest coverage. Ingenu, which is a legacy technology, Sigfox and NB-IOT have very limited U.S. coverage. Some private companies are currently using subscription-free, crowd-funded LoRaWAN networks to provide service to U.S. growers: however, with a very limited network footprint. Currently, cellular IoT does not perform well in rural areas without strong cellular data coverage.

In two weeks: Dr. Osroosh continues to discuss IoT strengths and limitations in part 2.

Get more info on applied environmental research in our