X
2927 Rate this article:
5.0

Monitoring Environmental Change With Remote Sensing

 

Geospatial analytics and artificial intelligence are transforming environmental consulting with timely, accurate, and actionable information.

 


Remote sensing technology has been a boon to environmental consultants. Imagery captured by satellites, aircraft and unmanned aerial vehicles (UAVs) can be used to detect even the most subtle changes to the natural environment. The relative ease with which valuable information can now be extracted from these image data sets allows environmental experts to spend more time determining and mitigating the root causes of identified changes.

The value of remote sensing in environmental preservation and remediation has multiplied exponentially in recent years. A proliferation in commercial imaging platforms makes it possible to find environmental issues anywhere on Earth and to monitor their impacts on habitats and other natural resources – on a yearly, monthly and even daily rate. Software automation supported by Artificial Intelligence (AI) has put complex image processing and analysis capabilities within reach of every environmental scientist.

There are many environmental applications of remotely sensed data, but analysis of change – whether good or bad – in land use and land cover is still among the most valuable. For example, environmental consultants are routinely contracted to monitor large development projects and energy extraction sites. They are charged with ensuring regulatory compliance by spotting anomalies that might indicate activities that are causing stress to surrounding habitats.

The beauty of modern imaging sensors is they can ‘see’ things the naked eye can’t. Infrared data collected along with visible bands can uncover anomalies in vegetation and water before a human on the ground notices anything. And imaging sensors aren’t the only things flying overhead. Lidar elevation scanners are now ubiquitous on aircraft and UAVs. Likewise, radar systems, with the ability to peer through clouds and darkness, capture their own valuable data sets from satellites and aircraft.

In addition to keeping an eye on regulated activities recognized to negatively impact the natural landscape, these remote sensing platforms are cost-effective methods of finding pollution and other forms of environmental damage that are otherwise unknown and unexpected. This information enables environmental scientists to determine the cause of previously undiscovered problems and trace them back to their sources.

 


 

More Remotely Sensed Data Than Ever Before

For the environmental consultant, this may be considered a Golden Age of remote sensing. The commercial availability of remotely sensed data has never been greater than it is today thanks to dramatic technological advances in sensors and the platforms that carry them.

 

SMALLER SATELLITES

The miniaturization of propulsion, stabilization and power subsystems has reduced the size of satellites to the point where entire constellations can be built and launched for less than the cost of a single legacy platform like Landsat and SPOT. These multi-satellite constellations have become the norm for new commercial Earth observation systems. The upshot is that several small satellites can be configured in orbit to capture repeat data over a given point on Earth at a rate of at least once per day. Even the most rapidly evolving environmental events can be monitored in near real-time.

ACCESSIBLE RADAR

Unlike passive optical sensors that can only operate in daylight, synthetic aperture radar (SAR) systems have their own power sources allowing them to capture data day or night, through clouds and in most weather conditions. The radar emissions reflect off the Earth’s surface and return valuable information to the sensor. This data can be processed into images that reveal details of the composition, shape, configuration and other characteristics of surface features. For decades the power sources required to operate these systems limited SAR systems to use in large satellites or aircraft. However, recent breakthroughs have put SAR sensors in small satellites – most notably Finland’s ICEYE – and in smaller aircraft. As a result, more SAR data is becoming available for use in environmental monitoring.

PROFESSIONAL UAVS

The cost of unmanned aerial vehicles, or drones, with the stability and power to carry mapping-grade imaging sensors and laser scanners (Lidar) has transformed the remote sensing industry. UAVs are now performing imaging and elevation mapping missions over small areas of interest at a cost and speed not possible for manned aircraft. They can also fly much closer to surface objects, getting incredibly detailed views of features. In addition, some drones are almost considered disposable, capturing data in conditions too hazardous for aircraft, such as volcanic eruptions and wildfires. Overall, this proliferation of affordable UAVs is making an enormous volume of high-resolution image data available for environmental applications.

ADVANCED SENSORS

Optical imaging sensors have gotten smaller and more sophisticated. Flying on small satellites, these digital cameras are capturing optical images with spatial resolution below a half meter (50 cm) and with discrete spectral capabilities spanning a multitude of visible, near infrared and shortwave infrared portions of the spectrum. This means that vegetation damage or water contamination – whether visible or not – can be detected in small areas. And these sensor enhancements aren’t limited to those on satellites. Airborne sensors have become more advanced and affordable, allowing aircraft to gather larger volumes of data faster and more economically than ever before.

 


 

Simplified Workflows With Advanced Analytics

As a result of the explosion in the availability of sensors and platforms, the challenge facing environmental consultants has shifted from having too little data to dealing with too much data. Remote sensing files themselves have grown bigger and more complex as a greater variety of sensors gather more data at finer granularity. Extracting useful information from Earth observation data sets has become a Big Data challenge.

 

ROBUST SOFTWARE SOLUTION

NV5 saw this data tsunami approaching and developed a solution in the form of its commercially available ENVI® software. Once a standard image processing package, ENVI has been enhanced and expanded to become a robust geospatial data analysis software that allows anyone with minimal training to extract meaningful information from all types of imagery – multispectral, hyperspectral, thermal, Lidar and SAR.

Leveraging decades of building geospatial solutions for the Defense and Intelligence communities, NV5 has developed scientifically-proven image analysis algorithms and integrated them into automated workflows. This enables users to derive answers to the most complicated geospatial queries from enormous data sets with a few clicks of the mouse. In addition, ENVI software gives more experienced users the means to create algorithms and build customized automated analysis processes of their own.

As noted earlier, change detection is among the environmental consultant’s most powerful tools. Two or more images acquired over the same area on different dates are simply loaded into ENVI, and with the push of a few buttons, the software identifies change that has occurred over time. Identified changes can be spatial, such as the erosion of a shorelines, or spectral, such as vegetative stress. Environmental consultants can use ENVI with a large number of images to determine the rate of change over time.

To facilitate certain image analysis applications, NV5 has bundled complex workflows into automated modules. The most popular of these is the Feature Extraction module (ENVI FX), which puts the power of the computer to work identifying features based on their spatial, spectral and textural characteristics. For example, rivers, lakes, fields, trees and coastlines can automatically be mapped from optical imagery, freeing up the environmental consultants to analyze specific objects of interest.

SOPHISTICATED FEATURE EXTRACTION

The feature extraction algorithms have become so sophisticated in ENVI, they can even differentiate objects by their condition. Using multispectral imagery, stressed or diseased plants or trees can be isolated automatically from healthy ones in farm fields or forest groves. If the spatial resolution is sufficient, the software can then tabulate the number of healthy or diseased trees in an area covering thousands of acres – a process that could take countless hours if done manually. These capabilities are automated in the ENVI Crop Science module. It should be noted the ENVI FX module works equally well with Lidar point clouds, extracting trees, buildings and power lines from the 3D data.

 


 

Making It Easier To Extract Information From Radar

SAR data contains rich information that often provides insights on features, land surfaces, and water bodies different from those revealed in optical imagery. “As more and more SAR satellites are launched, this once exclusive monitoring ability has become accessible for everyone,” according to Megan Gallaher, Senior Solutions Engineer at L3Harris Geospatial. “These radar data sets, however, are much more complex than either optical or Lidar and require a different processing tools to exploit. ENVI SARscape Analytics enable anyone to process and get actionable results from this data type.”

ENVI SARscape Analytics is an analysis module that integrates multiple processing functions and algorithms into simplified, and in some cases, automated workflows. As with the other ENVI modules, ENVI SARscape puts powerful capabilities into the hands of environmental consultants to quickly and easily extract meaningful results from SAR data.

“One of the most common uses of NV5's SAR technology by our environmental clients is to acquire and analyze images when its cloudy or at night to map the impact of storms or floods while events are still unfolding,” said Joey Griebel, Account Manager, NV5. “That’s a value that only SAR data can provide.”

Another popular application of SAR data that ENVI SARscape simplifies is the detection of deformation in land surfaces. A SAR technology called interferometry accurately measures elevation points and can pinpoint ground movement of a few millimeters from one data capture to the next. Utility operators are increasingly using SAR to monitor geohazards, such as slope creep, near their infrastructure that could signal a destructive landslide is coming.

“Oil and gas operators now routinely acquire SAR data around their hydrocarbon production activities to ensure subsurface fluid extraction is not causing the ground to subside,” said Griebel. “Subsidence can lead to environmental catastrophes created by broken well bores or damaged pipelines.”

 


 

Customizing Analytics With Deep Learning

The identification of specific objects in optical imagery, Lidar point clouds and SAR data sets has been greatly facilitated by automated recognition and extraction using ENVI FX and other modules. While manual mapping of certain items, such as roads, rivers, fields and trees has been replaced by automated methodologies, there are numerous other features that environmental consultants need to find and map.

The ENVI Deep Learning module is a solution to this challenge. Deep learning is a form of Artificial Intelligence that enables the user to ‘teach’ the software how to recognize nearly any feature or condition in the data from spatial, spectral or other characteristics. Whereas Machine Learning algorithms require a user to ‘show’ the computer hundreds of examples of the feature to be identified, Deep Learning uses iterative algorithms that can learn from just a dozen examples in some cases. “As with our other modules, you don’t have to be an Artificial Intelligence expert to use the ENVI Deep Learning module,” said Griebel. “The module has intuitive tools and workflows that enable users to easily label data and generate models with the click of a button.”

For more seasoned imagery experts, the module gives the option of fusing information layers from different remote sensing data sets to create more robust object classifiers. Griebel explained that environmental consultants have embraced the ENVI Deep Learning module for the significant savings in time it provides them, compared with manual processes. One early adopter has taught the module to find and count certain species of wildlife – as small as geese – in high-resolution satellite imagery to determine if the population is increasing or decreasing over time.

This task is incredibly time consuming using manual methods because the animals may number in the thousands, and they are never in the exact same locations from one image collection to another. Once trained, the algorithms evaluate the imagery covering a given study area in a matter of seconds. More importantly, the same Deep Learning algorithm can be applied repeatedly with new images as they are collected.

The time savings of Deep Learning is an obvious benefit, but Griebel said this Artificial Intelligence technique is having another dramatic impact on environmental consultants and other NV5 clients. The ability to identify and map objects and features has become so refined that users are expanding their applications of remotely sensed data to analyze geographic areas that are much larger than they would have ever considered with manual methods in the past. “They can extract features on a much bigger scale now,” he said.

 

Using ENVI’s proven analytics, additional data layers for deep learning can be created to accentuate features of interest. These data layers aid in the development of more robust and accurate classifiers by helping neural networks more easily learn where they need to focus.

 


 

End-To-End Monitoring And Analysis Services Becoming Popular

If today is the Golden Age of remote sensing thanks to the plethora of data collection platforms and sensors flying overhead, tomorrow will have to be called the Platinum age. The deployment of new satellites, aircraft and UAVs is showing no sign of slowing down. In fact, one published source has estimated the number of small commercial imaging satellites could double in the next two years alone.

While the increase in data variety, volume and revisit frequency is a net positive for all consumers of geospatial information, the negative impacts are starting to be felt as well. Specifically, the growing variety of data types and the sheer volume and size of data sets is stressing out end users.

Environmental consultants don’t have time to become subject matter experts on all the different imagery and data sets available to them. And many environmental organizations can’t keep up with the computing power required to process and archive extremely large data files.

“The majority of our environmental clients buy ENVI and do the work themselves, but there are others that contract our services team, especially for the ongoing monitoring of specific and larger areas of interest,” said Griebel.

In response, NV5 has rolled out geospatial solutions services customized to the needs of specific environmental applications. The client does not have to know which type of imagery or analysis algorithms to use. All the client does is describe what environmental questions need to be answered, and NV5 takes care of the rest – from choosing the right data set and creating a practical analytical workflow to providing the computer horsepower and formatting the answer in an easy-to-understand format. “Whether the environmental issue is a one-off problem or it requires long-term periodic monitoring, NV5 provides accurate and trusted answers for the client,” said Griebel.

 


 

DISCOVER MORE

ANSWERS YOU CAN TRUST

ENVI SOFTWARE

Process and analyze all types of imagery and data.

GEOSPATIAL SOLUTIONS FOR DISASTER MANAGEMENT

SOLUTIONS eBOOK

Enhance preparation and response with remote sensing technologies.

DISASTER MANAGEMENT WEBINAR

WEBINAR

See how to effectively use geospatial data in the disaster management cycle.