Introduction to Remote Sensing

Remote sensing is "the process of detecting and monitoring the physical characteristics of an area by measuring its reflected and emitted radiation at a distance from the targeted area" (USGS 2019).

While remote sensing is commonly used as a synonym for satellite data, the concept of remote sensing can also be applied to aerial photography or lidar from drones. The remote part of remote sensing means that you are gathering data from a distance.

A Brief History of US Satellite Imagery

The first images of the earth from space were captured in 1947 from a camera placed in a sub-orbital German V-2 rocket repurposed by the US after WW II.

Figure
First Space Image, 1947 (NASA 2009)

Although the Russians would beat the US into orbit with Sputnik I on 4 October 1957, the United States would be the first to return crude satellite images from a television camera onboard Explorer VI on 14 August, 1959.

Figure
Explorer VI Satellite Image, 1959 (NASA 1959)

The then-secret Corona defense intelligence satellite project would have beaten Explorer VI into space by a few months, but a string of technical failures delayed the first images until CORONA mission XIV on 18 August 1960 (CIA 2015). The satellites used film cameras to capture high resolution images that were then returned to earth in a re-entry capsule that was captured mid-air by a recovery airplane. As befits a cold-war era project the first high-resolution image from space was of the Russian Mys Shmidta Airfield on 18 August 1960.

Figure
Corona Spy Satellite Image of Mys Shmidta Airfield, 1960 (National Reconnaissance Office 1960)

The first truly functional civilian satellite remote sensing system was the Television Infrared Observation Satellite (TIROS) series, the first of which launched on 1 April 1960. This inaugurated the use of satellites for weather observation and forecasting.

Figure
TIROS Weather Image, 1960 (NASA 1960)

Applications of Remote Sensing

Satellite data and imagery has a wide variety of uses in the natural sciences in addition to its military and commercial value.

As an introduction to the wide variety of (perhaps unexpected) uses for remotely sensed data, skim this list of 100 Earth Shattering Remote Sensing Applications and Uses.

Figure
100 Earth Shattering Remote Sensing Applications and Uses

Spatial Resolution

Raster vs. Vector Data

Satellites almost always capture data as raster data. Raster data represent characteristics of areas on the earth as regular grids of rectangular pixels.

Figure
Pixels in a Remotely-Sensed Image

Raster data is often contrasted in GIS with vector data. Vector data stores locations as discrete geometric objects: points, lines or polygons.

Geographic phenomena can usually be represented using either vectors or rasters, but some types of phenomena are better suited to one or the other.

Figure
Points vs Lines vs Polygons vs Rasters
Spatial resolution represents the amount of area on the surface of the earth covered by each pixel. Spatial resolution is usually measured by the distance in meters between the centers of adjacent pixels.

The level of spatial resolution determines the level of detail that can be distinguished in a remotely sensed image. As you zoom in closer to the earth, the amount of information available diminishes and the images appear more fuzzy.

High resolution data is generally desired for more accurate analysis. However, higher resolution data also requires more storage space and processing power, and is harder and more expensive to capture. Deciding on what spatial resolution is appropriate for your work is often a trade-off between how accurate your analysis needs to be and how large your budget can be.

Spatial resolutions for satellite data commonly vary from one kilometer (MODIS), to 30 meters (Landsat), to five centimeters for military intelligence satellites.

Figure
Dallas at Medium Spatial Resolution (MODIS)
Figure
Dallas at High Spatial Resolution (Landsat)

Swath

Because there are technical and cost limitations to resolution that defines how much detail a satellite sensor can capture at any one time, satellite data capture follows a narrow path or swath along the ground. The width of this swath varies by different satellite systems based on their purpose.

Satellites can scan these swaths in two different ways:

Figure
Types of Swath

Temporal Resolution

The temporal resolution of available satellite data is how frequently data is available for any particular location on the surface of the earth. Some satellite systems return to the same location daily, while others that are designed to observe the entire earth may take days or weeks to return to the same location.

Temporal resolution is determined by the orbit of a satellite, which is the path that the satellite follows as it flies around the earth. The orbit is designed so the the centrifugal force of the circling of the satellite around the planet counterbalances the pull of gravity to keep the satellite aloft.

Orbits can have a number of different characteristics, which involve different types of movement relative to the earth:

While temporal resolution of captured data is determined by the orbit, the temporal resolution of usable data can be affected by cloud cover. If an area is obscured by clouds when the satellite passes over, that data may be unusable and no data can be captured for that area until the satellite passes again when there are no clouds or fewer clouds. Some sources like MODIS compensate for this by combining clear data from multiple passes. While this reduces the temporal resolution, it improves the availability of data.

Spectral Resolution

Remote sensing takes advantage of the emission and reflection of electromagnetic radiation by objects on the surface of the earth to capture what is where on the surface of the earth. The spectral resolution of remotely sensed data is the type of electromagnetic radiation represented in the data.

Electromagnetic Radiation

Objects reflect, absorb, and emit energy in a unique way, and at all times. Electromagnetic radiation originates from the vibration of electrons, atoms, and molecules, and is emitted in waves that are able to transmit energy from one place to another.

Different types of electromagnetic radiation are distinguished by the speed of their vibration.

The higher the temperature of an object, the faster its electrons vibrate and the shorter its peak wavelength of emitted radiation. Conversely, the lower the temperature of an object, the slower its electrons vibrate, and the longer its peak wavelength of emitted radiation.

The fundamental unit of electromagnetic phenomena is the photon, the smallest possible amount of electromagnetic energy of a particular wavelength. Photons are units of energy rather than matter, so they have no mass. The energy of a photon determines the frequency (and wavelength) of light that is associated with it. The greater the energy of the photon, the greater the frequency and vice versa.

Electromagnetic radiation is a part of our lives in many ways. Different frequencies of electromagnetic radiation have different propagation characteristics. These characteristics make different frequencies of electromagnetic radiation useful for different types of remote sensing.

Figure
The Electromagnetic Spectrum (Lawrence Berkeley National Laboratory 1996)

Electromagnetic radiation is different from particle radiation, which results from subatomic particles being thrown off by nuclear reactions, and is associated with radioactive materials like uranium and nuclear power plants. Particle radiation is often associated with electromagnetic radiation, but the primary health concern with any kind of radiation is ionization, which occurs when radiation pushes electrons out of atoms and leaves them as ions with a positive charge. With living cells, this ionization damages the cell DNA and can lead to cell death or mutations and cancer.

Bands

While early satellites captured only panchromatic (grayscale) visible light, contemporary satellites often have sensors that capture different ranges of frequencies or bands of electromagnetic radiation.

Satellites capture ranges of frequencies in bands. The spectral resolution of a satellite sensor is usually specified by the numbers of different bands available and the ranges of frequencies covered by each band. The appropriate spectral resolution depends on the purpose of the satellite.

For space imagery we are usually most interested in the red (430-480 THz), green (540-580 THz), and blue (610-670 THz) bands that the three different types of cone cells in our eye retinas can detect as visible light colors.

The diagram below shows the relationship of Landsat bands 1 - 9 to the visible light spectrum. Bands 10 and 11 captured by the Thermal Infrared Sensor (TIRS) are much lower frequencies and are not shown.

Figure
Landsat Bands

Analysis with Raster Data

While aerial and space-based imagery can have value in and of itself for visual interpretation, there are a variety of technical methods for extracting useful information from raster data.

NDVI

Different bands of electromagnetic radiation are useful for analyzing a different types of phenomena. For example, a combination of red and near-infrared bands called normalized difference vegetation index (NDVI) can be used to determine levels of vegetation in a particular area.

NDVI is especially useful for biologists and biogeographers involved in the study of phenology, which deals with the relations between climate and seasonal biological phenomena. In agriculture, NDVI can be used to identify areas of a farm or field that are not growing well.

NDVI is based on a characteristic that photosynthetic green plants tend to reflect infrared light to avoid overheating and reflect green light (which is why they appear green to our eyes), but absorb red light to power the process of photosynthesis.

This phenomena can be used with the Landsat 9 near infrared band (band 5) and the red band (band 4) to calculate an index that is highest in areas with large amounts of vegetation, and lower in areas of low vegetation.

Figure
Normalized Difference Vegetation Index

The range of the index is negative one to positive one.

When near infrared is high and red is low, that is when living photosynthetic plants are reflecting infrared and absorbing red, making NDVI high and closer to one.

1 - 0
------- = 1
1 + 0

When near infrared is low and red is high, such as with bare ground or water, NDVI is low and closer to negative one.

0 - 1
------- = -1
0 + 1

The normalization specified in this formula provides NDVI values that are useful regardless of variations in the level of lighting or the angle of view. It also makes it possible to distinguish between photosynthetic vegetation and non-photosynthetic green areas (like artificial turf). Accordingly, this gives NDVI an advantage over simply looking at an RGB image for areas that appear green.

Because NDVI creates a set of values from -1 to +1 and has no color of its own, maps of NDVI are commonly visualized with false-color that assigns different colors to ranges of values and makes areas with high and low NDVI easier to distinguish. In the example below of the area around Joliet, IL, colors range from red (low NDVI = urban areas or fallow fields) to green (high NDVI = trees or growing agricultural fields).

Figure
NDVI map of Joliet, IL in August of 2019

Classification

Classification of raster data is the process of grouping pixels of data into specific categories or classes. One common application of classification is land use or land cover classification that determines how specific areas of land are being used (cropland, forest, residential, manufacturing, etc.) Such data can be used by governments for taxation or urban planning, or by researchers to analyze environmental changes that result from human activities (such as urban sprawl or deforestation).

One notable example in the US is the National Land Cover Database, which is maintained by a consortium of federal agencies and based on Landsat satellite imagery and other supplementary datasets.

Figure
Land use around Joliet, IL from the National Land Cover Database

There are three primary techniques for image classification (GISGeography 2021):

Ground Control and Downlinks

As with GPS, satellites used for remote sensing are controlled through a mission operations center (MOC). All contemporary satellite data is returned to earth with radio signals through downlink stations that then relay that data to the MOC for processing, storage and communication. The photo below is of a particularly remote downlink station in the Arctic used by the Landsat system.

Figure
German Remote Sensing Data Center, Neustrelitz (DLR 2016)

Google Maps / Earth

Despite the name, when using satellite view in Google Maps or Google Earth, what you actually seeing is data from a variety of providers (both public and private) and a variety of platforms (satellites and airplanes) that has been compiled into a seamless whole by Google. Single scenes can actually be mosaics of data from multiple sources (Google 2020). Indeed, rather than giving a purely faithful picture of what you would see from space, this is an artistic representation of the Earth that is used to effectively communicate what is where so that users can interpret that data more clearly and act accordingly.

Figure
Google Maps Satellite View

The original provider and date of the data being viewed depends on the area being viewed as well as the zoom level (how close or far away you are from the ground). The providers (s) and date of data capture for a particular location and zoom level and is visible in Google Maps along the bottom of the window.

Figure
Google Maps Satellite View Providers

Threats to Satellite Systems

There are around 1,100 active satellites plus another 2,600 or so that have been decommissioned. This does not include as many as 500,000 pieces of "space junk" the size of a marble or larger that NASA tracks to help safeguard space operations. While most space junk will eventually return to earth, the constant addition of new satellites and launch stages, and the occasional collision of satellites (turning two satellites into multiple pieces of junk) means that the threat to space operations by space junk is increasing and irreversible.

As the list of space-faring nations grows, terrestrial conflicts could extend to actions against spaceborne systems, making military and civilian geospatial technology highly vulnerable to disruption or destruction by state and non-state actors.

All satellite systems are expensive to build, launch, maintain, and renew. As such, they are dependent upon political and economic support that is tenuous in the contemporary American political environment.