Spatial resolution refers to the smallest object that a sensor can detect on the ground, typically expressed as the size of a single pixel in the image. For instance, a spatial resolution of 1 meter means that one pixel represents a 1x1 meter area on the ground. High spatial resolution is crucial for applications requiring detailed imagery, such as urban planning or precision agriculture. Conversely, low spatial resolution is often acceptable for applications like weather monitoring, where broad coverage is more important than fine detail. Advances in satellite technologies have significantly improved spatial resolution, with some modern systems achieving sub-meter resolution. However, higher resolution often comes at the cost of reduced coverage area, low radiometric resolution or increased data volume. Selecting an appropriate spatial resolution depends on the specific requirements of the study or application.
Spectral resolution defines a sensor’s ability to distinguish between different wavelengths of light, typically measured by the number and width of spectral bands. Sensors with high spectral resolution, such as hyperspectral sensors, can detect subtle differences in material properties by capturing hundreds of narrow spectral bands. This capability is essential for applications like vegetation analysis, mineral mapping, and water quality assessment. In contrast, multispectral sensors, with fewer and broader bands, are suitable for general-purpose imaging and cost-effective monitoring. Higher spectral resolution, however, often involves trade-offs, such as increased data complexity and processing requirements. Recent technological advancements have enabled better integration of spectral data with machine learning techniques, enhancing analytical precision. Selecting the right spectral resolution involves balancing the need for detailed spectral information with the constraints of data storage and processing capabilities.
Radiometric resolution describes a sensor’s ability to measure
variations in the intensity of radiation it receives. It is determined
by the number of bits used to represent the data, with higher bit depths
allowing finer discrimination between intensity levels. For example, an
8-bit sensor can distinguish 256 levels, while a 16-bit sensor can
differentiate over 65,000 levels. This resolution is critical for
detecting subtle variations in reflectance, particularly in applications
like thermal imaging or climate studies. Higher radiometric resolution
enhances the sensor’s ability to detect slight changes in surface
properties, enabling more precise analysis. However, it also results in
larger data sizes, requiring efficient storage and processing
strategies. The choice of radiometric resolution depends on the study’s
sensitivity requirements and data handling capabilities. Recent
innovations are pushing the boundaries, enabling more precise
radiometric analyses in remote sensing.
In general, there are trade-offs between choosing a higher spatial
resolution, higher spectral resolution, and higher radiometric
resolution, where increasing one will decrease the other two.
Temporal resolution, also called revisit time, refers to the frequency with which a sensor captures images of the same area, often measured in days. It is a critical factor for monitoring dynamic processes, such as vegetation growth, urban expansion, or disaster response. High temporal resolution, such as daily or sub-daily revisits, allows for near-real-time analysis, making it essential for applications like weather forecasting and crop monitoring. Conversely, low temporal resolution is sufficient for studies requiring long-term trend analysis, such as glacier retreat or deforestation. Advances in satellite constellations, such as those by commercial providers, have drastically improved temporal resolution by enabling multiple revisits per day. However, higher temporal resolution often involves trade-offs in spatial or spectral resolution due to system limitations. Balancing these resolutions is key to designing effective monitoring strategies. Emerging technologies, such as unmanned aerial vehicles (UAV) or CubeSat satellite chains, further complement satellite systems, enhancing temporal resolution for localized studies.
To analyze if an object can be resolved by a sensor, we evaluate the spatial, radiometric, and spectral criteria:
An object is resolvable if its dimension (\(D\)) is larger than the sensor’s Ground Sampling Distance (GSD): \[D \geq \text{GSD} = \frac{H}{f} \cdot p\]
Contrast between the object and background is essential. The object is resolvable if: \[C = \frac{I_{\text{obj}} - I_{\text{background}}}{I_{\text{background}}} \geq \frac{1}{2^n}\] where \(n\) is the sensor’s radiometric resolution in bits.
Two objects are resolvable spectrally if the difference in their wavelengths exceeds the sensor’s bandwidth: \[\Delta \lambda \leq |\lambda_1 - \lambda_2|\]
An object can be resolved when all conditions are satisfied: \[D \geq \text{GSD}, \quad C \geq \frac{1}{2^n}, \quad \Delta \lambda \leq |\lambda_1 - \lambda_2|\] This combined analysis improves our understanding of sensor capabilities and the limits of detectability.
We can roughly categorize current missions as active and passive remote sensing operations. The active ones are radar and lidar, while the passive ones are multispectral/hyperspectral, thermal, and accelerometer-based ones. Active remote sensing can operate both day and night, and in the case of radar, it can also penetrate clouds and provide virtually gap-free dataset on its revisit time. Passive remote sensing collects the light reflected from the surface or region/object of interest (except accelerometer where no light is collected, but the acceleration or deceleration of the twin satellite is the measure) and require cloud-free conditions.
Multispectral imaging has multiple, but not so many, different wavelength windows in their sensors and record the light in these windows in separete detectors. Landsat and Sentinel-2 satellite series (especially Landsat-5 and 8, and Sentinel-2) have been widely used MSI missions providing consistent global data.
Accelerometer missions like GRACE measure gravity field variations, aiding hydrology and geophysics. Comparatively, they are the lowest altitude orbit satellites among earth observation satellites.
This kind of remote sensing technique is the most similar to taking a spectrum of a material in the laboratory, with quite narrow wavelength windows and high number of different bins to distinguish even minute dissimilarities. Even though low number of missions and even lower amount of data are available, they are of very high value in classification tasks. The HySIS mission by ISRO delivers hyperspectral data for mineral exploration.
Another beneficial sensor gives information regarding the temperature of the respective surface. Thermal sensors on Landsat-8 provide land surface temperature data. Applications include urban heat mapping. .
Radar is an active remote sensing operation, in which a microwave-length light is generated and sent to the object/region of interest and the received refleted signal from this carry different types of information from the point of reflection. Synthetic Aperture Radar (SAR) missions like Sentinel-1 are crucial for all-weather observations. . For most tasks, the radar signals sent and received are not influenced remarkably by clouds.
Lidar is similar to radar, but they use 532 or 1064 nm wavelength lasers, in which the 532 can also penetrate water to some extend. NASA’s ICESat-2 provides high-resolution elevation data, important for cryosphere studies. .
