resolution (10 m for blue, green, red, and NIR bands) provides more spatial details
than Landsat (30 m) and can be used to study spatial variability at the field to subfield
scales. The Sentinel-2 MSI includes four red edge bands that could be beneficial for
crop monitoring. Sentinel-2 data are freely available to the public and have shown
increasing uses in agricultural applications. Unfortunately, Sentinel-2 satellites do
not have thermal infrared bands that impact the detection of clouds at pixel level. The
lack of thermal infrared bands also limits the study on crop water use that requires
surface temperature in the land surface energy balance model.
The NASA Goddard Space Flight Center has produced the Harmonized Landsat
and Sentinel-2 (HLS) surface reflectance product to increase the temporal resolution.
HLS data products are co-registered, atmospherically corrected and gridded in the
Sentinel-2 tile (Claverie et al. 2018). The data can be used for time series analysis
directly. Version 1.4 HLS data over North America is available from NASA
Goddard Space Flight Center (https://hls.gsfc.nasa.gov/), and version 1.5 over the
globe is available from the NASA EarthData website (https://earthdata.nasa.gov/).
Other commercial satellite data, such as the WorldView and PlanetScope, provide
satellite imagery at very high spatial resolutions. The PlanetScope constellation with
hundreds of small satellites deployed provides a capacity for daily global coverage at
a lower cost. A technical challenge to use these data is the data inconsistency across
satellites and dates. Additional processes are needed to harmonize them for monitoring crop progress and conditions (Houborg and McCabe 2018).
2.3 Agricultural Applications
Remote sensing data have been widely used in agricultural applications, including
crop types mapping, crop growth condition monitoring, crop phenology detecting,
crop yields estimating, crop water use estimating, crop stress condition assessing,
and soil moisture retrieving. This section discusses the major applications using
satellite remote sensing.
2.3.1 Crop Type Identification
Identifying crop type and planting acreage is critical for estimating crop production.
Crop type map is a basis for many agricultural applications. Remote sensing imagery
provides spatial information that can be used to produce the wall-to-wall crop type
map. Crop type classification uses the distinct features of crops in the spectral and
temporal domains to separate different crop types. Remote sensing classification
includes supervised and unsupervised classifications. Most crop type classifications
were performed using a supervised classification method, which requires training
samples. The widely used and effective pixel-based methods include maximum
likelihood, decision tree, neural network, random forests, and support vector
2 Remote Sensing for Agriculture
15
than Landsat (30 m) and can be used to study spatial variability at the field to subfield
scales. The Sentinel-2 MSI includes four red edge bands that could be beneficial for
crop monitoring. Sentinel-2 data are freely available to the public and have shown
increasing uses in agricultural applications. Unfortunately, Sentinel-2 satellites do
not have thermal infrared bands that impact the detection of clouds at pixel level. The
lack of thermal infrared bands also limits the study on crop water use that requires
surface temperature in the land surface energy balance model.
The NASA Goddard Space Flight Center has produced the Harmonized Landsat
and Sentinel-2 (HLS) surface reflectance product to increase the temporal resolution.
HLS data products are co-registered, atmospherically corrected and gridded in the
Sentinel-2 tile (Claverie et al. 2018). The data can be used for time series analysis
directly. Version 1.4 HLS data over North America is available from NASA
Goddard Space Flight Center (https://hls.gsfc.nasa.gov/), and version 1.5 over the
globe is available from the NASA EarthData website (https://earthdata.nasa.gov/).
Other commercial satellite data, such as the WorldView and PlanetScope, provide
satellite imagery at very high spatial resolutions. The PlanetScope constellation with
hundreds of small satellites deployed provides a capacity for daily global coverage at
a lower cost. A technical challenge to use these data is the data inconsistency across
satellites and dates. Additional processes are needed to harmonize them for monitoring crop progress and conditions (Houborg and McCabe 2018).
2.3 Agricultural Applications
Remote sensing data have been widely used in agricultural applications, including
crop types mapping, crop growth condition monitoring, crop phenology detecting,
crop yields estimating, crop water use estimating, crop stress condition assessing,
and soil moisture retrieving. This section discusses the major applications using
satellite remote sensing.
2.3.1 Crop Type Identification
Identifying crop type and planting acreage is critical for estimating crop production.
Crop type map is a basis for many agricultural applications. Remote sensing imagery
provides spatial information that can be used to produce the wall-to-wall crop type
map. Crop type classification uses the distinct features of crops in the spectral and
temporal domains to separate different crop types. Remote sensing classification
includes supervised and unsupervised classifications. Most crop type classifications
were performed using a supervised classification method, which requires training
samples. The widely used and effective pixel-based methods include maximum
likelihood, decision tree, neural network, random forests, and support vector
2 Remote Sensing for Agriculture
15