necessary for efficient irrigation and drainage management. For example, the most
beneficial timing for irrigation is during the latter part of the reproductive growth
stages for soybeans versus the earlier tasseling period for corn. In addition, crop
progress information is critical for scheduling fertilization, pest management, and
harvesting operations at optimal times for achieving higher yields (Gao et al. 2017).
Crop progress varies by year and location and is affected by climate variation,
local weather, soil properties, environment changes, and anthropogenic activities. In
the United States, crop progress and condition are estimated using ground survey
data supplied by the trained reporters. These reporters provide visual observations
and subjective estimates of crop progress based on USDA NASS standard definitions. Crop growth stages, crop conditions, and farmers’ activities are reported each
week. The crop progress (CP) reports are summarized and released weekly during
the growing season from early April to late November. The report provides summaries at the agricultural statistic district (multiple counties) and state level to the
public (https://www.nass.usda.gov/Publications/State_Crop_Progress_and_Condi
tion/index.php). These reports do not discuss spatial variability within the agricultural statistical unit.
Remote sensing data are unique in providing spatial and temporal information for
crop monitoring. In recent years, remote sensing time series data have been used to
extract land surface phenology. These approaches use mathematical functions to fit
time series vegetation indices (VIs). Land surface phenology or phenological
parameters are extracted based on either a predefined VI threshold (Jonsson and
Eklundh 2004) or the curvatures of the fitting function (Zhang et al. 2003). Zhang
et al. (2003) developed a phenology program using a hybrid piecewise logistic
function, and the approach has been used to produce the MODIS phenology data
product since 2001. These approaches extract vegetation phenology using particular
features in time series VI data, which can be interpreted as remote sensing phenology. In order to relate phenology detected from remote sensing signals to the fieldobserved crop progress (or physiological stages), Sakamoto et al. (2010) developed a
two-step filtering approach to detect maize and soybean phenology using
MODIS data.
Global land surface phenology products are available at coarse spatial resolution
since 2001 (Zhang et al. 2003). However, the 500-m spatial resolution is still too
coarse for many crop fields. Gao et al. used a data fusion approach that combines the
temporal frequency of MODIS with the spatial resolution of the Landsat (Gao et al.
2006) to build daily time series VI at Landsat 30-m spatial resolution. The fused
Landsat-MODIS data were combined with Landsat observations to extract crop
phenology and then relate to crop growth stages in central Iowa from 2001 to
2016 (Gao et al. 2017). In addition, crop phenology mapping at medium spatial
resolution can be improved by combining Landsat and Sentinel-2 observations.
Land surface phenology at 30 m or finer pixels has been recently retrieved from
Landsat, Sentinel-2, and HLS (Bolton et al. 2020; Gao et al. 2017, 2020a, b; Zhang
et al. 2020).
2 Remote Sensing for Agriculture
17
beneficial timing for irrigation is during the latter part of the reproductive growth
stages for soybeans versus the earlier tasseling period for corn. In addition, crop
progress information is critical for scheduling fertilization, pest management, and
harvesting operations at optimal times for achieving higher yields (Gao et al. 2017).
Crop progress varies by year and location and is affected by climate variation,
local weather, soil properties, environment changes, and anthropogenic activities. In
the United States, crop progress and condition are estimated using ground survey
data supplied by the trained reporters. These reporters provide visual observations
and subjective estimates of crop progress based on USDA NASS standard definitions. Crop growth stages, crop conditions, and farmers’ activities are reported each
week. The crop progress (CP) reports are summarized and released weekly during
the growing season from early April to late November. The report provides summaries at the agricultural statistic district (multiple counties) and state level to the
public (https://www.nass.usda.gov/Publications/State_Crop_Progress_and_Condi
tion/index.php). These reports do not discuss spatial variability within the agricultural statistical unit.
Remote sensing data are unique in providing spatial and temporal information for
crop monitoring. In recent years, remote sensing time series data have been used to
extract land surface phenology. These approaches use mathematical functions to fit
time series vegetation indices (VIs). Land surface phenology or phenological
parameters are extracted based on either a predefined VI threshold (Jonsson and
Eklundh 2004) or the curvatures of the fitting function (Zhang et al. 2003). Zhang
et al. (2003) developed a phenology program using a hybrid piecewise logistic
function, and the approach has been used to produce the MODIS phenology data
product since 2001. These approaches extract vegetation phenology using particular
features in time series VI data, which can be interpreted as remote sensing phenology. In order to relate phenology detected from remote sensing signals to the fieldobserved crop progress (or physiological stages), Sakamoto et al. (2010) developed a
two-step filtering approach to detect maize and soybean phenology using
MODIS data.
Global land surface phenology products are available at coarse spatial resolution
since 2001 (Zhang et al. 2003). However, the 500-m spatial resolution is still too
coarse for many crop fields. Gao et al. used a data fusion approach that combines the
temporal frequency of MODIS with the spatial resolution of the Landsat (Gao et al.
2006) to build daily time series VI at Landsat 30-m spatial resolution. The fused
Landsat-MODIS data were combined with Landsat observations to extract crop
phenology and then relate to crop growth stages in central Iowa from 2001 to
2016 (Gao et al. 2017). In addition, crop phenology mapping at medium spatial
resolution can be improved by combining Landsat and Sentinel-2 observations.
Land surface phenology at 30 m or finer pixels has been recently retrieved from
Landsat, Sentinel-2, and HLS (Bolton et al. 2020; Gao et al. 2017, 2020a, b; Zhang
et al. 2020).
2 Remote Sensing for Agriculture
17