Abstract
The monitoring of crop development can benefit from the increased frequency of observation provided by modern geostationary satellites. This paper describes a four-year testing period from 2010 to 2014, during which satellite images from the world's first Geostationary Ocean Color Imager (GOCI) were used for spectral analyses of paddy rice in South Korea. A vegetation index was calculated from GOCI data based on the bidirectional reflectance distribution function (BRDF)-adjusted reflectance, which was then used to visually analyze the seasonal crop dynamics. These vegetation indices were then compared with those calculated using the Moderate-resolution Imaging Spectroradiometer (MODIS)-normalized difference vegetation index (NDVI) based on Nadir BRDF-adjusted reflectance. The results show clear advantages of GOCI, which provided four times better temporal resolution than the combined MODIS sensors, interpreting subtle characteristics of the vegetation development. Particularly in the rainy season, when data acquisition under clear weather conditions was very limited, it was possible to find cloudless pixels within the study sites by compiling GOCI images obtained from eight acquisition periods per day, from which the vegetation index could be calculated. In this study, ground spectral measurements from CROPSCAN were also compared with satellite-based vegetation products, despite their different index magnitude, according to systematic discrepancy, showing a similar crop development pattern to the GOCI products. Consequently, we conclude that the very high temporal resolution of GOCI is very beneficial for monitoring crop development, and has potential for providing improved information on phenology.
Highlights
Phenological changes in land surface vegetation, which are closely related to boundary-layer atmospheric dynamics, have been increasingly seen as important signals of year-to-year climate variations or even global environmental changes [1,2,3,4]
When comparing ground station particulate matter (PM2.5), we found that the overall root mean square error (RMSE) of the aerosol optical depth (AOD) was 0.123 [30]; it follows that the expected error in the surface reflectance using the Moderate-resolution Imaging Spectroradiometer (MODIS) daily AOD will be less than 3% in the 6S
For the intermediate-late-maturing rice paddy in Figure 3a, compared with the NBAR normalized difference vegetation index (NDVI) derived from MODIS, the Geostationary Ocean Color Imager (GOCI) BAR NDVI better reflects the annual tendency with less scattering from general crop seasonal dynamics
Summary
Phenological changes in land surface vegetation, which are closely related to boundary-layer atmospheric dynamics, have been increasingly seen as important signals of year-to-year climate variations or even global environmental changes [1,2,3,4]. To overcome the limitations of polar orbiting reflective wavelength sensors for interpreting vegetation development, various temporal smoothing techniques such as Fourier harmonics, threshold methods, and curve-fitting methods have been suggested to fill or smooth noise and sparse greenness observations from satellite images [17,18,19,20,21,22,23,24]. These techniques are effective for dealing with sporadic missing data, using them for long-term missing data during the cloudy monsoon period of crop growth may produce detrimental results. During the monsoon rainy season (called Jang-Ma in Korea) between
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
