Influência da geometria de aquisição sobre índices de vegetação e estimativas de IAF com dados MODIS, HYPERION e simulação PROSAIL para a soja

  • Breuning F
N/ACitations
Citations of this article
9Readers
Mendeley users who have this article in their library.

Abstract

In order to reduce cloud cover effects on the acquisition of optical images, the temporal resolution of the satellites can be improved by using large swath width sensors such as the Moderate Resolution Imaging Spectroradiometer (MODIS). MODIS acquires data over a given agricultural area with distinct view directions and angles in a short period of time. The off-nadir viewing provides important data to generate products using radiative transfer models, but it also introduces variability in the crop spectral response, especially when daily images are analysed. Even the hyperspectral Hyperion sensor, which acquires images over comparatively small areas (7.7 x 50 km) at nadir, can have its revisit time improved through side looking. Thus, crop studies using MODIS must consider directional effects (backscattering and forward scattering) resultant from the illumination-viewing geometry. The objective of this research was to evaluate directional effects, considering also different illumination and view configurations, on the reflectance of soybean canopies observed in distinct phenological stages. For this purpose, data acquired by MODIS and Hyperion, and reflectance simulation from the radiative transfer model PROSAIL, were used. A large soybean area located in Querência municipality, in Mato Grosso state (Brazil), was selected as study area. Two growing seasons were studied: 2004-2005 and 2005-2006. Initially, the PROSAIL was used to evaluate the influence of different soybean canopy parameters (e.g., leaf area index - LAI) and of distinct geometries of data acquisition (e.g., view zenith angle (VZA); solar zenith angle (SZA); relative azimuth angle (RAA) on the reflectance and vegetation indices. The other model parameters were kept constant. Using MODIS daily images (product MOD09), the impact of the directional effects on reflectance spectra of different soybean varieties were evaluated as a function of the crop development. Subsequently, the vegetation index (e.g., Normalized Difference Vegetation Index - NDVI; Enhanced Vegetation Index - EVI; modified Normalized Difference Water Index - NDWI) dependence on the directional effects was evaluated. The Wilcoxon non-parametric statistical test and a directional normalization of these indices, determined for backscattering and forward scattering, were used to identify the less sensible vegetation indices to VZA. Finally, the impact of the directional effects on LAI derived from empirical models, based on the relationship between NDVI and LAI, was performed. Results from PROSAIL simulations showed that the reflectance of the backscattering direction (predominance of sunlit canopy components viewed by the sensor) was higher than the reflectance of the forward scattering direction (predominance of shadow). Despite the LAI increase, the evaluation of the vegetation indices showed that the directional effects were still present. The comparison between the reflectance spectra simulated through PROSAIL and the observed MODIS reflectance spectra showed that PROSAIL underestimated the magnitude of the directional effects. However, as LAI increased, the similarity between both simulated and observed spectra was higher. Based on MODIS daily reflectance images, it was verified that the VZA can generate differences up to 20 % in the near infrared (NIR) reflectance. With the soybean canopy closure, the reflectance was less affected by the MODIS VZA. Results showed that NDVI and Greenness Normalized Difference Vegetation Index (GNDVI) were lesser sensible to directional effects and VZA than EVI and the remaining indices based on the NIR and short wave infrared (NDWI1640 e NDWI2120). The NDVI retrieved from the forward scattering direction was always higher than that derived from the backscattering viewing, but the contrary was observed for EVI. This result is associated with the higher dependence of NDVI on the red reflectance and of EVI on the NIR reflectance. The "Main Algorithm" (radiative transfer modelling) used in the generation of the MODIS LAI Product (MOD15A2) was gradually replaced by the "Backup Algorithm" (empirical modelling) with the soybean development, which is coincident with the peak of regional cloud cover. When the same soybean phenological stage and MODIS images acquired at consecutive days and in opposite view zenith angles were considered in the analysis, LAI estimates from empirical models presented larger values in the forward scattering than in the backscattering viewing. The largest LAI differences were found for NDVI values ranging from 0.70 to 0.85 (critical range). Results showed differences up to 3.2 when the global empirical model was used, which were reduced to values up to 1.5 when the "local" empirical model was utilized. In spite of the high temporal resolution of MODIS, due to factors such as cloud cover and viewing geometry, care is necessary when using their products and daily images for soybean studies.

Cite

CITATION STYLE

APA

Breuning, F. M. (2011). Influência da geometria de aquisição sobre índices de vegetação e estimativas de IAF com dados MODIS, HYPERION e simulação PROSAIL para a soja. Instituto Nacional de Pesquisas Espaciais. Retrieved from http://urlib.net/8JMKD3MGP7W/39D8T88

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free