ICAR Research Data Repository for Knowledge Management
KrishiKosh
View Archive Info| Field | Value | |
| Creator |
NILIMESH MRIDHA
|
|
| Contributor |
R. N. Sahoo
|
|
| Subject |
Assessing Crop Biophysical Parameters, Hyper-Spectral, Multispectral Remote Sensing Data, Radiative Transfer Modeling
|
|
| Description |
T-9116
Reliable crop growth monitoring in near real time over larger area is one of the important steps for precision agriculture and obtaining yield predictions before harvest time. Basic requirement for crop growth monitoring is the efficient tool for retrieving different biophysical parameters and its accuracy or reliability. Accurate quantitative estimation of vegetation biochemical and biophysical variables is also useful for a large variety of agricultural, ecological, and meteorological applications. With the advancement of space technology, advances in research to solve the complex problem of radiation interaction with the plant canopy components for quantification of different biophysical and biochemical parameters has been priority in recent years. However, keeping in view the limitation of empirical approaches like lack of generality of scale of application, subjectivity and scientific explanations, physical process based approach has been given preferred for biophysical parameters taking care non-Lambertian properties of soil and vegetation through bidirectional reflectance studies. Keeping in view above facts, the present study carried out to assess crop biophysical parameters from hyper-spectral and multispectral remote sensing data through radiative transfer modeling with the objectives (i) to evaluate the radiative transfer model and its inversion approach for retrieval of biophysical parameters of wheat and soybean using hyper-BRF data of ground based sensor, (ii) to upscale retrieval of biophysical products of wheat and soybean from satellite data through inversion of radiative transfer model and its evaluation and (iii) to study the spatial variability of crop growth conditions with respect to weather, soil and crop management variables. Field level experiments were conducted on wheat (var. PBW- 502) and soybean (var. Pusa 9814) crops in IARI field to evaluate the radiative transfer model PROSAIL and anisotropy study. Major wheat and soybean growing areas of Punjab and Haryana and south-western parts of Madhya Pradesh respectively were taken for upscaling experimental field scale study to regional for biophysical parameter retrieval and understanding their variability through developing composite crop health index. Two field surveys were conducted in the study area to collected in situ spectral and biophysical data required for the model during the period of Jan 31 to Feb 10, 2011 for wheat and Sep 10 to Sep 20, 2011 for soybean crop with the help of GPS synchronizing with the pass of MODIS satellites. Composite MODIS surface reflectance products (MOD09) of different period of survey were acquired and used.The BRF study at different wavelengths and at different growth stages of wheat and soybean canopy confirmed the anisotropic scattering behavior throughout optical wavelength region. The study reconfirms the strong and consistent anisotropic patterns of wheat and soybean reflectance in VIS and NIR regions in response to change in sun-target-sensor geometry and the magnitude was highest in the principal plane. The bi-directional reflectance was significantly higher showing hotspot in backward scattering direction in principal plane than forward scattering direction at all wavelengths. The hotspot becomes broader in soybean than wheat with the crop growth due to increase in LAI, leaf size and planophilic orientation. The wavelength dependency of BRF effect on crop canopy reflectance as d escribed through anisotropy factor (ANIF) and anisotropy index (ANIX) proved that plant geometry has very definite effects on anisotropy and erectophile wheat canopy showed higher anisotropy (about 1.5 t imes) than planophile soybean canopy and the BRF effects were very pronounced in visible region than NIR region where relatively low BRF effects were observed. The study showed that model simulated spectra closely matched with observed spectra for all the view zenith and azimuth angle combinations used in the experiment and also could simulate well the hotspot position in all the wavelength regions at all growth stages in both the crops. Overall across the observation dates, and wavelengths, comparison with observed spectra revealed that the relative error of the model for spectral simulation was 27 % for wheat and 20% for soybean. Performance of model in respect to simulation was better for soybean crop than wheat. Inversion of the model was done to retrieve biophysical parameters using look up table (LUT), artificial neural network (ANN) and genetic algorithm (GA) approaches. The retrieved biophysical parameters of wheat and soybean in field experiment were in good agreement with measured values for look up table (LUT) and genetic algorithm (GA) approach irrespective of the sensors whereas for artificial neural network, the performance was poor. In general, all approaches could capture the variability in measured biophysical parameters with estimation accuracy in order of LAI > CCC > Cab > Cw and relatively better in soybean than wheat. This study reconfirmed the better performance of model inversion compared to empirical regression for biophysical parameters estimation (except Cw) from satellite data on regional scale with reasonable accuracy. RMSE between estimates and in-situ measurements were lower for LAI (10 – 16 %) than for Cab (12– 21 %) by PROSAIL inversion from satellite data. Block level composite crop health index developed in the study by c ombining all the three retrieved parameters was better correlated with block average yield values of wheat and soybean crops than the individual correlation between yield and LAI or Cab or Cw and the areas (blocks) with higher CHI were found to produce proportionately higher yields and vice versa except some blocks where in spite of having low values of CHI, proportionate yield values are high and vice versa ascribing a combined effect of other factors on c rop yields as shown in this study. The conclusion drawn in this study may help in highlighting critical gaps involved in the retrieval of biophysical parameters by the latest version of PROSAIL model for necessary improvements on regional scale with improved accuracy. Such biophysical products could be used in agriculture for crop monitoring, stress detection, yield prediction and ecosystem sustainability evaluation. For the purpose of crop monitoring, a pre harvest composite crop health index may provide the information about the actual status of crop and prioritize the need of the site specific management practices needed for those particular zones. |
|
| Date |
2017-04-19T07:02:45Z
2017-04-19T07:02:45Z 2014 |
|
| Type |
Thesis
|
|
| Identifier |
http://krishikosh.egranth.ac.in/handle/1/5810009799
|
|
| Language |
en
|
|
| Format |
application/pdf
|
|
| Publisher |
Division of Agricultural Physics, Indian Agricultural Research Institute New Delhi
|
|