عنوان مقاله [English]
Introduction: Traditional methods of biophysical crop parameters (including biomass) estimation in the form of finite sampling or final weighing of the harvested products, is time consuming, costly and difficult. In recent years, the use of satellite imagery and remote sensing technology has been considered to estimate these parameters. So far, several vegetation indices have been developed and used to evaluate and estimate the bio-physiological and biochemical parameters of the crops. Because of the ease of using these indicators, this method is one of the most commonly used remote sensing techniques to estimate such parameters. Considering that such studies have not been carried out so far in Kermanshah province, the current study was carried out to estimate the corn biomass in a fertile plain of Kermanshah province (Mahidasht) using Landsat 8 satellite imagery.
Materials and Methods: The dry weight of the crop biomass was measured at the time of the satellite passing from 15 farms at the study area. During the corn growth period, there were 8 satellite images which downloaded from the American Geological Survey web site. In this study, 17 vegetation indices (NDVI- TNDVI- MNDVI- SAVI- OSAVI- VI1-VI2-VI3-PD311-PD312-PD321-RVI-NRVI- MIRV1-NIR*-DVI-IPVI) which in previous studies showed acceptable correlation with crop biomass were used. The correlation coefficient between the measured biomass and the corresponding values of the vegetation indices were used to evaluate the accuracy of the algorithms. For each fieldwork, the index with higher correlation coefficient was determined as the appropriate index for that stage of crop growth, and a regression relation was presented between the amount of corn biomass and the desired index. Finally, estimated values of the biomass based on the regression equations were compared with measured biomass using normalized mean square error (NRMSE).
Results: The measured values of the biomass were low at the beginning of the growth period and gradually increased until the seventh visit (August 26) and then decreased in the last visit (September 11). The average of biomass in 15 farms was measured as 40195 and 36741 kg / ha respectively in seventh and eighth fieldworks. Results of the study showed that the indices of PD311 for the first visit, PD321 for the second visit and the initial stages of growth, NIR* for the third, sixth, seventh and eighths, VI3 for the fourth visit, and the NRVI for the fifth visit, had the highest correlation coefficient with the measured values of biomass. The correlation coefficient of the appropriate index in the 8 fieldworks was 0.42, 0.5, 0.58, 0.71, 0.73, 0.66, 0.57 and 0.47, respectively. In overall, NIR * with the mean correlation coefficient of 0.52 was the most favorable index for the entire growth period. Based on values of NRMSE, it can be concluded that fitted relationships were able to estimate the amount of corn biomass except in the first stage of growth with a moderate to good accuracy. The amount of NRMSE in the last fieldwork, which is related to the final biomass yield, was 11.7%, indicating a good match between observed and predicted data.
Conclusion: The results of this study indicate that corn biomass can be estimated using vegetation indices with acceptable accuracy. The precision of this method was better for intermediate periods of crop growth than the early stages. It is better to use an appropriate vegetation index for each stage of crop growth instead of using an index for the entire crop growth period.