Estimativa da produtividade e estado nutricional da cultura do algodão irrigado via técnicas de sensoriamento remoto.

Abstract

Remote sensing techniques are a fast and inexpensive alternative to accurate assessments of nutritional status of a culture, and became valuable in cotton (Gossypium hirsutum L.) management to ensure high yield. In this sense, a field experiment was carried out in Apodi county, RN, Brazil, in order to estimate the lint yield and nutritional status of irrigated cotton regarding to nitrogen and phosphorous fertilization, using biophysical variables obtained through remote sensing techniques with data from a portable spectroradiometer and multispectral data from TM Landsat 5, providing ways to the recommendation of nitrogen fertilization. It was used a Randomized Block Design with 3 replications. The treatments included four N rates (0, 90, 180 and 270 kg ha"1) and four P205 rates (0, 120, 240 and 360 kg ha"1), and four assessment seasons, (40, 60, 80, 100 DAE). It was studied the effects of N and P combinations in the plots and the effects of assessment seasons in the subplots. The cotton spectral response was evaluated in the field by means of a portable spectroradiometer operating in the range of 350 to 1100 nm at 60, 80 and 100 DAE. Three multispectral images from Thematic Mapper of Landsat 5 were used, obtained in the days 01/11/2008, 17/11/2008 and 19/12/2008. Additionally, it has been determined the leaf chlorophyll content, N and P leaf content, SPAD indices, leaf area index, plant height, phytomass production of cotton plant and lint yield, as well as it was used meteorological data from the station installed in the experimental field (5°37'19" S and 37°49'06" W). The full flowering stage is the best time to estimate the biophysical variables by means of the vegetation indices obtained by both satellite imagery and data from the portable spectroradiometer. With the acquired data through the spectroradiometer it is possible to obtain mathematical models for all vegetation indices (IVs) used to evaluate the nutritional status and cotton plant vigor. The vegetation index with best performance in response to nitrogen fertilization was MTVI2, reaching its maximum value when applied 250 kg ha"1 of N. It is possible to estimate LAI and cotton phytomass yield using spectral indices obtained by spectroradiometer data with high accuracy. It has been observed high correlation coefficients between the IVs and the biophysical characteristics of cotton with significance at 1% by t test. In order to predict the IAF at the full flowering stage or at 100 DAE, the vegetation index with best fit was the NDVI, confirming its potential as a good estimator of the crop general conditions and canopy density. Phytomass production can be estimated in field by the NDVI, SAVI or the TVI with a good performance at 60 and 80 DAE, since the NDVI can be used as the best predictor in all growth stages, providing high accuracy. Furthermore, the multispectral data from TM Landsat 5 enabled the prediction of LAI through the vegetation indices with high correlation coefficients, showing that the method is suitable for the estimation of biophysical variables since the early flowering stage up to the complete flowering stage. MSAVI and MTVI2 were the best indices for predicting LAI at the early growth stages, while NDVI and TVI showed better performance in the full flowering stage. Phytomass production can be estimated by the model of Monteith (1972), if done during the full flowering stage of cotton or can also be estimated at the early flowering stage through the predictive model obtained with the NDVI, which showed the best performance for this purpose. The vegetation indices obtained from satellite images showed performance similar to those observed by field spectroradiometry in predicting the phytomass production of cotton, where the spectroradiometry has the advantage to be more flexible to schedule the assessment, as well as can be used for each treatment, providing better accuracy for smaller areas. Furthermore, it was observed that the IVs obtained from portable spectroradiometer exhibited better correlation to predict LAI than those indices acquired through satellite images in any of the evaluation phases. The lint yield estimated by the IAF or the phytomass production data acquired from satellite images were highly correlated with these parameters at 80 DAE, showing the capability of the CASA model for that purpose. The results show that both the field spectroradiometry and multispectral satellite data are adequate to obtain the biophysical variables and are useful for evaluating the status and vigor of cotton, making it possible for the correction of crop nitrogen deficiency, with full or partially yield recovering, using supplemental doses bounded by the SPAD index between 40 and 90 days after emergence applied through irrigation.