[Quantitative Identification of Yellow Rust and Powdery Mildew in Winter Wheat Based on Wavelet Feature].

Powdery mildew (Blumeria graminis) and stripe rust (Puccinia striiformis f. sp. Tritici) are two of the most prevalent and serious winter wheat diseases in the field, which caused heavy yield loss of winter wheat all over the world. It is necessary to quantitatively identify different diseases for spraying specific fungicides. This study examined the potential of quantitative distinction of powdery mildew and yellow rust by using hyperspectral data with continuous wavelet transform at canopy level. Spectral normalization was processing prior to other data analysis, given the differences of the groups in cultivars and soil environment. Then, continuous wavelet features were extracted from normalized spectral bands using continuous wavelet transform. Correlation analysis and independent t-test were used conjunctively to obtain sensitive spectral bands and continuous wavelet features of 350～1 300 nm, and then, principal component analysis was done to eliminate the redundancy of the spectral features. After that, Fisher linear discriminant models of powdery mildew, stripe rust and normal sample were built based on the principal components of SBs, WFs, and the combination of SBs & WFs, respectively. Finally, the methods of leave-one-out and 55 samples which have no share in model building were used to validate the models. The accuracies of classification were analyzed, it was indicated that the overall accuracies with 92.7% and 90.4% of the models based on WFs, were superior to those of SFs with 65.5% and 61.5%; However, the classification accuracies of Fisher 80-55 were higher but no different than leave-one-out cross validation model, which was possibly related to randomness of training samples selection. The overall accuracies with 94.6% and 91.1% of the models based on SBs & WFs were the highest; The producer' accuracies of powdery mildew and healthy samples based on SBs & WFs were improved more than 10% than those of WFs in Fisher 80-55. Focusing on the discriminant accuracy of different disease, yellow rust can be discriminated in the model based on both WFs and SBs & WFs with higher accuracy; the user' accuracy and producer' accuracy were all up to 100%. The results show great potential of continuous wavelet features in discriminating different disease stresses, and provide theoretical basis for crop disease identification in wide range using remote sensing image.

[Comparative Research on Estimating the Severity of Yellow Rust in Winter Wheat].

In order to improve the accuracy of wheat yellow rust disease severity using remote sensing and to find the optimum inversion model of wheat diseases, the canopy reflectance and disease index (DI) of winter wheat under different severity stripe rust were acquired. The three models of PLS (Partial Least Square), BP neural network using seven hyperspectral vegetation indices which have significant relationship with the occurrence of disease and vegetation index (PRI) were adopted to build a feasible regression model for detecting the disease severity. The results showed that PLS performed much better. The inversion accuracy of PLS method is best than of the VI (PRI, Photochemical Reflectance Index) and BP neural network models. The coefficients of determination (R2) of three methods to estimate disease severity between predicted and measured values are 0.936, 0.918 and 0.767 respectively. Evaluation was made between the estimated DI and the measured DI, indicating that the model based on PLS is suitable for monitoring wheat disease. In addition, to explore the different contributions of diverse types of vegetation index to the models, the paper attempts to use NDVI, GNDVI and MSR which on behalf of vegetation greenness and NDWI and MSI that represents the moisture content to be input variables of PLS model. The results showed that, for the wheat yellow rust disease, changes in chlorophyll content is more sensitive to the disease severity than the changes in water content of the canopy. However, the accuracy of the two models are both lower than predicted when participating in all seven vegetation indices, namely using several species of vegetation indices tends to be more accurate than that using single category. It indicated that it has great potential for evaluating wheat disease severity by using hyper-spectral remote sensing.

[Estimating Winter Wheat Nitrogen Vertical Distribution Based on Bidirectional Canopy Reflected Spectrum].

The vertical distribution of crop nitrogen is increased with plant height, timely and non-damaging measurement of crop nitrogen vertical distribution is critical for the crop production and quality, improving fertilizer utilization and reducing environmental impact. The objective of this study was to discuss the method of estimating winter wheat nitrogen vertical distribution by exploring bidirectional reflectance distribution function (BRDF) data using partial least square (PLS) algorithm. The canopy reflectance at nadir, +/-50 degrees and +/- 60 degrees; at nadir, +/- 30 degrees and +/- 40 degrees; and at nadir, +/- 20 degrees and +/- 30 degrees were selected to estimate foliage nitrogen density (FND) at upper layer, middle layer and bottom layer, respectively. Three PLS analysis models with FND as the dependent variable and vegetation indices at corresponding angles as the explicative variables were. established. The impact of soil reflectance and the canopy non-photosynthetic materials, was minimized by seven kinds of modifying vegetation indices with the ratio R700/R670. The estimated accuracy is significant raised at upper layer, middle layer and bottom layer in modeling experiment. Independent model verification selected the best three vegetation indices for further research. The research result showed that the modified Green normalized difference vegetation index (GNDVI) shows better performance than other vegetation indices at each layer, which means modified GNDVI could be used in estimating winter wheat nitrogen vertical distribution

[Research on universality of least squares support vector machine method for estimating leaf area index of winter wheat].

Leaf area index (LAI) is one of the most important parameters for evaluating winter wheat growth status and forecasting its yield. Hyperspectral remote sensing is a new technical approach that can be used to acquire the instant information of vegetation LAI at large scale. This study aims to explore the capability of least squares support vector machines (LS-SVM) method to winter wheat LAI estimation with hyperspectral data. After the compression of PHI airborne data with principal component analysis (PCA), the sample set based on the measured LAI data and hyperspectral reflectance data was established. Then the method of LS-SVM was developed respectively to estimate winter wheat LAI under four different conditions, to be specific, different plant type cultivars, different periods, different nitrogenous fertilizer and water conditions. Compared with traditional NDVI model estimation results, each experiment of LS-SVM model yielded higher determination coefficient as well as lower RMSE value, which meant that the LS-SVM method performed better than the NDVI method. In addition, NDVI model was unstable for winter wheat under the condition of different plant type cultivars, different nitrogenous fertilizer and different water, while the LS-SVM model showed good stability. Therefore, LS-SVM has high accuracy for learning and considerable universality for estimation of LAI of winter wheat under different conditions using hyperspectral data.

[Inversion of LAI by considering the hotspot effect for different geometrical wheat].

Aimed to deal with the limitation of canopy geometry to crop LAI inversion accuracy a new LAI inversion method for different geometrical winter wheat was proposed based on hotspot indices with field-measured experimental data. The present paper analyzed bidirectional reflectance characteristics of erective and loose varieties at red (680 nm) and NIR wavelengths (800 nm and 860 nm) and developed modified normalized difference between hotspot and dark-spot (MNDHD) and hotspot and dark-spot ratio index (HDRI) using hotspot and dark-spot index (HDS) and normalized difference between hotspot and dark-spot (NDHD) for reference. Combined indices were proposed in the form of the product between HDS, NDHD, MNDHD, HDRI and three ordinary vegetation indices NDVI, SR and EVI to inverse LAI for erective and loose wheat. The analysis results showed that LAI inversion accuracy of erective wheat Jing411 were 0.9431 and 0.9092 retrieved from the combined indices between NDVI and MNDHD and HDRI at 860 nm which were better than that of HDS and NDHD, the LAI inversion accuracy of loose wheat Zhongyou9507 were 0.9648 and 0.8956 retrieved from the combined indices between SR and HDRI and MNDHD at 800 nm which were also higher than that of HDS and NDHD. It was finally concluded that the combined indices between hotspot-signature indices and ordinary vegetation indices were feasible enough to inverse LAI for different crop geometrical wheat and multiangle remote sensing data was much more advantageous than perpendicular observation data to extract crop structural parameters.

[Comparative study on remote sensing invertion methods for estimating winter wheat leaf area index].

The present study aims to explore capability of different methods for winter wheat leaf area index inversion by integrating remote sensing image and synchronization field experiment. There were four kinds of LAI inversion methods discussed, specifically, support vector machines (SVM), discrete wavelet transform (DWT), continuous wavelet transform (CWT) and principal component analysis (PCA). Winter wheat LAI inversion models were established with the above four methods respectively, then estimation precision for each model was analyzed. Both discrete wavelet transform method and principal component analysis method are based on feature extraction and data dimension reduction, and multivariate regression models of the two methods showed comparable accuracy (R2 of DWT and PCA model was 0. 697 1 and 0. 692 4 respectively; RMSE was 0. 605 8 and 0. 554 1 respectively). While the model based on continuous wavelet transform suffered the lowest accuracy and didn't seem to be qualified to inverse LAL It was indicated that the nonlinear regression model with support vector machines method is the most eligible model for estimating winter wheat LAI in the study area.

[Inversion of leaf area index during different growth stages in winter wheat].

Being orientated to the low prescion of crop leaf area index (LAI) inversion using the same spectral vegetation index during different crop growth stages, the present paper analyzed the precision of LAI inversion by employing NDVI(normalized difference vegetation index). Ten vegetation indices were chosen including six broad-band vegetation indices and four narrow-band vegetation indices responding to vegetation cover to inverse LAI in different growth stages. Several conclusions were drawn according to the analysis. The determinant coefficient (R2) and root mean square error (RMSE) between LAI inversion value and true value were 0.5585 and 0.3209 respectively during the whole growth duraton. The mSR (modified simple ratio index) index was appropriate to inverse of LAI during earlier growth stages (before jointing stage) in winter wheat. The R2 and RMSE between LAI inversion value and true value were 0.7287 and 0.2971 respectively. The SR (simple ratio index) index was suitable enough to inverse of LAI during medium growth stages (from joingting stagess to heading stages). The R2 and RMSE between LAI inversion value and true value were 0.6546 and 0.3061 respectively. The NDVI (normalized difference vegetation index) index was proven to be fine to inverse LAI during later growth stages(from heading stage to ripening stage). The R2 and RMSE between LAI inversion value and true value were 0.6794 and 0.3164 respectively. Therefore it was indicated that the results of LAI inversion was much better inverse of winter wheat LAI choosing different vegetation indices during differen growth stages for winter wheat according to the change of vegetation cover and canopy reflectance than merely with NDVI to inverse LAI in the whole growth stages. It was concluded that the precision of LAI inversion was significantly improved with segmented models based on different vegetation indices.

[Geometric distortion correction for hyperspectral image using a rotating scan reflector].

Offner imaging spectrometer is a kind of pushbroom imaging system. Hyperspectral images acquired by Offner imaging spectrometers require relative motion of sensor and scene that is translation or rotation. Via rotating scan with a reflector at the front of sensor's len, large objects can be entirely captured. But for the changes in object distances, geometric distortion occurs. A formula of space projection from an object point to an image point by one capture was derived. According to the projection relation and slit's motion curve, the object points' coordinates on a reference plan were obtained with rotation angle for a variable. A rotating scan device using a reflector was designed and installed on an Offner imaging spectrometer. Clear images were achieved from the processing of correction algorithm.

[Monitoring the chlorophyll fluorescence parameter Fv/Fm in compact corn based on different hyperspectral vegetation indices].

In order to further assess the feasibility of monitoring the chlorophyll fluorescence parameter Fv/Fm in compact corn by hyperspectral remote sensing data, in the present study, hyperspectral vegetation indices from in-situ remote sensing measurements were utilized to monitor the chlorophyll fluorescence parameter Fv/Fm measured in the compact corn experiment. The relationships were analyzed between hyperspectral vegetation indices and Fv/Fm, and the monitoring models were established for Fv/Fm in the whole growth stages of compact corn. The results indicated that Fv/Fm was significantly correlated to the hyperspectral vegetation indices. Among them, structure-sensitive pigment index (SIPI) was the most sensitive remote sensing variable for monitoring Fv/Fm with correlation coefficient (r) of 0.88. The monitoring model of Fv/Fm was established on the base of SIPI, and the determination coefficients (r2) and the root mean square errors (RMSE) were 0.8126 and 0.082 respectively. The overall results suggest that hyperspectral vegetation indices can be potential indicators to monitor Fv/Fm during growth stages of compact corn.

[Study on relationship between new characteristic parameters of spectral curve and chlorophyll content for rice].

It is of significance to monitor chlorophyll content with hyperspectral data for crop growth diagnosis in field. In the study, with the point of view that spectral curve shapes display "tall, low, fat and thin" morphological changes, we proposed some new characteristic parameters from spectral curve such as the ascensive or degressive velocities of segments composing peak or valley shapes in spectral curve, and angles formed by the lines fitting the segments of two sides of peak or valley curves, and used the normalized spectra to analyze correlation between these parameters and rice chlorophyll content. The result shows that (1) there is a good negative correlation between rice chlorophyll content and normalized reflectance spectra from 520-740 nm; (2) characteristic parameters from green peak region of spectral curve display better correlation with rice chlorophyll content, which makes it possible to utilize the parameters to monitor crop chlorophyll content, and will provide new ideas and methods for carrying out crop growth diagnosis with hyperspectral data.

[Advance of the imaging spectral technique in diagnosis of the information of crop].

Labor intensive, time consuming, high technical requirements in operation and much affected by human factors is the limitation of diagnosing the crop information with conventional method, which could not make diagnosis real-time and rapid. Imaging spectral technique could simultaneously obtain the image and spectral information of crops. It could diagnose the growth and insects information of crop rapidly and non-destructively. In recent years, imaging spectroscopy has been widely used in diagnosis of the information of crop, so it provides technical support for agricultural informatization. In the present study, the principle of imaging spectroscopy was presented. The application progress of imaging spectroscopy technique in crop detection was investigated, including seed component detection, seed variety discrimination, seed disease and insect pest detection, field crop growth monitoring and field crop disease and insects monitoring. Then the paper analyzed difficulty of imaging spectroscopy for crop measurement, and the prospect of this technique was also discussed.

[A field-based pushbroom imaging spectrometer for estimating chlorophyll content of maize].

As an image-spectrum merging technology, the field-hperspectral imaging technology is a need for dynamic monitoring and real-time management of crop growth information acquiring at field scale in modern digital agriculture, and it is also an effective approach to promoting the development of quantitative remote sensing on agriculture. In the present study, the hyperspectral images of maize in potted trial and in field were acquired by a self-development push broom imaging spectrometer (PIS). The reflectance spectra of maize leaves in different layers were accurately extracted and then used to calculate the spectral vegetation indices, such as TCARI, OSAVI, CARI and NDVI. The spectral vegetation indices were used to construct the prediction model for measuring chlorophyll content. The results showed that the prediction model constructed by spectral index of MCARI/OSAVI had high accuracy. The coefficient of determination for the validation samples was R2 = 0.887, and RMSE was 1.8. The study indicated that PIS had extensive application potentiality on detecting spectral information of crop components in the micro-scale.

[Investigation of the hyperspectral image characteristics of wheat leaves under different stress].

The diagnosis of growing status and vigor of crops under various stresses is an important step in precision agriculture. Hyperspectral imaging technology has the advantage of providing both spectral and spatial information simultaneously, and has become a research hot spot. In the present study, auto-development of the pushbroom imaging spectrometer (PIS) was utilized to collect hyperspectral images of wheat leaves which suffer from shortage of nutrient, pest and disease stress. The hyperspectral cube was processed by the method of pixel average step by step to highlight the spectral characteristics, which facilitate the analysis based on the differences of leaves reflectance. The results showed that the hyperspectra of leaves from different layers can display nutrient differences, and recognize intuitively different stress extent by imaging figures. With the 2 nanometer spectral resolution and millimeter level spatial resolution of PIS, the number of disease spot can be qualitatively calculated when crop is infected with diseases, and, the area of plant disease could also be quantitatively analyzed; when crop suffered from pest and insect, the spectral information of leaves with single aphid and aphids can be detected by PIS, which provides a new means to quantitatively detect the aphid destroying of wheat leaf. The present study demonstrated that hyperspecral imaging has a great potential in quantitative and qualitative analysis of crop growth.

[Using extraction of red edge position to validate consistency of hyperspectral imaging and non-imaging data].

Using Pushbroom imaging spectrometer (PIS) and FieldSpec ProFR2500 (ASD), spectral reflectances of winter wheat and maize at different stages were collected synchronously. In order to validate the reliability of imaging spectral data, the red edge position of hyperspectral data for PIS and ASD were extracted by different algorithms, respectively. The following results were obtained: (1) The original spectrum of both instruments had high inosculation in red light region (670-740 nm); (2) With the spectra collected under laboratory condition (maize leaf), the extracted red edge position was is concentrated between 700 and 720 nm for the two instruments; (3) With the spectra collected undre field condition (wheat leaf), the extracted red edge position for PIS and ASD were different, the red edge position of PIS data was in 760 nm, while it was in 720 nm for ASD data. The main reason might be that the imaging spectral data were influenced by oxygen absorbtion; (4) the red edge rangeability of PIS and ASD were different, but the trends were the same. The above results could provide some references for hyperspectral imaging data's extensive application.

[Using hyperspectral remote sensing to estimate canopy chlorophyll density of wheat under yellow rust stress].

The canopy reflectance of winter wheat infected with different severity yellow rust was collected in the fields and canopy chlorophyll density (CCD) of the whole wheat was measured in the laboratory. The correlation was analyzed between hyperspectral indices and CCDs, the indices with relationship coefficients more than 0. 7 were selected to build the inversion models, and comparing the predicted results and measured results to test the models, the results showed the first derivative index (D750-D550)/(D750+D550) has higher prediction precision than other indices, while the next is first derivative index (D725-D702)/(D725+D702). Saturation analysis was performed for the above indices, the result indicated that when CCD was more than 12 microg x cm(-2), the first derivative index (D750-D550)/(D750+D550) was easiest to get to saturation level. Therefore, when CCD was less than 12 microg x cm(-2), the first derivative index (D750-D550)/(D750+D550) could be used to estimate wheat CCD and had higher prediction precision than other indices; and when CCD was more than 12 microg x cm(-2), the first derivative index (D725-D702)/(D725+D702) was not easiest to reach saturation level and could be used to estimate wheat CCD. There is a significant minus cor relation between CCD and disease index (DI), moreover, accurate estimation of CCD by using hyperspectral remote sensing not only can monitor wheat growth, but also can provide assistant information for identification of wheat disease. Therefore, this study has important meaning for prevention and reduction of disaster in agricultural field.

[Using canopy hyperspectral ratio index to retrieve relative water content of wheat under yellow rust stress].

The aim of this paper is to estimate canopy relative water contents (RWC) of winter wheat under yellow rust stress by using hyperspectral remote sensing. The canopy reflectance of winter wheat that infected different severity yellow rust was collected and the disease index (DI) of the wheat was investigated respectively in the fields, whereafter the wheat was sampled corresponding to the canopy reflectance measurements and the RWC of the whole wheat were measured in the Laboratory. The research showed that the canopy spectra reflectance gradually decreased in the near-infrared (NIR) region (900-1,300 nm) with RWC reduction, however, canopy spectra reflectance gradually increased in the short-wave-infrared (SWIR) region (1,300-2,500 nm), and there was just higher minus correlation between RWC and DI. Smoothing the canopy spectra, the ratio indices were built by using the sensitive bands for water in NIR and SWIR, and then the estimation RWC linear models were built by using ratio indices as variables, and the model inversion precision and stability were analyzed and compared for estimation RWC. The result indicated that the inversion precision and the stability of the model with ratio index R1,300/R1,200 as variable excel other models, the linear model's RMSE is 3.43, and the relative error is 4.78%. So, this study results not only can provide assistant information for diagnosing wheat disease but also can supply theories and methods for inversion vegetation RWC by using hyperspectral images in the future.

[Using the distance between hyperspectral red edge position and yellow edge position to identify wheat yellow rust disease].

The objective of the present paper is to identify healthy wheat and disease wheat by using hyeprspectral remote sensing as soon as possible. The canopy spectral reflectance of winter wheat infected by different severity yellow rust was measured and the disease indices (DI) were investigated in the field respectively. Smoothing the canopy spectra and calculating the first derivative values, the two methods were used to calculate the red edge position (REP) and yellow edge position (YEP) of the first derivative values: (a) maximum of the first derivative value; (b) Cho and Skidmore method. The result showed that REP gradually shifted to short-wave band, and the YEP gradually shifted to long-wave band with disease severity increasing, however, REP-YEP quickly became smaller. Analyzing and comparing the prediction precision of REP, YEP and REP-YEP for DI, the result indicated that the model REP-YEP as variable has the best estimation precision for DI than REP and YEP, the model estimation error is 6.22, and relative error is 14.3%, and it could identify healthy and disease wheat 12 days before the disease symptom apparently appeared. Therefore, this study not only can provide theory and technology for large areas monitoring of wheat disease by using hyperspectral remote sensing in the future, but also has the important meaning and practical application value for implementing precision agriculture.

[The analysis of consistency between HJ-1B and Landsat 5 TM for retrieving LST based on the single-channel algorithm].

To ascertain whether the thermal infrared image of HJ-1B which has the similar sensor parameter and setting to Landsat 5 TM6 image is applicable for retrieving the land surface temperature (LST), a comparison of retrieved LST between two types of sensors was conducted. Two scenes of thermal infrared images that came from different sensors were acquired in 5th, Apr 2009, which covered the same region in Beijing. To retrieve LST, a generalized single-channel algorithm developed by Jiménez-Muñoz and Sobrino was applied. The LST of study area for both images was thus generated. Based on the LST mapping results and corresponding statistics, an apparent trend could be observed which indicated the consistency in both LST value and its spatial distribution. Consequently, the performance of HJ-IB IRS serving as the data source for LST retrieval was assessed and illustrated in this study. Besides, a high temporal resolution as well as wide swath of the HJ-IRS data suggested its potential in application.

[Winter wheat growth spatial variation study based on temporal airborne high-spectrum images].

Precision agriculture technology is defined as an information-and technology-based agriculture management system to identify, analyze and manage crop spatial and temporal variation within fields for optimum profitability, sustainability and protection of the environment. In the present study, push-broom hyperspectral image sensor (PHI) image was used to investigate the spatial variance of winter wheat growth. The variable-rate fertilization contrast experiment was carried out on the National Experimental Station for Precision Agriculture of China during 2001-2002. Three airborne PHI images were acquired during the wheat growth season of 2002. Then contrast analysis about the wheat growth spatial variation was applied to the variable-rate fertilization area and uniformity fertilization area. The results showed that the spectral reflectance standard deviation increased significantly in red edge and short infrared wave band for all images. The wheat milky stage spectral reflectance has the maximum standard deviation in short infrared wave band, then the wheat jointing stage and wheat filling stage. Then six spectrum parameters that sensitive to wheat growth variation were defined and analyzed. The results indicate that parameters spatial variation coefficient for variable-rate experiment area was higher than that of contrast area in jointing stage. However, it decreased after the variable-rate fertilization application. The parameters spatial variation coefficient for variable-rate area was lower than that of contrast area in filling and milking stages. In addition, the yield spatial variation coefficient for variable-rate area was lower than that of contrast area. However, the yield mean value for variable-rate area was lower than that of contrast area. The study showed that the crop growth spatial variance information can be acquired through airborne remote sensing images timely and exactly. Remote sensing technology has provided powerful analytical tools for precision agriculture variable-rate management.

[Monitoring of winter wheat stripe rust based on the spectral knowledge base for TM images].

In most cases, the reversion model for monitoring the severity degree of stripe rust based on the hyperspectral information can not be directly applied by the satellite images with relatively broad bandwidth, while the airborne hyperspectral images can not be applied for large-scale monitoring either, due to the scale limitation of its data and high cost. For resolving this dilemma, we developed a monitoring method based on PHI images, which relies on the construction of spectral knowledge base of winter wheat stripe rust. Three PHI images corresponding to the winter wheat experimental field that included different severity degree of stripe rust were used as a medium to establish the spectral knowledge base of relationships between disease index (DI) and the simulated reflectance of TM bands by using the empirical reversion model of DI(%) and the relative spectral response (RSR) function of TM-5 sensor. Based on this, we can monitor and identify the winter wheat stripe rust by matching the spectral information of an untested pixel to the spectral knowledge base via Mahalanobis distance or spectral angle mapping (SAM). The precision of monitoring was validated by simulated TM pixels, while the effectiveness of identification was tested by pixels from TM images. The results showed that the method can provide high precision for monitoring and reasonable accuracy for identification in some certain growth stages of winter wheat. Based on the simulated TM pixels, the model performed best in the pustulation period, yielded a coefficient of determination R2 = 0.93, while the precision of estimates dropped in the milk stage, and performed worst in the jointing stage, which is basically inappropriate for monitoring. Moreover, by using the pixels from TM images, the infected pixels could be identified accurately in pustulation and milk stages, while failed to be identified in jointing stage. For matching algorithms, the Mahalanobis distance method produced a slightly better result than SAM method.