Pudong Liu

Evaluation of CALIPSO aerosol optical depth using AERONET and MODIS data over China

Aerosol optical depth (AOD) data from Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) were inter-compared and validated against ground-based measurements from Aerosol Robotic Network (AERONET) as well as Moderate Resolution Imaging Spectroradiometer (MODIS) over China during June 2006 to December 2012. We have compared the AOD between CALIOP and AERONET site by site using quality control flags to screen the AOD data. In general, CALIOP AOD is lower than AERONET due to cloud effect detected algorithm and retrieval uncertanty. Better agreement is apparent for these sites: XiangHe, Beijing, Xinglong, and SACOL. Low correlations were observed between CALIPSO and ground-based sunphotometer data in in south or east China. Comparison results show that the overall spatio-temporal distribution of CALIPSO AOD and MODIS AOD are basically consistent. As for the spatial distribution, both of the data show several high-value regions and low-value regions in China. CALIPSO is systematically lower than MODIS over China, especially over high AOD value regions for all seasons. As for the temporal variation, both data show a significant seasonal variation: AOD is largest in spring, then less in summer, and smallest in winter and autumn. Statistical frequency analysis show that CALIPSO AOD and MODIS AOD was separated at the cut-off points around 0.2 and 0.8, the frequency distribution curves were almost the same with AOD between 0.2 and 0.8, while AOD was smaller than 0.4, CALIPSO AOD gathered at the low-value region (0-0.2) and the frequency of MODIS AOD was higher than CALIPSO AOD with AOD greater than 0.8. CALIOP AOD values show good correlation with MODIS AOD for all time scales, particularly for yearly AOD with higher correlation coefficient of 0.691. Seasonal scatterplot comparisons suggest the highest correlation coefficient of 0.749 in autumn, followed by winter of 0.665, summer of 0.566, and spring of 0.442. Evaluation of CALIOP AOD retrievals provides prospect application for CALIPSO data.

Estimating chlorophyll content of spartina alterniflora at leaf level using hyper-spectral data

Spartina alterniflora, one of most successful invasive species in the world, was firstly introduced to China in 1979 to accelerate sedimentation and land formation via so-called “ecological engineering”, and it is now widely distributed in coastal saltmarshes in China. A key question is how to retrieve chlorophyll content to reflect growth status, which has important implication of potential invasiveness. In this work, an estimation model of chlorophyll content of S. alterniflora was developed based on hyper-spectral data in the Dongtan Wetland, Yangtze Estuary, China. The spectral reflectance of S. alterniflora leaves and their corresponding chlorophyll contents were measured, and then the correlation analysis and regression (i.e., linear, logarithmic, quadratic, power and exponential regression) method were established. The spectral reflectance was transformed and the feature parameters (i.e., “san bian”, “lv feng” and “hong gu”) were extracted to retrieve the chlorophyll content of S. alterniflora . The results showed that these parameters had a large correlation coefficient with chlorophyll content. On the basis of the correlation coefficient, mathematical models were established, and the models of power and exponential based on SDb had the least RMSE and larger R2 , which had a good performance regarding the inversion of chlorophyll content of S. alterniflora.

Erratum to: Estimating leaf chlorophyll contents by combining multiple spectral indices with an artificial neural network

Estimating leaf chlorophyll contents through leaf reflectance spectra is efficient and nondestructive, but the actual dataset always based on a single or a few kinds of specific species, has a limitation and instability for a common use. To address this problem, a combination of multiple spectral indices and a model simulated dataset are proposed in this paper. Six spectral indices are selected, including Blue Green Index (BGI), Photochemical Reflectance Index (PRI_5), Triangle Vegetation Index (TVI), Chlorophyll Absorption Ratio Index (CARI), Carotenoid Reflectance Index (CRI) and the green peak reflectance (R525). Both stepwise linear regression (SLR) and back-propagation artificial neural network (ANN) are used to combine the six spectral indices for the estimation of chlorophyll content (Cab). In addition, to overcome the limitation of actual dataset, a “big data” is applied by a within-leaf radiation transfer model (PROSPECT) to generate a large number of simulated samples with varying biochemical and biophysical parameters. 30% of the simulated dataset (SIM30) and an experimental dataset are used for validation. Compared with linear regression method, NN yields better result with R2 = 0.96 and RMSE = 5.80ug.cm−2 for Cab if validated by SIM30, while R2 = 0.95 and RMSE = 6.39ug.cm−2 for SLR. NN also gives satisfactory result with R2 = 0.80 and RMSE = 5.93ug.cm−2 for Cab if validated by LOPEX dataset, however, the SLR only gets 0.72 of R2 and 12.20ug.cm−2 of RMSE. The results indicate that integrating multiple spectral indices can improve the Cab estimating accuracy with a better stability in different kind of species and the model simulated dataset can make up the shortfall of actual measured dataset.

Identification of Phragmites australis and Spartina alterniflora in the Yangtze Estuary between Bayes and BP neural network using hyper-spectral data

The aim of this work was to identify the coastal wetland plants between Bayes and BP neural network using hyperspectral data in order to optimize the classification method. For this purpose, we chose two dominant plants (invasive S. alterniflora and native P. australis) in the Yangtze Estuary, the leaf spectral reflectance of P. australis and S. alterniflora were measured by ASD field spectral machine. We tested the Bayes method and BP neural network for the identification of these two species. Results showed that three different bands (i.e., 555 nm，711 nm and 920 nm) could be identified as the sensitive bands for the input parameters for the two methods. Bayes method and BP neural network prediction model both performed well (Bayes prediction for 88.57% accuracy, BP neural network model prediction for about 80% accuracy), but Bayes theorem method could give higher accuracy and stability.

Estimation of Chlorophyll content of Phragmites australis based on PROSPECT and DART models in the saltmarsh of Yangtze Estuary

Phragmites australis is a native dominant specie in the Yangtze Estuary, which plays a key role in the structure and function of wetland ecosystem. One key question is how to estimate the Chlorophyll content quickly and effectively at large scales, which could be used to reflect the growth condition and calculate the vegetation productivity. The aim of this work was to estimate Chlorophyll content of P. australis based on the PROSPECT and DART (Discrete Anisotropic Radiative Transfer) model. A total of 6 widely used Vegetation indices (VIs) were chosen (i.e., Normalized Difference Vegetation Index (NDVI), Structure Insensitive Pigment Index (SIPI), Colouration Index (COI), Simple Ratio Index (SR), Cater Index (CAI), and Red-edge Position Linear Interpolation (REP_Li)) and calculated, and then the relationship between VIs and Cab were analyzed. Results showed that COI and SIPI were sensitive to the leaf chlorophyll content as the chlorophyll content changes at the leaf scale. Meanwhile, no obvious saturation phenomenon was observed for these two indices compared to other indices.

Simulation and analysis of NDVI performance based on vegetation canopy radiative transfer model

This paper uses PROSAIL model to simulate vegetation canopy reflectance under different chlorophyll contents and Leaf area index (LAI). The changes of NDVIs with different LAIs and chlorophyll contents are analyzed. A simulated spectral dataset was built firstly by using PROSIAL vegetation radiative transfer model with various vegetation chlorophyll concentrations and leaf area index. The responses of NDVIs to LAIs are quantitatively analyzed further based on the dataset. The results show that chlorophyll contents affect canopy reflectance mainly in visible band. Canopy reflectance decreases with an increasing chlorophyll content. Under the same LAI value, NDVI values increase with an increase chlorophyll contents. Under constant content of chlorophyll, NDVIs increases with an increasing LAI. When the value of LAI is less than5, the canopy reflectance is significantly affected by soil background. When value of LAI is higher than5, the earth surface is almost completely covered with vegetation. The increase in LAI has little effect on canopy reflectance and NDVIs consequently. NDVIs increases with the adding of chlorophyll content, when chlorophyll is higher than 40, the rangeability of NDVIs is becoming stable.

The impacts of bandwidths on the estimation of leaf chlorophyll concentration using normalized difference vegetation indices

The aim of this work is to estimate leaf chlorophyll concentration with 6 different normalized difference vegetation indices (NDVIs) under 4 bandwidths (1, 5, 10 and 20 nm). A popular leaf radiative transfer model(i.e. PROSPECT) was used to simulate the leaf reflectance spectra from 400-800nm under varying chlorophyll concentrations. Then 6 combinations of bands sensitive to chlorophyll concentrations were chosen for the calculation of their NDVIs. Simulated spectral response functions were applied to calculate the synthesis reflectance spectra at the intervals of 5, 10 and 20 nm respectively, and then corresponding NDVIs were calculated. The change of correlation coefficients between the NDVIs and the leaf chlorophyll concentrations were examined. Results showed that some NDVIs had a good performance with increasing bandwidth, whereas response of different NDVIs to the 4 bandwidths were different generally. Our results suggested that the improvement of spectral resolution was not necessary for some NDVIs to estimate leaf chlorophyll.

Analysis of optimal narrow band RVI for estimating foliar photosynthetic pigments based on PROSPECT model

Remote sensing is an effective tool to estimate foliar pigments contents with the analysis of vegetation index. The crucial issue is how to choose the optimal bands-combination to conduct the vegetation index. In this study, RVI, a vegetation index computed by the reflectance of Red and NIR bands, has been used to estimate the contents of chlorophyll and carotenoid. The reflectance of the two bands forming the narrow band RVI was simulated by the PROSPECT model. The possible combinations of narrow band RVI were examined from 400 nm to 800 nm. The results showed that: At the leaf level, estimation of chlorophyll content can be identified in narrow band RVI. Ranges for these bands included: (1) 549-589nm, 616-636nm or 729-735nm combined with 434-454nm; (2) 663-688nm, 710-717nm, 719-728nm or 730- 739nm combined with 549-561nm; (3) 663-688nm combined with 569-615nm. However, no valid narrow-band RVI for the estimation of carotenoid content was successfully identified. Our results also showed that two rules should be followed when choosing optimal bands-combination: (1) the selected bands must have minimal interference from other biochemical constituents; (2) there should be distinct differences between the sensitivities of the bands selected for particular pigments.

Estimating leaf photosynthetic pigments information by stepwise multiple linear regression analysis and a leaf optical model

Leaf pigments are key elements for plant photosynthesis and growth. Traditional manual sampling of these pigments is labor-intensive and costly, which also has the difficulty in capturing their temporal and spatial characteristics. The aim of this work is to estimate photosynthetic pigments at large scale by remote sensing. For this purpose, inverse model were proposed with the aid of stepwise multiple linear regression (SMLR) analysis. Furthermore, a leaf radiative transfer model (i.e. PROSPECT model) was employed to simulate the leaf reflectance where wavelength varies from 400 to 780 nm at 1 nm interval, and then these values were treated as the data from remote sensing observations. Meanwhile, simulated chlorophyll concentration (Cab), carotenoid concentration (Car) and their ratio (Cab/Car) were taken as target to build the regression model respectively. In this study, a total of 4000 samples were simulated via PROSPECT with different Cab, Car and leaf mesophyll structures as 70% of these samples were applied for training while the last 30% for model validation. Reflectance (r) and its mathematic transformations (1/r and log (1/r)) were all employed to build regression model respectively. Results showed fair agreements between pigments and simulated reflectance with all adjusted coefficients of determination (R2) larger than 0.8 as 6 wavebands were selected to build the SMLR model. The largest value of R2 for Cab, Car and Cab/Car are 0.8845, 0.876 and 0.8765, respectively. Meanwhile, mathematic transformations of reflectance showed little influence on regression accuracy. We concluded that it was feasible to estimate the chlorophyll and carotenoids and their ratio based on statistical model with leaf reflectance data.

Assimilation of remote sensing data into crop growth model to improve the estimation of regional winter wheat yield

Accurate regional crop growth monitoring and yield prediction is very critical for the national food security assessment and sustainable development of agriculture, especially for China, which has the largest population in the world. Remote sensing data and crop growth model have been successfully used in the crop production prediction. However, both of them have inherent limitation and uncertainty. The data assimilation method which combines crop growth model and remotely sensed data has been proven to be the most effective method in regional yield estimation. The aim of this paper is to improve the estimation of regional winter wheat yield of crop growth model by using data assimilation schemes with Ensemble Kalman Filter (EnKF) algorithm. WOrld FOod STudies (WOFOST) crop growth model was chosen as the crop growth model which was calibrated and validated by the field measured data. MODIS Leaf Area Index (LAI) values were used as remote sensing observations to adjust the LAI simulated by the WOFOST model based on EnKF. The results illustrate that the EnKF algorithm has significantly improved the regional winter wheat yield estimates over the WOFOST simulation without assimilation in both potential and water-limited modes. Although this study clearly implies that the assimilation of the remotely sensed data into crop growth model with EnKF algorithm has the potential to improve the prediction of regional crop yield and has great potential in agricultural applications, high resolution meteorological data and detailed crop field management are necessary to reach a high accuracy of regional crop yield estimation.