Discriminating the papyrus vegetation (Cyperus papyrus L.) and its co-existent species using random forest and hyperspectral data resampled to HYMAP

Techniques for discriminating swamp wetland species are critical for the rapid assessment and proactive management of wetlands. In this study, we tested whether the random forest (RF) algorithm could discriminate between papyrus swamp and its co-existent species (Phragmites australis, Echinochloa pyramidalis and Thelypteris interrupta) using in situ canopy reflectance spectra. Canopy spectral measurements were taken from the species using analytical spectral devices but later resampled to Hyperspectral Mapper (HYMAP) resolution. The RF algorithm and a simple forward variable selection (FVS) technique were used to identify key wavelengths for discriminating papyrus swamp and its co-existence species. The method yielded 10 wavelengths located in the visible and short-wave infrared portions of the electromagnetic spectrum with a lowest out-of-bag (OOB) estimate error rate of 9.5% and .632+?bootstrap error of 8.95%. The use of RF as a classification algorithm resulted in overall accuracy of 90.50% and a kappa value of 0.87, with individual class accuracies ranging from 93.73% to 100%. Additionally, the results from this study indicate that the RF algorithm produces better classification results than conventional classification trees (CTs) when using all HYMAP wavelengths (n?=?126) and when using wavelengths selected by the FVS technique.     

Estimating aboveground biomass of grassland having a high canopy cover: an exploratory analysis of in situ hyperspectral data

To improve the estimation of aboveground biomass of grassland having a high canopy cover based on remotely sensed data, we measured in situ hyperspectral reflectance and the aboveground green biomass of 42 quadrats in an alpine meadow ecosystem on the Qinghai–Tibetan Plateau. We examined the relationship between aboveground green biomass and the spectral features of original reflectance, first-order derivative reflectance (FDR), and band-depth indices by partial least squares (PLS) regression, as well as the relationship between the aboveground biomass and narrow-band vegetation indices by linear and nonlinear regression analyses. The major findings are as follows. (1) The effective portions of spectra for estimating aboveground biomass of a high-cover meadow were within the red-edge and near infrared (NIR) regions. (2) The band-depth ratio (BDR) feature, using NIR region bands (760–950 nm) in combination with the red-edge bands, yields the best predictive accuracy (RMSE?=?40.0 g m^?2) for estimating biomass among all the spectral features used as independent variables in the partial least squares regression method. (3) The ratio vegetation index (RVI2) and the normalized difference vegetation index (NDVI2) proposed by Mutanga and Skidmore (Mutanga, O. and Skidmore, A.K., 2004a Mutanga, O. and Skidmore, A. K. 2004a. Narrow band vegetation indices solve the saturation problem in biomass estimation. International Journal of Remote Sensing, 25: 1–16.  [Google Scholar], Narrow band vegetation indices solve the saturation problem in biomass estimation. International Journal of Remote Sensing, 25, pp. 1–6) are better correlated to the aboveground biomass than other VIs (R ²?=?0.27 for NDVI2 and 0.26 for RVI2), while RDVI, TVI and MTV1 predicted biomass with higher accuracy (RMSE?=?37.2 g m^?2, 39.9 g m^?2 and 39.8 g m^?2, respectively). Although all of the models developed in this study are probably acceptable, the models developed in this study still have low accuracy, indicating the urgent need for further efforts.     

Assessment of plant nitrogen stress in wheat (Triticum aestivum L.) through hyperspectral indices

A field experiment with wheat was conducted with four different nitrogen and four different water stress levels, and hyperspectral reflectances in the 350–2500 nm range were recorded at six crop phenostages for two years (2009–2010 and 2010–2011). Thirty-two hyperspectral indices were determined using the first-year reflectance data. Plant nitrogen (N) status, characterized by leaf nitrogen content (LNC) and plant nitrogen accumulation (PNA), showed the highest R ² with the spectral indices at the booting stage. The best five predictive equations for LNC were based on the green normalized difference vegetation index (GNDVI), normalized difference chlorophyll index (NDCI), normalized difference₇₀₅ (ND₇₀₅) index, ratio index-1dB (RI-1dB) and Vogelman index a (VOGa). Their validation using the second-year data showed high R ² (>0.80) and ratio of performance to deviation (RPD; >2.25) and low root mean square error (RMSE; <0.24) and relative error (<10%). For PNA, five predictive equations with simple ratio pigment index (SRPI), photochemical reflectance index (PRI), modified simple ratio₇₀₅ (mSR₇₀₅), modified normalized difference₇₀₅ (mND₇₀₅) and normalized pigment chlorophyll index (NPCI) as predicting indices yielded the best relations with high R ² > 0.80. The corresponding RMSE and RE of these ranged from 1.39 to 1.13 and from 24.5% to 33.3%, respectively. Although the predicted values show good agreement with the observed values, the prediction of LNC is more accurate than PNA, as indicated by higher RMSE and very high RE for the latter. Hence, the plant nitrogen stress of wheat can be accurately assessed through the prediction of LNC based on the five identified reflectance indices at the booting stage.     

Calibration and validation of hyperspectral indices for the estimation of broadleaved forest leaf chlorophyll content,leaf mass per area,leaf area index and leaf canopy biomass

This article aims at finding efficient hyperspectral indices for the estimation of forest sun leaf chlorophyll content (CHL, µg cm_leaf^? 2), sun leaf mass per area (LMA, g_{dry matter} m_leaf^? 2), canopy leaf area index (LAI, m²_leaf m_soil^? 2) and leaf canopy biomass (B_leaf, g_{dry matter} m_soil^? 2). These parameters are useful inputs for forest ecosystem simulations at landscape scale. The method is based on the determination of the best vegetation indices (index form and wavelengths) using the radiative transfer model PROSAIL (formed by the newly-calibrated leaf reflectance model PROSPECT coupled with the multi-layer version of the canopy radiative transfer model SAIL). The results are tested on experimental measurements at both leaf and canopy scales. At the leaf scale, it is possible to estimate CHL with high precision using a two wavelength vegetation index after a simulation based calibration. At the leaf scale, the LMA is more difficult to estimate with indices. At the canopy scale, efficient indices were determined on a generic simulated database to estimate CHL, LMA, LAI and Bleaf in a general way. These indices were then applied to two Hyperion images (50 plots) on the Fontainebleau and Fougères forests and portable spectroradiometer measurements. They showed good results with an RMSE of 8.2 µg cm^? 2 for CHL, 9.1 g m^? 2 for LMA, 1.7 m² m^? 2 for LAI and 50.6 g m^? 2 for Bleaf. However, at the canopy scale, even if the wavelengths of the calibrated indices were accurately determined with the simulated database, the regressions between the indices and the biophysical characteristics still had to be calibrated on measurements. At the canopy scale, the best indices were: for leaf chlorophyll content: ND_chl = (ρ₉₂₅ ? ρ₇₁₀)/(ρ₉₂₅ + ρ₇₁₀), for leaf mass per area: ND_LMA = (ρ₂₂₆₀ ? ρ₁₄₉₀)/(ρ₂₂₆₀ + ρ₁₄₉₀), for leaf area index: D_LAI = ρ₁₇₂₅ ? ρ₉₇₀, and for canopy leaf biomass: ND_Bleaf = (ρ₂₁₆₀ ? ρ₁₅₄₀)/(ρ₂₁₆₀ + ρ₁₅₄₀).     

Forest attribute imputation using machine-learning methods and ASTER data: comparison of k-NN,SVR and random forest regression algorithms

In a small case study of mixed hardwood Hyrcanian forests of Iran, three non-parametric methods, namely k-nearest neighbour (k-NN), support vector machine regression (SVR) and tree regression based on random forest (RF), were used in plot-level estimation of volume/ha, basal area/ha and stems/ha using field inventory and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data. Relevant pre-processing and processing steps were applied to the ASTER data for geometric and atmospheric correction and for enhancing quantitative forest parameters. After collecting terrestrial information on trees in the 101 samples, the volume, basal area and tree number per hectare were calculated in each plot. In the k-NN implementation using different distance measures and k, the cross-validation method was used to find the best distance measure and optimal k. In SVR, the best regularized parameters of four kernel types were obtained using leave-one-out cross-validation. RF was implemented using a bootstrap learning method with regularized parameters for decision tree model and stopping. The validity of performances was examined using unused test samples by absolute and relative root mean square error (RMSE) and bias metrics. In volume/ha estimation, the results showed that all the three algorithms had similar performances. However, SVR and RF produced better results than k-NN with relative RMSE values of 28.54, 25.86 and 26.86 (m³ ha^–1), respectively, using k-NN, SVR and RF algorithms, but RF could generate unbiased estimation. In basal area/ha and stems/ha estimation, the implementation results of RF showed that RF was slightly superior in relative RMSE (18.39, 20.64) to SVR (19.35, 22.09) and k-NN (20.20, 21.53), but k-NN could generate unbiased estimation compared with the other two algorithms used.     

A fuzzy regression model based on distances and random variables with crisp input and fuzzy output data: a case study in biomass production

Least-squares technique is well-known and widely used to determine the coefficients of a explanatory model from observations based on a concept of distance. Traditionally, the observations consist of pairs of numeric values. However, in many real-life problems, the independent or explanatory variable can be observed precisely (for instance, the time) and the dependent or response variable is usually described by approximate values, such as “about £300\pounds300” or “approximately $500”, instead of exact values, due to sources of uncertainty that may affect the response. In this paper, we present a new technique to obtain fuzzy regression models that consider triangular fuzzy numbers in the response variable. The procedure solves linear and non-linear problems and is easy to compute in practice and may be applied in different contexts. The usefulness of the proposed method is illustrated using simulated and real-life examples.     

In situ examination of the relationship between various vegetation indices and canopy phenology in an evergreen coniferous forest,Japan

We examined the relationship between four vegetation indices and tree canopy phenology in an evergreen coniferous forest in Japan based on observations made using a spectral radiometer and a digital camera at a daily time step during a 4 year period. The colour of the canopy surface of Japanese cedar (Cryptomeria japonica) changed from yellowish-green to whitish-green from late May to July and turned reddish-green in winter. The normalized difference vegetation index (NDVI), enhanced vegetation index (EVI) and plant area index (PAI) showed no seasonality. In contrast, the green–red ratio vegetation index (GRVI) increased from March to June and then decreased gradually from July to December, resulting in a bell-shaped curve. GRVI revealed seasonal changes in the colour of the canopy surface. GRVI correlated more positively with the evaluated maximum photosynthetic rate for the whole forest canopy, A _max, than did NDVI or EVI. These results suggest the possibility that GRVI is more useful than NDVI and EVI for capturing seasonal changes in photosynthetic capacity, as the green and red reflectances are strongly influenced by changes in leaf pigments in this type of forest.     

Associated changes in physiological parameters and spectral reflectance indices in olive (Olea europaea L.) leaves in response to different levels of water stress

This study aimed to determine whether modification of physiological parameters could be detected remotely by monitoring the spectral reflectance of olive leaves in response to different degrees of drought. Three different drought intensities were simulated: (a) a mild drought by feeding abscisic acid to detached branches; (b) a rapid and severe drought by detaching leaves and letting them dry over several hours; (c) a relatively slow drought caused by withholding water to potted olive plants. The three degrees of stress affected gas exchange and chlorophyll fluorescence. When the inhibition of photosynthesis occurred within an hour it was not accompanied by a parallel reduction in chlorophyll concentration in the carotenoid to chlorophyll ratio. Consequently, changes in spectral reflectance in the visible region, e.g. in PRI (photochemical reflectance index) and FRI (fluorescence reflectance indices) were not significantly induced. In contrast, when the inhibition of photosynthesis caused by slow developing drought was prolonged (i.e. more than 24 hours) and led to a decrease in chlorophyll concentration and to a simultaneous increase in carotenoid to chlorophyll ratio, there were significant changes in the visible region of the leaf spectral reflectance and, in turn, in PRI and FRI. We defined 16 new reference wavelengths, from visible to SWIR regions, which are sensitive to both fast‐developing and slow‐developing stresses. These reference wavelengths were used to develop an algorithm, the Relative Reflectance Increment (RRI), that was linearly related to changes in relative water content (RWC, r ² = 0.733). This algorithm showed that the 1455 nm wavelength is highly affected by drought. This wavelength was therefore used to elaborate the water content reflectance index that was inversely related to RWC (r ² = 0.702).     

Methodology to map the spread of an invasive plant (Lantana camara L.) in forest ecosystems using Indian remote sensing satellite data

The primary objective of this paper is to evaluate the utility of different Indian remote sensing sensors for detection, mapping and patch size estimation of Lantana camara L. (Kurri). The latter, of the family Verbinaceae, is one of the most aggressive invasive plant species and has colonized large areas of forest land in the Himalayan foothills (Shiwalik range). The State Forest Departments of India are planning to develop a suitable strategy to halt its invasion. The first step in this direction is to have accurate information on the location and spread of the plant in spatial format. The test site is part of the forest of the Rajaji National Park, Uttarakhand. Indian Remote Sensing-Linear Imaging Self-Scanning Sensor (IRS-LISS) III (multi-spectral, 23.5 m), IRS-LISS IV (multi-spectral, 5.8 m), Cartosat-1 (Panchromatic, 2.5 m) and a merged image of LISS IV and Cartosat-1 using Brovey fusion techniques were used to map Lantana camara L. Further improvement was obtained using texture analysis. The study demonstrates the potentiality of LISS IV and Cartosat-1 data for detection and mapping of Lantana camara L. The results show the feasibility of developing a semi-automated procedure to map and analyse the distribution of Lantana in forest areas.     

On intra-annual EVI variability in the dry season of tropical forest: A case study with MODIS and hyperspectral data

Considerable controversy is associated with dry season increases in the Enhanced Vegetation Index (EVI), observed using the Moderate Resolution Imaging Spectroradiometer (MODIS), compared with field-based estimates of decreasing plant productivity. Here, we investigate potential causes of intra-annual variability by comparing EVI from mature forest with field-measured Leaf Area Index (LAI) to validate space-based observations. EVI was calculated from 19 nadir and off-nadir Hyperion images in the 2005 dry season, and inspected for consistency with MODIS observations from 2004 to 2009. The objective was to evaluate the possible influence of the view-illumination geometry and of canopy foliage and leaf flush on the EVI. Spectral mixture models were used to evaluate the relationship between EVI and the shade fraction, a measure that varies with pixel brightness. MODIS LAI values were compared with LAI estimated using hemispherical photographs taken in two field campaigns in the dry season. To keep LAI and leaf flush conditions as constant variables and vary solar illumination, we used airborne Hyperspectral Mapper (Hymap) data acquired over mature forest from another region on the same day but with two distinct solar zenith angles (SZA) (29° and 53°). Results showed that intra-annual variability in MODIS and nadir Hyperion EVI in the dry season of tropical forest were driven by solar illumination effects rather than changes in LAI. The reflectance of the MODIS and Hyperion blue, red and near infrared (NIR) bands was higher at the end of the dry season because of the predominance of sunlit canopy components for the sensors due to decreasing SZA from June (44°) to September (26°). Because EVI was highly correlated with the reflectance of the NIR band used to generate it (r of + 0.98 for MODIS and + 0.88 for Hyperion), this vegetation index followed the general NIR pattern, increasing with smaller SZA towards the end of the dry season. Hyperion EVI was inversely correlated with the shade fraction (r = − 0.93). Changes in canopy foliage detected from MODIS LAI data were not consistent with LAI estimates from hemispherical photographs. Although further research is necessary to measure the impact of leaf flush on intra-annual EVI variability in the Querência region, analysis of Hymap data with fixed LAI and leaf flush conditions confirmed the influence of the illumination effects on the EVI.     

Isolation and in silico evaluation of antidiabetic molecules of Cynodon dactylon (L.)

Cynodon dactylon is a potential source of metabolites such as flavanoids, alkaloids, glycosides and β-sitosterol and has been traditionally employed to treat urinary tract and other microbial infections and dysentery. The present work attempts to evaluate the activity of C. dactylon extracts for glycemic control. Aqueous extracts of C. dactylon analyzed by HPLC–ESI MS have identified the presence of apigenin, luteolin, 6-C-pentosyl-8-C-hexosyl apigenin and 6-C-hexosyl-8-C-pentosyl luteolin. Evaluation of hypoglycemic activity through an extensive in silico docking approach with PPARγ (Peroxisome Proliferator-Activated Receptor), GLUT-4 (glucose transporter-4) and SGLT2 (sodium glucose co-transporter-2) revealed that luteolin, apigenin, 6-C-pentosyl-8-C-hexosyl apigenin, 6-C-hexosyl-8-C-pentosyl luteolin interact with SGLT2. Interactions of these molecules with Gln 295 and Asp 294 residues of SGLT2 have been shown to compare well with that of the phase III drug, dapagliflozin. These residues have been proven to be responsible for sugar sensing and transport. This work establishes C. dactylon extract as a potential SGLT2 inhibitor for diabetic neuropathy thus enabling a possibility of this plant extract as a new alternative to existing diabetic approaches.     

Estimating aboveground carbon in a catchment of the Siberian forest tundra: Combining satellite imagery and field inventory

This study was part of an interdisciplinary research project on soil carbon and phytomass dynamics of boreal and arctic permafrost landscapes. The 45 ha study area was a catchment located in the forest tundra in northern Siberia, approximately 100 km north of the Arctic Circle.The objective of this study was to estimate aboveground carbon (AGC) and assess and model its spatial variability. We combined multi-spectral high resolution remote sensing imagery and sample based field inventory data by means of the k-nearest neighbor (k-NN) technique and linear regression.Field data was collected by stratified systematic sampling in August 2006 with a total sample size of n = 31 circular nested sample plots of 154 m² for trees and shrubs and 1 m² for ground vegetation. Destructive biomass samples were taken on a sub-sample for fresh weight and moisture content. Species-specific allometric biomass models were constructed to predict dry biomass from diameter at breast height (dbh) for trees and from elliptic projection areas for shrubs.Quickbird data (standard imagery product), acquired shortly before the field campaign and archived ASTER data (Level-1B product) of 2001 were geo-referenced, converted to calibrated radiances at sensor and used as carrier data. Spectral information of the pixels which were located in the inventory plots were extracted and analyzed as reference set. Stepwise multiple linear regression was applied to identify suitable predictors from the set of variables of the original satellite bands, vegetation indices and texture metrics. To produce thematic carbon maps, carbon values were predicted for all pixels of the investigated satellite scenes. For this prediction, we compared the kNN distance-weighted classifier and multiple linear regression with respect to their predictions.The estimated mean value of aboveground carbon from stratified sampling in the field is 15.3 t/ha (standard error SE = 1.50 t/ha, SE% = 9.8%). Zonal prediction from the k-NN method for the Quickbird image as carrier is 14.7 t/ha with a root mean square error RMSE = 6.42 t/ha, RMSE_r = 44%) resulting from leave-one-out cross-validation. The k-NN-approach allows mapping and analysis of the spatial variability of AGC. The results show high spatial variability with AGC predictions ranging from 4.3 t/ha to 28.8 t/ha, reflecting the highly heterogeneous conditions in those permafrost-influenced landscapes. The means and totals of linear regression and k-NN predictions revealed only small differences but some regional distinctions were recognized in the maps.     

Relationships of leaf nitrogen concentration and canopy nitrogen density with spectral features parameters and narrow-band spectral indices calculated from field winter wheat (Triticum aestivum L.) spectra

Remote sensing is a promising tool that provides quantitative and timely information for crop stress detection over large areas. Nitrogen (N) is one of the important nutrient elements influencing grain yield and quality of winter wheat (Triticum aestivum L.). In this study, canopy spectral parameters were evaluated for N status assessment in winter wheat. A winter wheat field experiment with 25 different cultivars was conducted at the China National Experimental Station for Precision Agriculture, Beijing, China. Wheat canopy spectral reflectance over 350–2500 nm at different stages was measured with an ASD FieldSpec Pro 2500 spectrometer (Analytical Spectral Devices, Boulder, CO, USA) fitted with a 25° field of view (FOV) fibre optic adaptor. Thirteen narrow-band spectral indices, three spectral features parameters associated with the absorption bands centred at 670 and 980 nm and another three related to reflectance maximum values located at 560, 920, 1690 and 2230 nm were calculated and correlated with leaf N concentration (LNC) and canopy N density (CND). The results showed that CND was a more sensitive parameter than LNC in response to the variation of canopy-level spectral parameters. The correlation coefficient values between LNC and CND, on the one hand, and narrow-band spectral indices and spectral features parameters, on the other hand, varied with the growth stages of winter wheat, with no predominance of a single spectral parameter as the best variable. The differences in correlation results for the relationships of CND and LNC with narrow-band spectral indices and spectral features parameters decreased with wheat plant developing from Feekes 4.0 to Feekes 11.1. The red edge position (REP) was demonstrated to be a good indicator for winter wheat LNC estimation. The absorption band depth (ABD) normalized to the area of absorption feature (NBD) at 670 nm (NBD670) was the most reliable indicator for winter wheat canopy N status assessment.     

Simultaneous retrieval of CO2 and aerosols in a plume from hyperspectral imagery: application to the characterization of forest fire smoke using AVIRIS data

Hyperspectral imagery is a widely used technique to study atmospheric composition. For several years, many methods have been developed to estimate the abundance of gases. However, existing methods do not simultaneously retrieve the properties of aerosols and often use standard aerosol models to describe the radiative impact of particles. This approach is not suited to the characterization of plumes, because plume particles may have a very different composition and size distribution from aerosols described by the standard models given by radiative transfer codes. This article presents a new method to simultaneously retrieve carbon dioxide (CO₂) and aerosols inside a plume, combining an aerosol retrieval algorithm using visible and near-infrared (VNIR) wavelengths and a CO₂ estimation algorithm using shortwave infrared (SWIR) wavelengths. The microphysical properties of the plume particles, obtained after aerosol retrieval, are used to calculate their optical properties in the SWIR. Then, a database of atmospheric terms is generated with the radiative transfer code, Moderate Resolution Atmospheric Transmission (MODTRAN). Finally, pixel radiances around the 2.0 μm absorption feature are used to retrieve the CO₂ abundances. After conducting a signal sensitivity analysis, the method was applied to two airborne visible/infrared imaging spectrometer (AVIRIS) images acquired over areas of biomass burning. For the first image, in situ measurements were available. The results show that including the aerosol retrieval step before the CO₂ estimation: (1) induces a better agreement between in situ measurements and retrieved CO₂ abundances (the CO₂ overestimation of about 15%, induced by neglecting aerosols has been corrected, especially for pixels where the plume is not very thick); (2) reduces the standard deviation of estimated CO₂ abundance by a factor of four; and (3) causes the spatial distribution of retrieved concentrations to be coherent.     

Comparison of two hyperspectral imaging and two laser-induced fluorescence instruments for the detection of zinc stress and chlorophyll concentration in bahia grass (Paspalum notatum Flugge.)

Bahia grass (Paspalum notatum Flugge.) plants were grown in silica sand and irrigated daily with one of five levels of Zn (0, 0.5, 25, 50, or 100 mg l⁻¹) to determine the effects of the heavy metal on the growth and development of plant canopies. Healthy and stressed plants were measured with two hyperspectral imagers, laser-induced fluorescence spectroscopy (LIFS), and laser-induced fluorescence imaging (LIFI) systems in order to determine if the four handheld remote sensing instruments were equally capable of detecting plant stress and measuring canopy chlorophyll levels in bahia grass. Symptoms of bahia grass plants grown at deficient (0 mg l⁻¹) or toxic (25, 50, or 100 mg l⁻¹) concentrations of Zn were dominated by leaf chlorosis and plant stunting. Leaf fresh weight, leaf dry weight, CO₂ assimilation, total chlorophyll, and leaf thickness followed (+) quadratic models in which control plants (0.5 mg l⁻¹ Zn) exhibited higher responses than plants grown at either deficient or toxic levels of Zn. Normalized difference vegetation index [NDVI=(NIR−Red)/(NIR+Red)] and ratio vegetation index [RVI=R₇₅₀/R₇₀₀, in which R denotes reflectance] values were calculated for calibrated digital images from both hyperspectral imagers. The NDVI and RVI values from both hyperspectral imagers were fit best by (+) quadratic models when treatments were constrained between 0 and 100 mg l⁻¹ Zn, but were fit best by linear regression models with (−) slopes when treatments were constrained between 0.5 and 100 mg l⁻¹ Zn. Furthermore, both NDVI and RVI algorithms were effective in predicting the concentrations of chlorophyll in canopies of bahia grass grown at the various levels of Zn. In contrast, red/far-red (R/FR) fluorescence ratios estimated from leaf fluorescence values measured with the LIFS and LIFI instruments were fit best by (−) quadratic models when treatments were constrained between 0 and 100 mg l⁻¹ Zn, but were fit best by linear regression models with (+) slopes when treatments were constrained between 0.5 and 100 mg l⁻¹ Zn. A series of regression analyses were conducted among plant biometric, biochemical, and leaf anatomical parameters (treated as independent variables) and the remote sensing algorithms, NDVI, RVI, blue/green (BL/GR), and R/FR (treated as dependant variables). In general, residuals were significantly higher for NDVI and RVI models compared to the BL/GR and R/FR models indicating that the NDVI and RVI algorithms were able to measure total chlorophyll and plant biomass more accurately than the BL/GR and R/FR algorithms. However, unique capabilities of LIFS and LIFI instruments continue to argue for the development of laser-induced fluorescence remote sensing technologies.     

Comparing the performance of multispectral vegetation indices and machine-learning algorithms for remote estimation of chlorophyll content: a case study in the Sundarbans mangrove forest

Optical vegetation indices (VIs) have been used to retrieve and assess biophysical variables from satellite reflectance data. These indices, however, also are sensitive to a number of confounding factors, such as canopy geometry, soil optical properties, and solar position. This suggests that VIs should be used cautiously for biophysical parameter estimation. Among biophysical variables, chlorophyll content is of particular importance as an indicator of photosynthetic activity. The goal of this study is to investigate the performance of multispectral optical VIs for chlorophyll content estimation in the world’s largest mangrove forest, the Sundarbans, and to compare these with machine-learning algorithms (MLAs). To this end, we have investigated the performance of 15 multispectral VIs and six state-of-the-art MLAs that are widely used for adaptive data fitting. The MLAs are Artificial Neural Networks (ANNs), Genetic Algorithm (GA), Gaussian Processes for Machine Learning (GPML), Kernel Ridge Regression (KRR), Locally Weighted Polynomials (LWP), and Multivariate Adaptive Regression Splines (MARS). We use an in situ data set of reflectance and chlorophyll measurements to develop and validate our models. Each MLA was evaluated 500 times with random partitions of training and validation data. Results showed that the weight optimization and term selection used within GA produce the most reliable chlorophyll content estimation. However, green normalized difference VI (GNDVI) is a simple and computationally efficient VI that produces results that are nearly as accurate as GA in terms of model fit and performance. Results also show that all methods except ANNs and MARS produce a quasi-linear relationship between spectral reflectance and chlorophyll content. Statistical transformations of GNDVI and chlorophyll content have the capability of further reducing model error.     

Land-use/cover classification in a heterogeneous coastal landscape using RapidEye imagery: evaluating the performance of random forest and support vector machines classifiers

Mapping of patterns and spatial distribution of land-use/cover (LULC) has long been based on remotely sensed data. In the recent past, efforts to improve the reliability of LULC maps have seen a proliferation of image classification techniques. Despite these efforts, derived LULC maps are still often judged to be of insufficient quality for operational applications, due to disagreement between generated maps and reference data. In this study we sought to pursue two objectives: first, to test the new-generation multispectral RapidEye imagery classification output using machine-learning random forest (RF) and support vector machines (SVM) classifiers in a heterogeneous coastal landscape; and second, to determine the importance of different RapidEye bands on classification output. Accuracy of the derived thematic maps was assessed by computing confusion matrices of the classifiers’ cover maps with respective independent validation data sets. An overall classification accuracy of 93.07% with a kappa value of 0.92, and 91.80 with a kappa value of 0.92 was achieved using RF and SVM, respectively. In this study, RF and SVM classifiers performed comparatively similarly as demonstrated by the results of McNemer’s test (Z = 1.15). An evaluation of different RapidEye bands using the two classifiers showed that incorporation of the red-edge band has a significant effect on the overall classification accuracy in vegetation cover types. Consequently, pursuit of high classification accuracy using high-spatial resolution imagery on complex landscapes remains paramount.