Mapping leaf area index over a mixed natural forest area in the flooding season using ground-based measurements and Landsat TM imagery

Leaf area index (LAI) is an important structural parameter in terrestrial ecosystem modelling and management. Therefore, it is necessary to conduct an investigation on using moderate-resolution satellite imagery to estimate and map LAI in mixed natural forests in southeastern USA. In this study, along with ground-measured LAI and Landsat TM imagery, the potential of Landsat 5 TM data for estimating LAI in a mixed natural forest ecosystem in southeastern USA was investigated and a modelling method for mapping LAI in a flooding season was developed. To do so, first, 70 ground-based LAI measurements were collected on 8 April 2008 and again on 1 August 2008 and 30 July 2009; TM data were calibrated to ground surface reflectance. Then univariate correlation and multivariate regression analyses were conducted between the LAI measurement and 13 spectral variables, including seven spectral vegetation indices (VIs) and six single TM bands. Finally, April 08 and August 08 LAI maps were made by using TM image data, a multivariate regression model and relationships between April 08 and August 08 LAI measurements. The experimental results indicate that Landsat TM imagery could be used for mapping LAI in a mixed natural forest ecosystem in southeastern USA. Furthermore, TM4 and TM3 single bands (R ² > 0.45) and the soil adjusted vegetation index, transformed soil adjusted vegetation index and non-linear vegetation index (R ² > 0.64) have produced the highest and second highest correlation with ground-measured LAI. A better modelling result (R ²?=?0.78, accuracy?=?73%, root mean square error (RMSE)?=?0.66) of the 10-predictor multiple regression model was obtained for estimating and mapping April 08 LAI from TM data. With a linear model and a power model, August 08 LAI maps were successfully produced from the April 08 LAI map (accuracy?=?79%, RMSE?=?0.57), although only 58–65% of total variance could be accounted for by the linear and non-linear models.     

Inversion of a forest reflectance model to estimate structural canopy variables from hyperspectral remote sensing data

The potential of canopy reflectance modelling to retrieve simultaneously several structural variables in managed Norway spruce stands was investigated using the “Invertible Forest Reflectance Model”, INFORM. INFORM is an innovative extension of the FLIM model, with crown transparency, infinite crown reflectance and understory reflectance simulated using physically based sub-models (SAILH, LIBERTY and PROSPECT). The INFORM model was inverted with hyperspectral airborne HyMap data using a neural network approach. INFORM based estimates of forest structural variables were produced using site-specific ranges of stand structural variables. A relatively simple three layer feed-forward backpropagation neural network with two input neurons, one neuron in the hidden layer and three output neurons was employed to map leaf area index (LAI), crown coverage and stem density.To identify the optimum 2-band spectral subset to be used in the inversion process, all 2-band combinations of the HyMap dataset were systematically evaluated for model inversion. Field measurements of structural variables from 39 forest stands were used to validate the maps produced from HyMap imagery. Using two HyMap wavebands at 837 nm and 1148 nm the obtained accuracy of the LAI map amounts to an rmse of 0.58 (relative rmse = 18% of mean, R² = 0.73). With HyMap data resampled to Landsat TM spectral bands and using two “optimum” bands at 840 nm and 1650 nm, rmse was 0.66 and relative rmse 21%. In contrast to approaches based on empirical relations between spectral vegetation indices and structural variables, the main advantage of the inversion approach is that it does not require previous calibration.     

Retrieval of leaf area index for a coniferous forest by inverting a forest reflectance model

The aim of this paper was to serve as a pilot study for running a physically based forest reflectance model through an operational forest management data base in Finnish coniferous forests. The LAI values of 250 boreal coniferous stands were retrieved with the physically based model by inversion from a SPOT HRVIR1 image. The use of three spectral vegetation indices (NDVI, RSR and MSI) in LAI estimation was tested for the same stands. Ground-truth LAI was based on an allometric model which can be applied to routine stand inventory data. Stand reflectances were computed as an average of reflectances of the pixels located within the digital stand borders.The relationships of LAI and spectral vegetation indices calculated from the SPOT data were very scattered. RSR exhibited the widest range of values (and the highest correlation with LAI), suggesting it to be more dynamic than MSI or NDVI. Inversion of the reflectance model was done twice: first using as simultaneous input three wavelength bands (red, NIR and MIR), then only the red and NIR bands. The aim was to observe whether including the MIR band in the inversion would improve the inverted LAI estimates or if using only the red and NIR bands would result in the same reliability of inverted values. The motivation for examining the influence of the MIR band resulted from several recent studies from the boreal zone which suggest that the pronounced understory effect could be minimized by the inclusion of the MIR band. The LAI values inverted by the model were slightly larger than the ground-truth LAI values. A minor improvement in LAI estimates was observed after the inclusion of the MIR band in reflectance model inversion. The errors in the ground-truth LAI were uncertain and the background understory reflectance was expected to be highly variable. Thus, the quality of the data used may be to a large extent responsible for the observed low utility of the tested channels.     

Needle chlorophyll content estimation through model inversion using hyperspectral data from boreal conifer forest canopies

Leaf chlorophyll content in coniferous forest canopies, a measure of stand condition, is the target of studies and models linking leaf reflectance and transmittance and canopy hyperspectral reflectance imagery. The viability of estimation of needle chlorophyll content from airborne hyperspectral optical data through inversion of linked leaf level and canopy level radiative transfer models is discussed in this paper. This study is focused on five sites of Jack Pine (Pinus banksiana Lamb.) in the Algoma Region (Canada), where field, laboratory and airborne data were collected in 1998 and 1999 campaigns. Airborne hyperspectral CASI data of 72 bands in the visible and near-infrared region and 2 m spatial resolution were collected from 20×20 m study sites of Jack Pine in 2 consecutive years. It was found that needle chlorophyll content could be estimated at the leaf level (r²=0.4) by inversion of the PROSPECT leaf model from needle reflectance and transmittance spectra collected with a special needle carrier apparatus coupled to the Li-Cor 1800 integrating sphere. The Jack Pine forest stands used for this study with LAI>2, and the high spatial resolution hyperspectral reflectance collected, allowed the use of the SPRINT canopy reflectance model coupled to PROSPECT for needle chlorophyll content estimation by model inversion. The optical index R750/R710 was used as the merit function in the numerical inversion to minimize the effect of shadows and LAI variation in the mean canopy reflectance from the 20×20 m plots. Estimates of needle pigment content from airborne hyperspectral reflectance using this linked leaf-canopy model inversion methodology showed an r²=0.4 and RMSE=8.1 μg/cm² when targeting sunlit crown pixels in Jack Pine sites with pigment content ranging between 26.8 and 56.8 μg/cm² (1570-3320 μg/g).     

Modelling forest stand volume and tree density using Landsat ETM+ data

Estimation of stand volume and tree density in a large area using remotely sensed data has considerable significance for sustainable management of natural resources. In this paper, we explore likely relationships between forest stand characteristics and Landsat Enhanced Thematic Mapper Plus (ETM+) reflectance values. We used multivariate regression technique to predict stand volume and tree density. The result showed that a linear combination of greenness and difference vegetation index (DVI) were better predictors of stand volume (adjusted R² = 43%; root mean square error (RMSE) = 97.4 m³ ha^?1) than other ETM+ bands and vegetation indices. In addition, the regression model with ETM4 (near infrared band) and ETM5 (first shortwave band) as independent variables was a better predictor of tree density (adjusted R² = 73.4%; RMSE = 170.13 ha^?1) than other combinations of ETM⁺ bands and vegetation indices. Results obtained from this study demonstrate the significant relationship between forest stand characteristics and ETM+ reflectance values and the utility of transformed bands in modelling stand volume and tree density. Based on the results of this study, we conclude that ETM+ data are useful to estimate forest volume and density and to gain insights into its structural characteristics in our study area. Forest managers could use ETM+ data for gaining insights into stand characteristics and generating maps required for developing forest management plans and identifying locations within stands that require treatments and other interventions.     

Estimation of forest leaf area index from SPOT imagery using NDVI distribution over forest stands

Leaf area index (LAI) is a key parameter of atmosphere–vegetation exchanges, affecting the net ecosystem exchange and the productivity. At regional or continental scales, LAI can be estimated by remotely‐sensed spectral vegetation indices (SVI). Nevertheless, relationships between LAI and SVI show saturation for LAI values greater than 3–5. This is one of the principal limitations of remote sensing of LAI in forest canopies. In this article, a new approach is developed to determine LAI from the spatial variability of radiometric data. To test this method, in situ measurements for LAI of 40 stands, with three dominant species (European beech, oak and Scots pine) were available over 5 years in the Fontainebleau forest near Paris. If all years and all species are pooled, a good linear relationship without saturation is founded between average stand LAI measurements and a model combining the logarithm of the standard deviation and the skewness of the normalized difference vegetation index (NDVI) (R ² = 0.73 rmse = 1.08). We demonstrate that this relation can be slightly improved by using different linear models for each year and each species (R ² = 0.82 rmse = 0.86), but the standard deviation is less sensitive to the species and the year effects than the mean NDVI and is therefore a performing index.     

Canopy level estimations of chlorophyll and LAI for two tropical species (teak and bamboo) from Hyperion (EO1) data

Current economic development in tropical regions (especially in India, China, and Brazil) is putting tremendous pressure on tropical forest cover. Some of the dominant and economically important species are planted at large scale in these countries. Teak and bamboo are two important species of tropical regions because of their commercial and conservation values. Accurate estimates of foliar chemistry can help in evaluating the health status of vegetation in these regions. An attempt has been made to derive canopy level estimation of chlorophyll and leaf area index (LAI) for these species utilizing Hyperion data. Partial least square (PLS) regression analysis was carried out to identify the correlation between measured parameters (chlorophyll and LAI) and Hyperion reflectance spectra. PLS regression identified 600–750 nm as a sensitive spectral region for chlorophyll and 1000–1507 nm for LAI. The PLS regression model tested in this study worked well for the estimation of chlorophyll (R ² = 0.90, root mean square error (RMSE) = 0.182 for teak and R ² = 0.84, RMSE = 0.113 for bamboo) and for the estimation of LAI (R ² = 0.87, RMSE = 0.425). The lower predictive error obtained indicates the robustness of the data set and also of the applicability of the PLS regression analysis. Wavelengths recognized by the PLS regression model were utilized for the development of vegetation indices for estimating chlorophyll and LAI. Predictive performances of the developed simple ratios (SRs) were evaluated using the cross-validation method. SR 743/692 gave the best results for the prediction of chlorophyll with the leave-one-out cross-validation (LOO-CV) method (R ² = 0.73, RMSE = 0.28 for teak and R ² = 0.71, RMSE = 0.15 for bamboo). The normalized difference ratio (ND 1457/1084) gave the best results for the prediction of LAI with LOO-CV (R ² = 0.66, RMSE = 0.57). Ratios developed here can be tested for teak and bamboo cover spread in tropical regions with similar environmental conditions.     

Impact of understory vegetation on forest canopy reflectance and remotely sensed LAI estimates

Forest leaf area index (LAI), is an important variable in carbon balance models. However, understory vegetation is a recognized problem that limits the accuracy of satellite-estimated forest LAI. A canopy reflectance model was used to investigate the impact of the understory vegetation on LAI estimated from reflectance values estimated from satellite sensor data. Reflectance spectra were produced by the model using detailed field data as input, i.e. forest LAI, tree structural parameters, and the composition, distribution and reflectance of the forest floor. Common deciduous and coniferous forest types in southern Sweden were investigated. A negative linear relationship (r² = 0.6) was observed between field estimated LAI and the degree of understory vegetation, and the results indicated better agreement when coniferous and deciduous stands were analysed separately. The simulated spectra verified that the impact of the understory on the reflected signal from the top of the canopy is important; the reflectance values varying by up to ± 18% in the red and up to ± 10% in the near infra-red region of the spectra due to the understory. In order to predict the variation in LAI due to the understory vegetation, model inversions were performed where the input spectra were changed between the minimum, average and maximum reflectance values obtained from the forward runs. The resulting variation in LAI was found to be 1.6 units on average. The LAI of the understory could be predicted indirectly from simple stand data on forest characteristics, i.e. from allometric estimates, as an initial step in the process of estimating LAI. It is suggested here that compensation for the effect of the understory would improve the accuracy in the estimates of canopy LAI considerably.     

Estimating attributes of deciduous forest cover of a sanctuary in India utilizing Hyperion data and PLS analysis

Continuous and comprehensive evaluation of biochemical and biophysical attributes of forest ecosystems is a key aspect for monitoring their health status in the current global change scenario. Traditional methods of monitoring forest cover such as inventorying are time consuming, cost intensive, and untimely in delivering the output. The present study was carried out to monitor three important deciduous forest covers of India (teak, bamboo, and mixed), utilizing Hyperion (EO1) data of two seasons and partial least squares regression analysis. Attributes measured were canopy chlorophyll, nitrogen, cellulose, lignin, and biomass of tree trunks. Measured attributes showed a wider range, indicating variation in the growth phase of the covers. PLS models developed in this study showed higher R² values (0.63–0.90 for chlorophyll and nitrogen, 0.52–0.80 for cellulose and lignin, 0.80–0.86 for bole biomass). From the spectral data analysis we conclude that PLS regression with selected bands is better for the computation of specific biochemical parameters. For parameters such as bole biomass, reflectance spectra of 165 bands worked better. Developed models are advantageous for monitoring two important tropical covers (teak and bamboo) by utilizing space-borne data. A PLS model developed for teak-cover biomass worked well with mixed species cover (tested as an independent data set), indicating the applicability of the model across similar tropical covers.     

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.     

Leaf area index estimation with MODIS reflectance time series and model inversion during full rotations of Eucalyptus plantations

The leaf area index (LAI) of fast-growing Eucalyptus plantations is highly dynamic both seasonally and inter-annually, and is spatially variable depending on pedo-climatic conditions. LAI is very important in determining the carbon and water balance of a stand, but is difficult to measure during a complete stand rotation and at large scales. Remote-sensing methods allowing the retrieval of LAI time series with accuracy and precision are therefore necessary. Here, we tested two methods for LAI estimation from MODIS 250m resolution red and near-infrared (NIR) reflectance time series. The first method involved the inversion of a coupled model of leaf reflectance and transmittance (PROSPECT4), soil reflectance (SOILSPECT) and canopy radiative transfer (4SAIL2). Model parameters other than the LAI were either fixed to measured constant values, or allowed to vary seasonally and/or with stand age according to trends observed in field measurements. The LAI was assumed to vary throughout the rotation following a series of alternately increasing and decreasing sigmoid curves. The parameters of each sigmoid curve that allowed the best fit of simulated canopy reflectance to MODIS red and NIR reflectance data were obtained by minimization techniques. The second method was based on a linear relationship between the LAI and values of the GEneralized Soil Adjusted Vegetation Index (GESAVI), which was calibrated using destructive LAI measurements made at two seasons, on Eucalyptus stands of different ages and productivity levels. The ability of each approach to reproduce field-measured LAI values was assessed, and uncertainty on results and parameter sensitivities were examined. Both methods offered a good fit between measured and estimated LAI (R² = 0.80 and R² = 0.62 for model inversion and GESAVI-based methods, respectively), but the GESAVI-based method overestimated the LAI at young ages.     

Estimating the leaf area index in Indian tropical forests using Landsat-8 OLI data

Leaf area index (LAI) is a key vegetation biophysical parameter and is extensively used in modelling of phenology, primary production, light interception, evapotranspiration, carbon, and nitrogen dynamics. In the present study, we attempt to spatially characterize LAI for natural forests of Western Ghats India, using ground based and Landsat-8 Operational Land Imager (OLI) sensor satellite data. For this, 41 ground-based LAI measurements were carried out across a gradient of tropical forest types, viz. dry, moist, and evergreen forests using LAI-2200 plant canopy analyser, during the month of March 2015. Initially, measured LAI values were regressed with 15 spectral variables, including nine spectral vegetation indices (SVIs) and six Landsat-8 surface reflectance (ρ) variables using univariate correlation analysis. Results showed that the red (ρ_red), near-infrared (ρ_NIR), shortwave infrared (ρ_SWIR1, ρ_SWIR2) reflectance bands (R² > 0.6), and all SVIs (R² > 0.7) except simple ratio (SR) have the highest and second highest coefficient of determination with ground-measured LAI. In the second step, to select significant (high R², low root mean square error (RMSE), and p-level < 0.05) SVIs to determine the best representative model, stepwise multiple linear regression (SMLR) was implemented. The results indicate that the SMLR model predicted LAI with better coefficient of determination (R² = 0.83, RMSE = 0.78) using normalized difference vegetation index, enhanced vegetation index, and soil-adjusted vegetation index variables compared to the univariate approach. The predicted SMLR model was used to estimate a spatial map of LAI. It is desirable to evaluate the stability and potentiality of regional LAI models in natural forest ecosystems against the operationally accepted Moderate Resolution Imaging Spectroradiometer (MODIS) global LAI product. To do this, the Landsat-8 pixel-based LAI map was resampled to 1 km resolution and compared with the MODIS derived LAI map. Results suggested that Landsat-8 OLI-based VIs provide significant LAI maps at moderate resolution (30 m) as well as coarse resolution (1 km) for regional climate models.     

Global sensitivity analysis of PROSAIL model parameters when simulating Moso bamboo forest canopy reflectance

PROSAIL is a combination of the leaf optical properties spectra (PROSPECT) model and the scattering by arbitrarily inclined leaves (SAIL) canopy bidirectional reflectance model. When modelling forest canopy reflectance using the PROSAIL radiative transfer model, the sensitivities of parameters can affect the modelling accuracy. Traditionally, sensitivities have been assessed using local sensitivity analysis (LSA); however, drawbacks to this approach include a lack of consideration for coupled effects between different parameters. In this study, parameter sensitivities in the PROSAIL model were calculated using two global sensitivity analysis (GSA) methods (the Extended Fourier Amplitude Sensitivity Test (EFAST) method and the Morris method), field measurements, and Landsat 5 Thematic Mapper (TM) data for a Moso bamboo forest. The results of GSA were compared with those of LSA in order to identify the key parameters impacting the Moso bamboo forest canopy reflectance, and to provide a reference for model optimization and vegetation canopy inversion improvement. The results showed that: (1) the sensitivities of six major input parameters of the PROSAIL model were generally consistent with the sorting orders of the two GSA methods, but were not in accordance with those from the LSA method, especially in the mid-infrared band; (2) coupled effects among parameters acting on reflectance simulation in visible light bands were greater than those in infrared bands; (3) the simulated canopy reflectance was evaluated using Landsat 5 TM data, and the results simulated based on LSA analysis showed higher error than those based on GSA analysis, because the LSA method ignored the influence of some parameters on canopy reflectance, e.g. leaf mesophyll structure (N), average leaf angle (ALA), leaf water content (C_w), and leaf dry matter content (C_m). However, GSA was able to fully consider the coupled effects among parameters, and thus identified the sensitive parameters impacting on reflectance more accurately.     

Seasonal reflectance trends of hemiboreal birch forests

MODIS, AVHRR and SPOT VEGETATION satellite images have recently been used to track coarse scale seasonal vegetation dynamics of boreal and temperate forests. However, the understanding of driving factors of reflectance seasonality at forest stand level is still in its infancy, and has only preliminarily been linked to, for example, forest structure or site fertility. We present results from a study on the seasonal reflectance trends of 145 hemiboreal birch stands in Estonia from budburst to initial senescence. A time series comprising 32 high resolution Landsat ETM+, TM and SPOT HRVIR, HRV images from April to September was assembled for analyzing empirical reflectance courses of birch stands. The most noteworthy seasonal reflectance dynamics were observed in the red and NIR channels, changes in the green and SWIR spectral channels were relatively small. The most stable period in stand reflectance in all the spectral channels occurred in midsummer i.e. when stand leaf area index (LAI) reached its highest level and changes in solar angle were the smallest. A twenty-day difference was observed between the reflectance development of birch stands growing on infertile and fertile sites. Next, to provide an explanation for the observed reflectance changes, we simulated the mean seasonal reflectance trajectories of the study stands at 10 day intervals for the same period using a radiative transfer model (FRT). Simulated seasonal reflectance courses for the different site fertility classes followed the general pattern of the measured courses. Simulation results indicated that the main driving factors for reflectance seasonality for all the site fertility classes in the red and green bands were stand LAI and leaf chlorophyll content, in the NIR band stand LAI, and in the SWIR band LAI and general water content. Finally, we discuss current limitations related to applying forest radiative transfer models in investigating the driving factors of seasonal reflectance changes in the boreal zone.     

Effects of woody elements on simulated canopy reflectance: Implications for forest chlorophyll content retrieval

An important bio-indicator of actual plant health status, the foliar content of chlorophyll a and b (Cab), can be estimated using imaging spectroscopy. For forest canopies, however, the relationship between the spectral response and leaf chemistry is confounded by factors such as background (e.g. understory), canopy structure, and the presence of non-photosynthetic vegetation (NPV, e.g. woody elements)—particularly the appreciable amounts of standing and fallen dead wood found in older forests. We present a sensitivity analysis for the estimation of chlorophyll content in woody coniferous canopies using radiative transfer modeling, and use the modeled top-of-canopy reflectance data to analyze the contribution of woody elements, leaf area index (LAI), and crown cover (CC) to the retrieval of foliar Cab content. The radiative transfer model used comprises two linked submodels: one at leaf level (PROSPECT) and one at canopy level (FLIGHT). This generated bidirectional reflectance data according to the band settings of the Compact High Resolution Imaging Spectrometer (CHRIS) from which chlorophyll indices were calculated. Most of the chlorophyll indices outperformed single wavelengths in predicting Cab content at canopy level, with best results obtained by the Maccioni index ([R₇₈₀ − R₇₁₀] / [R₇₈₀ − R₆₈₀]). We demonstrate the performance of this index with respect to structural information on three distinct coniferous forest types (young, early mature and old-growth stands). The modeling results suggest that the spectral variation due to variation in canopy chlorophyll content is best captured for stands with medium dense canopies. However, the strength of the up-scaled Cab signal weakens with increasing crown NPV scattering elements, especially when crown cover exceeds 30%. LAI exerts the least perturbations. We conclude that the spectral influence of woody elements is an important variable that should be considered in radiative transfer approaches when retrieving foliar pigment estimates in heterogeneous stands, particularly if the stands are partly defoliated or long-lived.     

Comparison of different regression models and validation techniques for the assessment of wheat leaf area index from hyperspectral data

Leaf area index (LAI) is one of the most important plant parameters when observing agricultural crops and a decisive factor for yield estimates. Remote-sensing data provide spectral information on large areas and allow for a detailed quantitative assessment of LAI and other plant parameters. The present study compared support vector regression (SVR), random forest regression (RFR), and partial least-squares regression (PLSR) and their achieved model qualities for the assessment of LAI from wheat reflectance spectra. In this context, the validation technique used for verifying the accuracy of an empirical–statistical regression model was very important in order to allow the spatial transferability of models to unknown data. Thus, two different validation methods, leave-one-out cross-validation (cv) and independent validation (iv), were performed to determine model accuracy. The LAI and field reflectance spectra of 124 plots were collected from four fields during two stages of plant development in 2011 and 2012. In the case of cross-validation for the separate years, as well as the entire data set, SVR provided the best results (2011: R²_cv = 0.739, 2012: R²_cv = 0.85, 2011 and 2012: R²_cv = 0.944). Independent validation of the data set from both years led to completely different results. The accuracy of PLSR (R²_iv = 0.912) and RFR (R²_iv = 0.770) remained almost at the same level as that of cross-validation, while SVR showed a clear decline in model performance (R²_iv = 0.769). The results indicate that regression model robustness largely depends on the applied validation approach and the data range of the LAI used for model building.     

Discrete return lidar-based prediction of leaf area index in two conifer forests

Leaf area index (LAI) is a key forest structural characteristic that serves as a primary control for exchanges of mass and energy within a vegetated ecosystem. Most previous attempts to estimate LAI from remotely sensed data have relied on empirical relationships between field-measured observations and various spectral vegetation indices (SVIs) derived from optical imagery or the inversion of canopy radiative transfer models. However, as biomass within an ecosystem increases, accurate LAI estimates are difficult to quantify. Here we use lidar data in conjunction with SPOT5-derived spectral vegetation indices (SVIs) to examine the extent to which integration of both lidar and spectral datasets can estimate specific LAI quantities over a broad range of conifer forest stands in the northern Rocky Mountains. Our results show that SPOT5-derived SVIs performed poorly across our study areas, explaining less than 50% of variation in observed LAI, while lidar-only models account for a significant amount of variation across the two study areas located in northern Idaho; the St. Joe Woodlands (R² = 0.86; RMSE = 0.76) and the Nez Perce Reservation (R² = 0.69; RMSE = 0.61). Further, we found that LAI models derived from lidar metrics were only incrementally improved with the inclusion of SPOT 5-derived SVIs; increases in R² ranged from 0.02–0.04, though model RMSE values decreased for most models (0–11.76% decrease). Significant lidar-only models tended to utilize a common set of predictor variables such as canopy percentile heights and percentile height differences, percent canopy cover metrics, and covariates that described lidar height distributional parameters. All integrated lidar-SPOT 5 models included textural measures of the visible wavelengths (e.g. green and red reflectance). Due to the limited amount of LAI model improvement when adding SPOT 5 metrics to lidar data, we conclude that lidar data alone can provide superior estimates of LAI for our study areas.     

Assessment of water quality in Lake Garda (Italy) using Hyperion

For testing the integration of the remote sensing related technologies into the water quality monitoring programs of Lake Garda (the largest Italian lake), the spatial and spectral resolutions of Hyperion and the capability of physics-based approaches were considered highly suitable. Hyperion data were acquired on 22nd July 2003 and water quality was assessed (i) defining a bio-optical model, (ii) converting the Hyperion at-sensor radiances into subsurface irradiance reflectances, and (iii) adopting a bio-optical model inversion technique. The bio-optical model was parameterised using specific inherent optical properties of the lake and light field variables derived from a radiative transfer numerical model. A MODTRAN-based atmospheric correction code, complemented with an air/water interface correction was used to convert Hyperion at-sensor radiances into subsurface irradiance reflectance values. These reflectance values were comparable to in situ reflectance spectra measured during the Hyperion overpass, except at longer wavelengths (beyond 700 nm), where reflectance values were contaminated by severe atmospheric adjacency effects. Chlorophyll-a and tripton concentrations were retrieved by inverting two Hyperion bands selected using a sensitivity analysis applied to the bio-optical model. The sensitivity analysis indicated that the assessment of coloured dissolved organic matter was not achievable in this study due to the limited coloured dissolved organic matter concentration range of the lake, resulting in reflectance differences below the environmental measurement noise of Hyperion. The chlorophyll-a and tripton image-products were compared to in situ data collected during the Hyperion overpass, both by traditional sampling techniques (8 points) and by continuous flow-through systems (32 km). For chlorophyll-a the correlation coefficient between in situ point stations and Hyperion-inferred concentrations was 0.77 (data range from 1.30 to 2.16 mg m^− 3). The Hyperion-derived chlorophyll-a concentrations also match most of the flow-through transect data. For tripton, the validation was constrained by variable re-suspension phenomena. The correlation coefficient between in situ point stations and Hyperion-derived concentrations increased from 0.48 to 0.75 (data range from 0.95 to 2.13 g m^− 3) if the sampling data from the re-suspension zone was avoided. The comparison of Hyperion-derived tripton concentrations and flow-through transect data exhibited a similar mismatch. The results of this research suggest further studies to address compatibilities of validation methods for water body features with a high rate of change, and to reduce the contamination by atmospheric adjacency effects on Hyperion data at longer wavelengths in Alpine environment. The transferability of the presented method to other sensors and the ability to assess water quality independent from in situ water quality data, suggest that management relevant applications for Lake Garda (and other subalpine lakes) could be supported by remote sensing.     

An evaluation of EO-1 hyperspectral Hyperion data for chlorophyll content and leaf area index estimation

Estimation of chlorophyll content and the leaf area index (LAI) using remote sensing technology is of particular use in precision agriculture. Wavelengths at the red edge of the vegetation spectrum (705 and 750 nm) were selected to test vegetation indices (VIs) using spaceborne hyperspectral Hyperion data for the estimation of chlorophyll content and LAI in different canopy structures. Thirty sites were selected for the ground data collection. The results show that chlorophyll content and LAI can be successfully estimated by VIs derived from Hyperion data with a root mean square error (RMSE) of 7.20–10.49 μg cm^?2 for chlorophyll content and 0.55–0.77 m² m^?2 for LAI. The special index derived from three bands provided the best estimation of the chlorophyll content (RMSE of 7.19 μg cm^?2 for the Modified Chlorophyll Absorption Ratio Index/Optimized Soil-Adjusted Vegetation Index (MCARI/OSAVI₇₀₅)) and LAI (RMSE of 0.55 m² m^?2 for a second form of the MCARI (MCARI2₇₀₅)). These results demonstrate the possibilities for analysing the variation in chlorophyll content and LAI using hyperspectral Hyperion data with bands from the red edge of the vegetation spectrum.