Bunkei Matsushita

Sensitivity of the Enhanced Vegetation Index (EVI) and Normalized Difference Vegetation Index (NDVI) to Topographic Effects: A Case Study in High-density Cypress Forest

Vegetation indices play an important role in monitoring variations in vegetation. The Enhanced Vegetation Index (EVI) proposed by the MODIS Land Discipline Group and the Normalized Difference Vegetation Index (NDVI) are both global-based vegetation indices aimed at providing consistent spatial and temporal information regarding global vegetation. However, many environmental factors such as atmospheric conditions and soil background may produce errors in these indices. The topographic effect is another very important factor, especially when the indices are used in areas of rough terrain. In this paper, we theoretically analyzed differences in the topographic effect on the EVI and the NDVI based on a non-Lambertian model and two airborne-based images acquired from a mountainous area covered by high-density Japanese cypress plantation were used as a case study. The results indicate that the soil adjustment factor “L” in the EVI makes it more sensitive to topographic conditions than is the NDVI. Based on these results, we strongly recommend that the topographic effect should be removed in the reflectance data before the EVI was calculated—as well as from other vegetation indices that similarly include a term without a band ratio format (e.g., the PVI and SAVI)—when these indices are used in the area of rough terrain, where the topographic effect on the vegetation indices having only a band ratio format (e.g., the NDVI) can usually be ignored.

Integrating remotely sensed data with an ecosystem model to estimate net primary productivity in East Asia

Abstract This paper describes a method of integrating remotely sensed data with an ecosystem model to estimate net primary productivity (NPP) in East Asia. We improved the Boreal Ecosystem Productivity Simulator (BEPS) model for global NPP estimation by incorporating a new land cover map and employed a robust Normalized Difference Vegetation Index–Leaf Area Index (NDVI–LAI) algorithm. Using this method, we produced a map showing the distribution of annual NPP in East Asia in 1998 and calculated that the mean NPP for that area in that year was 634 g C/m 2 /year. Comparing the estimated NPP obtained from model computation with the observed NPP obtained from an NPP database, we found that the estimated NPP closely approximates the observed NPP, with an average error of −20%. We checked the accuracy of a six-biome land cover map using a Geographic Information System (GIS) data set for Japan [Data Sets for GIS on the Natural Environment, Japan (DS_GIS_NEJ), Japan Environment Agency, Ver. 2, 1999] and how the accuracy of the map affects NPP estimation. Results show that an accurate land cover map is essential if one is to accurately and reliably estimate NPP, and it is especially crucial if one is to estimate the NPP of an individual biome (e.g., for crop prediction).

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, 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.

An automatic method for burn scar mapping using support vector machines

Wildfires release large amounts of carbon, smoke and aerosols that strongly impact the global climatic system. Burn scar is an important parameter when modelling the impact of wildfires on the ecosystem and the climatic system. We have developed an automatic burn scar mapping method using daily Moderate Resolution Imaging Spectroradiometer (MODIS) data, in which the Global Environment Monitoring Index (GEMI), a vegetation index VI3T and a new index, GEMI‐Burn scar (GEMI‐B), were used together to enhance the differences between burned and unburned pixels related to vegetation photosynthesis, surface temperature and vegetation water content, respectively, and an automatic region growing method based on Support Vector Machines (SVMs) was used to classify burn scars without any predefined threshold. A case study was carried out to validate the new method at the border area between Mongolia and China, where a wildfire took place in May 2003. The results show that the burn scar area extracted by the new method is consistent with that from Landsat Thematic Mapper (TM) data with high accuracy. The sound performance of the new technique is due to the following reasons: (1) multiple features of burn scar spectra were combined and used, (2) a reasonable assumption was made stating that the neighbourhoods of active fires (hotspots) are most likely to be burn scars, (3) an SVM classifier was adopted that works well with a small number of training samples, and (4) an iterative classification procedure was developed that is capable of running continuous training for the SVM classifier to deal with the transitionary features of burn scar pixels. The results suggest that the new index GEMI‐B and automatic mapping method based on SVMs have the potential to be applied to near real‐time burn scar mapping in grassland areas.

How do dams affect freshwater fish distributions in Japan? Statistical analysis of native and nonnative species with various life histories

We examined the effects of dams on freshwater fish species based on data collected during 1990–2004 from 200 drainage systems in Japan. Of the 76 fish species examined, the occurrence of 20 species within Petromyzontidae, Cyprinidae, Cobitididae, Salmonidae, Cottidae, and Gobiidae was negatively affected by the presence of dams located in the downstream reaches of fish survey sites, whereas the occurrence of 12 species within Cyprinidae, Adrianichthyidae, Centrarchidae, and Gobiidae was positively associated with the presence of dams. A significantly higher proportion of the fishes with a negative damming effect were diadromous species as compared to the fishes with a positive damming effect. Conversely, the latter group had a significantly higher proportion of nonnative species than the former. A significant interaction existed between the effects of damming and the effects of elevation on family-specific species richness. Families dominated by native migratory species showed a greater reduction in the number of species above dams at lower elevations, whereas families represented primarily by nonnative species had higher species richness above dams at higher elevations, except for Centrarchidae, which was always higher in species richness above dams regardless of elevation. Based on our findings, dams in Japan have adversely affected native freshwater fishes by blocking their migration routes, favoring nonnative fishes, or altering existing habitats.

An Enhanced Three-Band Index for Estimating Chlorophyll-a in Turbid Case-II Waters: Case Studies of Lake Kasumigaura, Japan, and Lake Dianchi, China

A three-band index was previously proposed and successfully utilized to estimate the chlorophyll-a concentration (Chl-a) in case-II waters. However, this index shows uncertainties in highly turbid situations. In this study, an enhanced three-band index is proposed to solve this problem. Since the new index employs bands that are identical to those of the original threeband index, it can be applied to Medium Resolution Imaging Spectrometer (MERIS) data. The performance of the index was evaluated using the data collected from two turbid Asian lakes: Lake Kasumigaura, Japan, and Lake Dianchi, China. The results showed that the Chl-a predicted by the enhanced threeband index was strongly correlated with the measured Chl-a (R2 > 0.83), and the root-mean-square error (rmse) and the normalized root-mean-square error (NRMS) were both reduced for the two lakes (for Lake Kasumigaura, rmse from 13.97 to 8.68 mg · m-3 and NRMS from 19.01% to 12.30%; for Lake Dianchi, rmse from 41.29 to 15.28 mg · m-3 and NRMS from 35.83% to 21.34%). These findings imply that, if accurately atmospheric-corrected MERIS data are available, the enhanced three-band index could be used for mapping Chl-a even in highly turbid case-II waters.

Retrieval of Inherent Optical Properties for Turbid Inland Waters From Remote-Sensing Reflectance

Remote estimation of inherent optical properties (IOPs) for water bodies cannot only provide indicators of water quality, but also be used in the study on biological and biogeochemical processes of waters. The quasi-analytical algorithm (QAA) is a simple and effective method to retrieve IOPs from remote-sensing reflectance (Rrs). The QAA has been widely validated and applied in oceans, but its application in inland waters is far less extensive. In this paper, the QAA was enhanced to retrieve IOPs for turbid inland waters based on the bandwidths of Medium Resolution Imaging Spectrometer (MERIS). The enhancement was achieved by proposing a semi-analytical model to estimate the spectral slope of particle backscattering, as well as a novel estimation model for phytoplankton absorption coefficient at 443 nm. Two data sets (i.e., noise-free synthetic data and in-situ data) were collected to assess the performance of the enhanced algorithm. Results show that the algorithm yields almost error-free estimations for total absorption and backscattering coefficients and estimations for phytoplankton absorption at 443 nm with acceptable accuracy in the case of synthetic data set. For the in-situ data set, the algorithm retrieves the total absorption coefficients (ranging 0.337-8.331 m-1) with root-mean-square-error in log scale (RMSE) and bias in log scale lower than 0.130 and 0.094, respectively, and phytoplankton absorption at 443 nm (ranging 0.378-4.669 m-1) with RMSE and bias in log scale of 0.151 and 0.096, respectively. These results indicate the potential of the enhanced QAA to accurately retrieve the IOPs from MERIS satellite observations for inland waters.

Developing a MODIS-based index to discriminate dead fuel from photosynthetic vegetation and soil background in the Asian steppe area

Dead fuel (DF) coverage and biomass are important parameters for wildfire danger rating and fire behaviour modelling. Although a hyperspectral Cellulose Absorption Index (CAI) has been proven to be a good tool for discriminating non-photosynthetic vegetation (NPV), plant litter and crop residues, the narrow bands are not well suited for multi-band Moderate Resolution Imaging Spectroradiometer (MODIS) to achieve global and real-time wildfire risk assessment. Meanwhile, Landsat TM-based indices, such as Normalized Difference Index (NDI), Soil Adjusted Corn Residue Index (SACRI) and Crop Residue Index Multiband (CRIM), have been proposed to extract NPV only in the case of two components of NPV and soil. This study intended to discriminate DF and estimate its coverage in three-component mixtures of photosynthetic vegetation (PV), DF and soil using MODIS simulated data. The methods used included: (1) analysing field spectra of PV, DF and soil; (2) developing a four-band Dead Fuel Index (DFI) based on MODIS band ranges; and (3) simulating spectral mixtures to determine the lower thresholds for DF detection. DFI as well as NDI, SACRI, CRIM and CAI were evaluated based on individual spectra and linearly simulated spectral mixtures of PV, DF and soil. Results suggested that DFI was the best index for DF discrimination, with a minimal fractional coverage of only about 0.20 in one pixel required to confirm the existence of DF with MODIS simulated data. However, DFI is less sensitive to long-term DFs. For top-of-atmosphere (TOA) reflectance simulation under different atmospheric conditions, the minimal fractions were 0.32, 0.42, 0.53 and 0.64 for optical depths at 550 nm of 0.2 (clear), 0.4, 0.6 and 0.8 (hazy), respectively. The results of this study suggest that DFI has good potential to estimate DF coverage in steppe areas.

A new algorithm for estimating chlorophyll-a concentration from multi-spectral satellite data in case II waters: a simulation based on a controlled laboratory experiment

This paper presents the spectral decomposition algorithm (SDA), a new algorithm for estimating chlorophyll‐a concentration in case II waters using multi‐spectral satellite data, which is based on a simulation in a controlled laboratory experiment. The SDA is composed of two key steps. The first of these is to consider the mixed reflectance spectrum of a given pixel as a linear combination of three basic components: clear water, non phytoplankton suspended sediments (NPSS), and phytoplankton. The second step is to use a decomposition coefficient (Cp ) obtained from the first step as an independent variable in the chlorophyll‐a estimation model, instead of the single band reflectance, band ratio or arithmetic calculation of bands used in conventional methods. The simulated results for the Landsat TM data showed that bands 1, 3 and 4 are useful wavelengths for estimating chlorophyll‐a concentrations. In the case of a water body with chlorophyll‐a concentrations ranging from 0 to 105 µg l−1 and NPSS concentrations ranging from 0 to 100 mg l−1, the RMSE of the estimation model of chlorophyll‐a concentrations based on the SDA was 13.7 µg l−1, reduced by nearly half of that for conventional methods (the RMSE was 25.6 µg l−1 for the band ratio, and 25.5 µg l−1 for the arithmetic calculation of bands). The results of a two‐factor ANOVA (without replication) highlight that the decomposition coefficient Cp contains information from phytoplankton far more than from NPSS. However, Cp values were still changed with the addition of NPSS, due mainly to the influence of the interaction of optical properties among phytoplankton, NPSS and water, which occurred in both horizontal and vertical directions in the water bodies. Considering the basic components as a nonlinear combination in a water area may reduce the effect of NPSS on Cp values from that of their linear combination. In addition, the influence of coloured dissolved organic matter (CDOM), which is generally considered as one of the optically active substances in case II waters, was ignored according to the practical conditions of our study area, Lake Kasumigaura, Japan in this paper. Users can set the basic components (or endmembers) freely according to the conditions of their own study area when the SDA is used (e.g. considering the CDOM as a basic component in the SDA).

A New Method to Define the VI-Ts Diagram Using Subpixel Vegetation and Soil Information: A Case Study over a Semiarid Agricultural Region in the North China Plain.

The VI-Ts diagram determined by the scatter points of the vegetation index (VI) and surface temperature (Ts) has been widely applied in land surface studies. In the VI-Ts diagram, dry point is defined as a pixel with maximum Ts and minimum VI, while wet point is defined as a pixel with minimum Ts and maximum VI. If both dry and wet points can be obtained simultaneously, a triangular VI-Ts diagram can be readily defined. However, traditional methods cannot define an ideal VI-Ts diagram if there are no full ranges of land surface moisture and VI, such as during rainy season or in a period with a narrow VI range. In this study, a new method was proposed to define the VI-Ts diagram based on the subpixel vegetation and soil information, which was independent of the full ranges of land surface moisture and VI. In this method, a simple approach was firstly proposed to decompose Ts of a given pixel into two components, the surface temperatures of soil (Tsoil) and vegetation (Tveg), by means of Ts and VI information of neighboring pixels. The minimum Tveg and maximum Tsoil were then used to determine the wet and dry points respectively within a given sampling window. This method was tested over a 30 km × 30 km semiarid agricultural area in the North China Plain through 2003 using Advanced Spaceborne Thermal Emission Reflection Radiometer (ASTER) and MODerate-resolution Imaging Spectroradiometer (MODIS) data. The wet and dry points obtained from our proposed method and from a traditional method were compared with those obtained from ground data within the sampling window with the 30 km × 30 km size. Results show that Tsoil and Tveg can be obtained with acceptable accuracies, and that our proposed method can define reasonable VI-Ts diagrams over a semiarid agricultural region throughout the whole year, even for both cases of rainy season and narrow range of VI.