Crop classification plays an important role in effective and controllable agricultural management. In this paper, nine scenes of Landsat8_OLI data in 2013 were collected for cotton planting area extraction in Shawan country of Xinjiang Uygur Autonomous Region, China. Normalized Difference Vegetation Index (NDVI) time series were generated to characterize the phenological pattern of each crop type. The optimal temporal reflectance image was chosen by analyzing the difference of NDVI profile between cotton and other crop types. The hierarchical classification strategy was performed on the three features of NDVI time series, NDVI statistics and reflectance. Firstly a simple decision tree was built on NDVI statistics and reflectance to extract vegetation cover; then various types of crops were distinguished by support vector machine (SVM) and maximum likelihood supervised classifier (MLC), thereby cotton plating area was extracted. A comprehensive evaluation for the cotton extraction map was performed using human-computer interaction visual interpretation and ground truth data. Results showed that MLC achieved the accuracy of 97.56% for cotton extraction. The cotton extraction map was very consistent with the ground truth data. This multi-temporal classification method is promising for crop extraction even for land cover classification.

The light-use efficiency (LUE) is one of critical parameters in the terrestrial ecosystem production studies. However, it is still a challenge how to up-scale LUE from canopy to the landscape/regional scales. One potential solution is to use automated multi-angle tower-based remote sensing platforms, which observe canopy reflectance with high spatial, temporal, spectral and angle resolution. Although some published paper on the LUE in boreal and temperate forests had used continuous multi-angle measurements of the surface reflectance, lack of study in literature investigated the vegetation physiological parameters of cropland using the surface reflectance with high spatio-temporal and high spectral data with multiple angles. To improve our understanding of physiological status of cropland, the maize within the footprint of the Daman Superstation flux tower site of Heihe Watershed Allied Telemetry Experiment Research (HiWATER) was employed in this study. Based on the observed reflectance and flux data, a Bidirectional Reflectance Distribution Function (BRDF) of vegetation index (Photochemical Reflectance Index, PRI and Vegetation Index using the Universal Pattern Decomposition method, VIUPD) at continuous time series was established by integrating of a semi-empirical kernel-driven BRDF model (RossThick-LiSparse), a footprint model (the Simple Analytical Footprint model on Eulerian coordinates for scalar Flux, SAFE-f) and a LUE model. Besides, based on the sky-condition (direct/diffused radiation) data, the relationships between the vegetation index (PRI and VIUPD) and sunlit/shaded LUE under corresponding sky conditions were established. Taking maize field as an example, the measurements were obtained during June to August, 2012. The relationships between PRI and LUE for sunlit and shaded leaves were: PRIsu=0.06339log(LUEsu) + 0.04882, PRIsh= 0.02675log(LUEsh) + 0.01619, where, the subscript su, sh represent sunlit and shaded leaves respectively; p< 0.0001, R2 (Coefficient of determination) equal 0.6443 and 0.6081 for sunlit and shaded, respectively. Then the LUE was up-scaled to landscape/regional scales based on these relationships and sky conditions, and it can be used for the estimation of gross primary productivity (GPP) of cropland using a LUE-based model with high accuracy.

Due to the budget and technical limitations, remote sensing sensors trade spatial resolution, swath width and spectral resolution. Consequently, no sensor can provide high spatial resolution, high temporal resolution and high spectral resolution simultaneously. However, the ability of earth observation at fine resolution is urgently needed for global change science. One possible solution is to “blend” the reflectance from high spatial resolution with less frequency coverage (e.g. Landsat), daily, global data (e.g. MODIS, Moderate Resolution Imaging Spectroradiometer), and high spectral resolution with low re-visit cycle (e.g. Hyperion). However, the previous algorithms for blending multi-source remotely-sensed data have some shortcomings, especially for less consideration of hyperspectral information. To overcome this shortcoming, this study has developed a SPAtial-Temporal-Spectral blending model (SPATS) which can simulate surface reflectance with high spatial-temporal-spectral resolution. SPATS is based on existing spatial-temporal image fusion model and spatialspectral image fusion model. Taking Landsat, Hyperion and MODIS data as example, the performance of SPATS was tested with both simulated and observed satellite data, especially at heterogeneous landscapes. The results show that the high spatial-temporal-spectral resolution reflectance data can be applied to a new investigation into how global landscapes are changing at different temporal scales.

The Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) on the Mars Reconnaissance Orbiter (MRO), which has high spectral and spatial resolution, has greatly enhanced our understanding of the composition of hydrous minerals on Mars. Researchers have successfully identified and mapped the hydrous minerals using the spectral features of water absorption based on CRISM data. But more quantitative information will yield strong constraints on the formation conditions of the hydrous minerals and even on ancient habitable environments on Mars[1]. However, there exist few studies on quantitative retrieval of hydrous minerals mainly due to the following problems: 1) the nonlinear model usually complicated, which limits its scope of application; 2) the mixture of minerals at visible-infrared wavelengths is nonlinear, and large errors will occur if the linear model is used directly; 3) The distribution characteristics (relatively low concentration, scattered distribution, and unclear or unknown background minerals) of hydrous minerals result in difficulty in determining the spectra and the number of endmembers from the image itself, which is usually needed for unmixing.

The detection of hydrous minerals on Mars is of great importance for revealing the early water environment as well as possible biotic activity. However, few studies focus on quantitatively retrieving hydrous minerals for some difficulties. In this letter, we studied the area around the Mars Science Laboratory (MSL) landing site, to identify hydrous minerals and retrieve their abundance. Firstly, the distribution of hydrous minerals was extracted using their water absorption features. Then, a sparse unmixing algorithm was applied along with the CRISM spectral library to retrieve the abundance of hydrous minerals in this area. As a result, seven hydrous minerals were quantitatively retrieved, e.g. actinolite, montmorillonite, saponite, jarosite and so forth, and the total concentration of all hydrous minerals was as high as 40 vol% near the lower reaches of Mount Sharp. Our results were consistent with results from related research and the in-situ analysis of the MSL rover Curiosity.

This study proposed a multidimensional remote sensing data storage structure related to spatial, temporal and spectral dimensions, the integrated organization of this multidimensional remote sensing data for long time series, and the multidimensional analysis from spatial, temporal, spectral dimensions. Five data formats were included within the multidimensional data storage, which were TSQ (Temporal Sequential), TSP (Temporal Sequential Pixel), TIB (Temporal Interleaved by Band), TIP (Temporal Interleaved by Pixel) and TIS (Temporal Interleaved by Spectrum). Quick extraction for spectral, temporal, band cube, temporal cube, and feature cube can be accomplished. Meanwhile, one data format can be transformed to another data format based on five multidimensional data formats. First, a set of basic transformation was proposed to achieve the relationship between part data format. Second, indirect transformation can be achieved according to the set of basic transformation. For the demand of multidimensional conversion and extraction of four-dimensional remote sensing data, the proposed four dimensional remote sensing data structure of this study can achieve the integration organization and management of multidimensional remote sensing data, which could support the application of multidimensional analysis.

Landsat satellites series provide large amounts of data for both the regional and global vegetation time series observation. As the currently operational Landsat satellites, Landsat 7 and Landsat 8 take the responsibility of multi-decadal Landsat imagery. However, the updating or changing of Landsat sensors will bring in a certain degree of bias for the long-term continuity research. To minimize this bias between ETM+ on board Landsat 7 and OLI on board Landsat 8, we intended to obtain the relationship between these two sensors for further research based on long-term Landsat data. Firstly, the wheat reflectance spectra from the field experiments was used to simulate for ETM+ and OLI satellite-based values. Then the linear regression relationship between two sensors was established via three vegetation indices which were the Normalized Difference Vegetation Index (NDVI), Soil-adjusted Vegetation Index (SAVI) and Enhanced Vegetation Index (EVI). Lastly, three study areas (A: Harvard Forest, B: Hulun Buir Grassland and C: Kahurangi National Park in New Zealand) were selected to verify the performance of the regression relationships by using images of ETM+ and OLI for the closed available time period. The results showed that: (1) the values of OLI was 1%-3% higher than ETM+ when only the instruments setting difference was been taken into consideration. (2) the continuity between the two Landsat sensors improved averaging 3% after the calibration by the regression relationship. Thus, the results based on regression relationship of this records offer the potential of effective calibration for the Landsat time series research related to ETM+ and OLI sensors.

Hyperspectral imaging can provide both continuously spatial and spectral information of the Earth\'s surface that allows us mapping of the regional geological and mineral resources information. One of the most successful applications of hyperspectral imaging remote sensing identified was regional geological and mineral resource surveys. A light-weight airborne hyperspectral imaging system (LAHIS) has been developed in China. The hardware of the compact LAHIS include hyperspectral VNIR imaging sensors, a hyperspectral SWIR imaging sensors, high resolution optical remote sensor and a POS (IMU and DGPS). The weight of the system is less than 23kg. The VNIR hyperspectral imaging sensors measures incoming radiation in 250 contiguous spectral channels in the 400–1000 nm wavelength range with spectral resolution of better than 2-3 nm and creates images of 334 pixels for a line of targets with a nominal instantaneous field of view (IFOV) of ~1 mrad. The SWIR hyperspectral imaging sensors measures incoming radiation in 256 contiguous spectral channels in the 1000–2500 nm wavelength range with spectral resolution of better than 6 nm and creates images of 320 pixels for a line of targets with a nominal instantaneous field of view (IFOV) of ~2 mrad. The 400 to 2500nm spectral range provides abundant information about the regional geological and mineral resources information. Two ground mineral scan experiment and an UAV carried flying experiment has been done. The experiment results show the LAHIS have achieved relative high performance levels in terms of signal to noise ratio and image quality. The potential applications for light-weight airborne hyperspectral imaging system in the regional geological and mineral resources survey are tremendous.

Lots of researches have been increasingly focusing on time series analysis of remote sensing datasets, deriving phenology time and trajectory parameters by carve fitting and detecting changes due to natural or artificial factors. For these applications extraction of various characteristic parameters is an indispensable and fundamental procedure. However, there is a lack of an integrated method currently to manage long time-series remote sensing imagery, meanwhile as a direct access to extracting time series of various characteristic parameters. In this paper we propose a user-friendly program for managing time series of remote sensing datasets, what’s more, extracting time series data for areas like point, rectangle and general polygon, according to user-defined formula automatically computing and constructing time series of various characteristic parameters. In addition, spectral data for one day, after a point or scope is selected, is able to be extracted and processed using general spectral analysis methods. This program tries to manage four dimensional (including time, spatial and spectral dimensions) remote sensing datasets, and be applicable to outputting time series of characteristic parameters and spectral data, providing an innovatively fast and flexible tool for time-series studies.

The telluric O2-B and O2-A have been proved to be capable of solar-induced vegetation fluorescence (SIF) retrieval based on Fraunhofer line depth (FLD) principle. However, most FLD based algorithms mainly aim for SIF detection in O2-A, not suitable in O2-B. One of the critical reasons is that it is very difficult to model the sudden varying reflectance around O2-B band located in the red-edge spectral region (about 680-800 nm). In order to resolve this issue, this study proposes a new method based on the established inverted Gaussian reflectance model (IGM) and FLD principle using hyperspectral radiative transfer simulations with 1 nm bandwidth in 400-1000 nm range. Results show that the proposed method can better capture the spectrally non-linear characterization of the reflectance in 680-800 nm and thereby enables retrieval in O2-B, yielding much more accurate SIFs than typical FLD methods, including sFLD, 3FLD and iFLD.

Remotely sensed solar-induced chlorophyll fluorescence (SIF) has been considered an ideal probe in monitoring global vegetation photosynthesis. However, challenges in accurate estimate of faint SIF (less than 5% of the total reflected radiation in near infrared bands) from the observed apparent reflected radiation greatly limit its wide applications. Currently, the telluric O2-B (~688nm) and O2-A (~761nm) have been proved to be capable of SIF retrieval based on Fraunhofer line depth (FLD) principle. They may still work well even using conventional ground-based commercial spectrometers with typical spectral resolutions of 2~5 nm and high enough signal-to-noise ratio (e.g., the ASD spectrometer). Nevertheless, almost all current FLD based algorithms were mainly developed for O2-A, a few concentrating on the other SIF emission peak in O2-B. One of the critical reasons is that it is very difficult to model the sudden varying reflectance around O2-B band located in the red-edge spectral region (about 680-800 nm). Diurnal