Qingxi Tong

Evaluation of methods for soil surface moisture estimation from reflectance data

The estimation of soil moisture from reflectance measurements in the solar spectral domain (400-2500 nm) was investigated. For this purpose, 18 soils representing a large range of permanent characteristics was considered. Reflectance data were measured in the laboratory during the soil drying process with a high spectral resolution spectroradiometer. Five approaches were compared. The first one was based on single-band reflectance and on the normalization of reflectance data by the reflectance of the corresponding soil under dry conditions. The second and the third approaches were based on either reflectance derivatives or absorbance derivatives. The fourth and fifth approaches were based on the differences of reflectance and absorbance observed in two non-consecutive bands. In the first step, the relationships were calibrated over half the dataset (nine soils) with emphasis on the selection of the most pertinent spectral bands. Results showed that, for the first approach, the bands corresponding to the highest water absorption capacities (1944 nm) yielded the best soil moisture retrieval performances. For the second and third approaches, the bands corresponding to sharp edges of the water absorption features performed better (1834 nm for the reflectance derivatives and 1622 nm for the absorbance derivatives). The fourth and fifth approaches that can be considered as a generalization of the derivative approach when bands are no longer consecutive, the best performances were achieved when the bands were not separated too much. The best overall retrieval performances were achieved with the absorbance derivatives and the difference of absorbance, confirming the non-linear character of the relationship between soil moisture and reflectance. The previously calibrated relations were tested over the evaluation dataset made of the nine remaining soils. It showed additionally that the normalization of reflectance values by that observed under dry conditions was only partly minimizing soil type effects. The best performances for the lowest soil moisture values (<0.20 m 3 /m 3 ) were obtained with the reflectance derivatives. However, because of the non-linear behaviour for the highest soil moisture values, the derivatives of absorbance and difference of absorbance provided the best estimates for these moisture levels. Results were discussed in view of potential applications of high spectral resolution sensors.

Estimating winter wheat plant water content using red edge parameters

Remote sensing of plant water content is difficult because the absorption band sensitive to foliar liquid water is also sensitive to the atmospheric vapour. A method using non-water-absorption spectral parameters to evaluate plant water content (PWC) would be valuable. In our experiment, canopy spectra of 48 winter wheat treatments with different varieties, different fertilization and irrigation levels were measured by an ASD FieldSpec FR spectrometer in six different growth stages from erecting stage to milking stage, and the PWCs of the related wheat plant samples were also measured. Significant positive coefficients of correlation were observed between PWC and spectral reflectance in 740–930u2009nm region in all of the six different growth stages, which indicates that the NIR spectral reflectance increases due to the effect of PWC on the leaf internal structure. This mechanism also affects the red edge spectrum in 680–740u2009nm region. The spectral reflectance increases more rapidly and the red edge becomes steeper if PWC is higher. The coefficients of correlation between PWC and red edge width, derived from the inverted-Gaussian model, are significant at the 0·999 confidence-level, which is more reliable than WI and NDWI, and the statistical models for PWC based on red edge width were set up in all the six different growth stages. In addition, LAI and canopy chlorophyll density (CCD) are also related to red edge parameter, such as red edge position and red edge width. It seems that PWC plays a more important role in red edge width than LAI and CCD due to the effect of PWC on the leaf internal structure, and that CCD plays a more important role in red edge position than LAI and PWC.

A patch-based image classification by integrating hyperspectral data with GIS

Hyperspectral remote sensing data provide detailed spectral information and are widely used for pixel‐based image classification. However, without considering spatial correlation among neighbouring pixels, a generated thematic map may have a ‘salt‐and‐pepper’ appearance. With the development of the Geographic Information System (GIS), the spatial relationship between a pixel and its neighbours can be recorded readily and used together with remote sensing data. The objective of this study was to integrate hyperspectral data with the GIS for effective thematic mapping. To date, GIS data have been used mainly in field surveys or training field selection for remote sensing data interpretation. Here we propose a patch‐classification based on integration of the GIS with remote sensing data. The classification results obtained by using this method can be easily saved in a vector format as used for GIS files. Computational cost is decreased compared with a pixel‐by‐pixel classification. The issue of how to identify pure or mixed patches is addressed and a three‐level simple and effective checking method is developed. A case study is presented with a hyperspectral data set recorded by the Pushbroom Hyperspectral Imager (PHI) and related GIS data.

Hyperspectral remote sensing in China

In recent years, hyperspectral remote sensing has stepped into a new stage in China. There are some advanced hyperspectral imagers and CCD cameras developed by Chinese institutes and companies. Pushbroom Hyperspectral Imager (PHI) and Operative Modular Imaging Spectrometer (OMIS) have presented the level of airborne hyperspectral imagers in China, which have been developed by the Chinese Academy of Sciences. A narrow band hyperspectral digital camera system (HDCS) was developed and tested in 2000, the center of wavelength of which can be changed to fit different applications. There is also a kind of Fourier Imaging Spectrometer developed in China. Accordingly, Chinese scholars have created a number of models to meet different application problems. Some new models for hyperspectral remote sensing are provided. They are Hyperspectral Data Classification Model, POS Dat Geometric Correction Model, Derivative Spectral Model (DSM), Multi-temporal Index Image Cube Model (MIIC), Hybrid Decision Tree Model (HDT) and Correlation Simulating Analysis Model (CSAM). Some successful applications are provided and evaluated.

Kernel adaptive filter (SRSSHF) and quality improvement method for hyperspectral imaging based on spectral dimension recognition and spatial dimension smoothing according to CSAM

According to the advanced feature of hyperspectral image and Correlation Simulating Analysis Model (CSAM), a new simple but efficient kernel-adaptive filter (SRSSHF) especially for hyperspectral image is suggested in this paper. It is achieved not based on the traditional sigma (standard deviation) statistics in spatial dimension, but on the valid-pixel judge in spectral dimension and the intellectualized shift convolution in spatial dimensions. So its criteria is based on the intrinsic property of objects by adequately utilizing the spectral information that hyperspectral affords. Such a filter also is an adaptive filter, and its kernel size theoretically has no strong influence on the filter results. What it concentrates is the feature of signal itself but not the speckle noise, its criterion is in spectral dimension, and multiple iteration is available. So the tradeoff of spatial texture is not necessary. It is applied to filter and improve quality of PHI hyperspectral images acquired both in Changzhou, China and Nagano, Japan, and a >200 looks iteration and a comparison with other typical adaptive filters also are tried. It shows that SRSSHF can smooth whole the internal of a homogeneous area while ideally keep and, as well as, enhance the edges well. As good results are achieved, this paper suggests that SRSSHF on the base of CSAM is a relative ideal filter for HRS images. Some other features of SRSSHF are also discussed in this paper.

Automatic flat field algorithm for hyperspectral image calibration

Image spectra calibration is of great importance for further processing and feature extraction. In this paper, an automated flat field reflectance calibration algorithm (AFFT) is proposed. This algorithm is an improvement to the traditional flat field transformation calibration. It is based on the fact that the so-called flat field is a flat block of high brightness and relative flat spectral response, and at a certain wavelength range (.e.g. 500-700nm) the brightness or radiance of the flat field is a certain multiple of the average spectrum of the image. Because the average image spectrum spectrum usuall is relatively flat, so a certain multiple of the average spectrum can be regarded as the criterion (or threshold) to select flat field pixels. So such parameters as wavelength range, multiple increment between flat field and the average image spectrum and number of the largest area block are set to determine the useful flat field so that an average spectrum of the flat field is obtained. By using this flat field spectrum as solar/atmospheric response, hyperspectral image can be calibrated to reflectance image. In the end, AFFT was validated by one PHI image acquired in Japan, 2000. It turns out that AFFT is effective to search all the flat fields which meet the fixed terms automatically and promptly, the spectra transformed by this method are much smoother and reliable to some extent.

Spectral unmixing and image classification supported by spatial knowledge

Usually the spectral unmixing and endmember extraction were based on the spectral statistics algorithm. In this paper, spatial knowledge, such as field patch information, was involved in the pure pixel selecting. In this way, endmember extraction was not only carried out in spectral space but also considering the spatial location of pixels. In addition, these known background information can also improve the accuracy of image classification, and also can be used to intellectually separate pixels and evaluate each sub-pixels different attributes.

Hyperspectral remote sensing monitoring on hot wastewater of Futtsu Power Plant in Tokyo Bay, Japan, using airborne Operational Modular Imaging Spectrometer (OMIS)

An experimental study in monitoring the hot wastewater which is discharged into sea by the Futtsu Power Plant on the east coast of Tokyo Bay, Japan, was carried out in August-September, 2001, by using airborne hyperspectral remote sensing (HRS) sensor OMIS (Operational Modular Imaging Spectrometer). The fundamental progress of experiment, features of OMIS HRS image, data progressing and information extraction technologies, and a primary but successful result are introduced in detail. A new algorithm to extract the features and the infection extension of hot wastewater is developed and suggested in this paper. The algorithm adequately uses the whole spectral range of OMIS according to the general spectral responding characters of water. The water in the whole area is extracted by its spectral features in VNIR at first and then the polluted water is picked out from it by combine-using the MIR and TIR information. As a result, a temperature distribution map is successfully achieved in a test area and some other abnormal points are popped out and therefore some other pollution sources are discovered successfully in the whole scopes. The relatively good results in this paper show that hyperspectral remote sensing technology has a great prospect in detecting ocean and coastal environment both in qualitatively and quantitatively, at least for the hot wastewater. And an OMIS system with the algorithm suggested in this paper is operational for monitoring the infection features of hot wastewater.

Vegetation classification and soil moisture calculation using land surface temperature (LST) and vegetation index (VI)

In this paper, the temperature-missivity separating (TES) method and normalized difference vegetation index (NDVI) are introduced, and the hyperspectral image data are analyzed using land surface temperature (LST) and NDVI channels which are acquired by Operative Module Imaging Spectral (OMIS) in Beijing Precision Agriculture Demonstration Base in Xiaotangshan town, Beijing in 26 Apr, 2001. Firstly, the 6 kinds of ground targets, which are winter wheat in booting stage and jointing stage, bare soil, water in ponds, sullage in dry ponds, aquatic grass, are well classified using LST and NDVI channels. Secondly, the triangle-like scatter-plot is built and analyzed using LST and NDVI channels, which is convenient to extract the information of vegetation growth and soils moisture. Compared with the scatter-plot built by red and near-infrared bands, the spectral distance between different classes are larger, and the samples in the same class are more convergent. Finally, we design a logarithm VIT model to extract the surface soil water content (SWC) using LST and NDVI channel, which works well, and the coefficient of determination, R2, between the measured surface SWC and the estimated is 0.634. The mapping of surface SWC in the wheat area are calculated and illustrated, which is important for scientific irrigation and precise agriculture.

Image classification supported by digital geomorphology model

In some complicated terrain area, such a loess plateau of China, it is very difficult to get higher accuracy of landuse classification only depending on the traditional spectral statistics methods, especially the image pixel size is much larger than the geomorphology units. In order to improve the image classification results, large scale relief map has been used to create the digital geomorphology model(DGM). DGM can be used to do the pixel unmixing works, specially reducing the influence of terrain shadow. Applying fuzzy mathematics theory, the DGM has been used to correct the digital image classification result, so as to create more accurate landuse map. In addition, this method is also helpful to find some minor objects in low spatial resolution images.