Information on the ecological and physiological status of crops is essential for growth diagnostics and yield prediction. Within-field or between-field spatial information is required, especially with the recent trend toward precision agriculture, which seeks the efficient use of agrochemicals, water, and energy. The study of carbon and nitrogen cycles as well as environmental management on local and regional scales requires assessment of the spatial variability of biophysical and ecophysiological variables, scaling up of which is also needed for scientific and decision-making purposes. Remote sensing has great potential for these applications because it enables wide-area, non-destructive, and real-time acquisition of information about ecophysiological conditions of vegetation. With recent advances in sensor technology, a variety of electromagnetic signatures, such as hyperspectral reflectance, thermal-infrared temperature, and microwave backscattering coefficients, can be acquired for both plants and ecosystems using ground-based, airborne, and satellite platforms. Their spatial and temporal resolutions have both recently been improved. This article reviews the state of the art in the remote sensing of plant ecophysiological data, with special emphasis on the synergy between remote sensing signatures and biophysical and ecophysiological process models. Several case studies for the optical, thermal, and microwave domains have demonstrated the potential of this synergistic linkage. Remote sensing and process modeling methods complement each other when combined synergistically. Further research on this approach is needed for a wide range of ecophysiological and ecosystem studies, as well as for practical crop management.

View full abstract

Particulate contamination control is one of the most important processes to maintain the performance of an optical sensor through its assembly, test and storage phases. The quantitative evaluation of surface cleanliness level of a satellite is the most essential for the contamination control. There are two kinds of ways to measure the surface cleanliness level. One is the way to set up collector plates as close to a satellite as possible to make accumulate suspended particulates on their surfaces, the other is to make contaminated particulates adhere to tapes and count them with a microscope. Both methods are complementary and it is necessary to compare their accuracies for the same sample to evaluate the availability of each method. In this study, six kinds of adhesive tapes were selected and their particulate collection efficiencies were compared with the method with collector plates. It was found that the collection efficiency depended not only on adhesive force but other characteristics of tapes, such as its backing materials and thickness. The cleanroom-tape with moderate adhesion had ability to strip out particles larger than 10, um, but significant amount of particles smaller than l0, um were left on objective surfaces. From the observed particle size distributions at Tanegashima Space Center, it was found that the particles with their size of 10, um had the largest contribution to area coverage and particles larger than 10/um accounted for 80% of total area coverage. The axial ratio distribution of stripped particles was almost the same as that before measured stripping though apparent axial shape of large particles stripped from objective surfaces rounded slightly due to the deformation of adhesive. Although there is a problem of remains adhesive, it was found that the tape method is effective for collecting large particles. Its handiness allows us to use for the periodical check method of the contamination control as a complementary method to the collector plates.

View full abstract

A GPR (Ground Penetration Radar) is a narrow bandwidth, high radar range device, where improvement of resolution is greatly desired to resolve the closely buried targets and clear imaging of targets. However, it is not easy to get high resolution, as the GPR signals are very weak and enveloped inside the noise. The MUSIC (Multiple Signal Classification) algorithm, which has its fame for its super resolution capacity, has been implemented. In addition, conventional spectral estimation technique, FFT (Fast Fourier Transform), has also been implemented for high precision receiving signal level. As well as this, CPM (Combined Processing Method) was performed for the signal processing and image reconstruction of closely located GPR targets. In this paper we proposed the method to improve the SNR (Signal to Noise ratio) of MUSIC and CPM. In order to support the proposal, a detailed simulation has been preformed analyzing SNR. Moreover, a laboratory experiment has also been performed for the thorough investigation and maximum resolution capacity of CPM. The experiment result demonstrates the CPM has the vertical resolution of 2 cm at 800 MHz frequency bandwidth in a water medium. All the simulation and experimental results are presented.

View full abstract

This paper develops a new method for evaluating rock mass slope stability by using the remote sensing on the ground, and reports the results of applying the method to an actual slope. The slope consists of andesite which has various degrees of weathering. We observe the spectral reflectance and temperature of the slope surface. According to the characteristic of temperature, the slope surface is divided into many areas. At each area, we calculate La b values from the observed spectral reflectance. At each area, we measure the seismic wave velocity of slope surface and the ultrasonic wave velocity of rock blocks. From the experimental relationship between the b value and the seismic wave velocity, etc., it is possible to estimate the seismic wave velocity of slope surface and the ultrasonic wave velocity from the observed b value. When getting the seismic velocity, etc., we can estimate the slope stability by using the conventional relationships. This result shows the possibility that we can evaluate the stability of rock mass slope, based on the remote sensing technique.

View full abstract

The purpose of this study is to propose a model to calculate quantity of generated heat from road traffic for heat island analysis. In this paper, we proposed and verified the method of simultaneous estimation of temperature and emissivity of automobile's surface using thermal images. We measured spectrum emissivity using sample tip of automobile's surface materials. When observed at an angle of incidence similar to the one observed by infrared thermography from the building rooftop, long-wavelength radiation from the wall and the window surface of the opposite buildings reflected automobile's surface could not be ignored. Based on these data, we proposed the method of simultaneous estimation of "emissivity and surface temperature of an automobile's body by compensating of reflection of radiant temperature observed by infrared thermography with 8-14μm bands. To evaluate the application of this method, radiant temperature distribution of bonnets, roofs, and glasses of cars were observed by infrared thermography. The results show that surface temperature measured by thermocouple and corrected surface temperature by thermal image almost agreed. So this method is effective for estimation of emissivity and surface temperature of an automobile's body.

View full abstract

We conducted a cross-calibration experiment on airborne L-band SAR systems (JAXA/CRL Pi-SAR and JPL AIRSAR) in the Tottori Sand Dune. This experiment forms a part of the Pacific Rim-2000 campaign conducted from July to October in 2000. Both Pi-SAR and AIRSAR acquired their data over the test site within two hours on October 4, 2000. The aircraft track altitudes were about 8000 m. Four rectangular trihedral corner reflectors with an 80 cm leg-length were deployed in order to estimate absolute calibration coefficients. We also deployed four dihedral corner reflectors with different rotation angles for polarimetric calibration. This report describes the results of calibration experiment for Pi-SAR and AIRSAR systems conducted on the Tottori Sand Dune.
The Pi-SAR distortion matrices for both transmitting and receiving systems were calculated by using the data containing one trihedral reflector and two dihedral reflectors with different rotation angles. Gain balance between HH and VV was not good this time because Pi-SAR had some hardware degradation. The absolute calibration coefficient for the HH channel has almost the same as in the previous calibration experiment in 1998. However, we found that the VV channel had degraded more than 3 dB. The improvements of gain balance and crosstalk between H and V channels were confirmed after the calibration using the extracted distortion matrices.
We compared Pi-SAR data with AIRSAR data at point targets as well as distributed targets. Relatively large differences were observed in low backscattering targets with VV and HV channels. These differences are presumably due to increased noise equivalent backscattering coefficients caused by the hardware degradation of Pi-SAR V system data chain.

View full abstract

RGB values from CCD cameras are device-dependent, and are not able to be simply converted into international color standard CIE XYZ (and Lab) values. We have developed a calibration method from these RGB values to CIE Lab values by using a standard color sample set. We achieved the accuracy of ΔEab 2.23 (standard deviation of error×2.5) by applying a couple of calculation techniques to the calibration. First, accuracies of X, Y and Z estimations in the lower range are enhanced, while those in the higher are diminished. In the converting equations from XYZ to Lab, XYZ values are powered by third. Hence, estimation error in low XYZ region reduces Lab estimation accuracy more effectively. The second method is interactive modifications of X and Z values. Values of a and b is derived by Y and X, and Y and Z, respectively. For example, a combination of larger estimation error in Y and smaller estimation error of Z yields very high deviation of b. To avoid this, we modified Z and X values based on Y estimation error, i.e. reduced X is applied to regression equations when Y is underestimated.

View full abstract