The feasibility of a unique greenhouse, named Gradient Biome, is now being examined extensively in the University of Tokyo. It is a large chamber (length: 200 m, width: 50 m, height: 40 m) in which elements of the weather, such as the air temperature and humidity, are controlled in such a way as to create a continuous gradient from the tropical to the frigid zones on the earth. In the Gradient Biome, ecosystems (mainly plants) corresponding to each test climate are introduced and the possible responses of these ecosystems to forecasted global warming will be observed. Since one of the expected responses is the shift of the ecosystem(s) toward a more suitable environment, there should be no artificial obstacles preventing the shift. This requirement is difficult to satisfy as the environments, (air temperature and humidity), blend and become homogeneous. This paper presents the results of the numerical and experimental studies conducted to find ways that the air temperature and humidity could be graduated (rather than homogeneous) in the Gradient Biome. From the results of numerical simulation, validated by comparison with experiment, we suggest a method to create a continuous gradient of the environmental variables inside the Gradient Biome.

View full abstract

To reduce irremovable radiometric distortion on images observed with remote sensing sensors, reflectance adjustment approach based on the positional relationship between soil line and full vegetation cover line in red reflectance (R ) and near infrared reflectance (NIR ) space and raNDVI (Reflectance-adjusted Normalized Difference Vegetation Index) were proposed. The raNDVI was subsequently compared its reflectance adjustment effect with other conventional vegetation indices, NDVI, SAVI and MSAVI by evaluating errors of each pixel value between two years in three different vegetation type regions. As a result of this experiment, it confirmed that the proposed approach properly corrected the ratio of R to NIR  for unifying the vegetation index values in the preserved vegetation areas. It is also showed that this method had suitably changed the length and direction of the reflectance correction according to the positional gap between whole scatter plots and origin of R, NIR  feature space on each image.

View full abstract

Vertical plant area density profiles of rice canopy at different four growth stages were estimated using a high-resolution portable scanning lidar. 3D point cloud data on the canopies were collected by a portable scanning lidar with laser beams inclined at zenith angle of 30°. This laser beam zenith angle was selected to avoid obstruction of the beam into the lower canopy by the upper part. Based on the lidar-derived images, PAD values were computed in each growth stage by using the voxel-based canopy profiling method. The values were compared with ones directly measured by stratified clipping. The root mean square errors were 0.36 to 1.81 m² m^-3 at each growth stage and 1.20 m² m^-3 across all growth stages. Plant area index was also estimated, with absolute percentage errors of 4.0 to 20.6% at each growth stage and of 9.9% across all growth stages.

View full abstract