• Korean Journal of Remote Sensing
• /
• v.29 no.1
• /
• pp.95-104
• /
• 2013

In this study, a novel method is proposed for retrieving relative surface temperature from single-channel middle infra-red (MIR, 3-5 ${\mu}m$) remotely sensed data. In order to retrieve absolute temperature from MIR data, it is necessary to accommodate at least atmospheric effects, surface emissivity and reflected solar radiance. Instead of retrieving kinematic temperature of each target, we propose an alternative to retrieve the relative temperature between two targets. The core idea is to minimize atmospheric effects by assuming that the differential at-sensor radiance between two targets experiences the same atmospheric effects. To reduce effective simplify atmospheric parameters, each atmospheric parameter was examined by MODTRAN and MIR emissivity derived from ASTER spectral libraries. Simulation results provided a required accuracy of 2 K for materials with a temperature of 300 K within 0.1 emissivity errors. The algorithm was tested using MODIS band 23 MIR day time images for validation. The accuracy of retrieved relative temperature was $0.485{\pm}1.552$ K. The results demonstrated that the proposed algorithm was able to produce relative temperature with a required accuracy from only single-channel radiance data. However, this method has limitations when applied to materials having very low temperatures using day time MIR images.

• Korean Journal of Remote Sensing
• /
• v.33 no.5_1
• /
• pp.599-605
• /
• 2017

Accurate cloud discrimination in satellite images strongly affects accuracy of remotely sensed parameter produced using it. Especially, cloud contaminated pixel over ocean is one of the major error factors such as Sea Surface Temperature (SST), ocean color, and chlorophyll-a retrievals,so accurate cloud detection is essential process and it can lead to understand ocean circulation. However, static threshold method using real-time algorithm such as Moderate Resolution Imaging Spectroradiometer (MODIS), Advanced Himawari Imager (AHI) can't fully explained reflectance variability over ocean as a function of relative positions between the sun - sea surface - satellite. In this paper, we assembled a reflectance spectral library as a function of Solar Zenith Angle (SZA) and Viewing Zenith Angle (VZA) from ocean surface reflectance with clear sky condition of Advanced Himawari Imager (AHI) identified by NOAA's cloud products and spectral library is used for applying the Dynamic Time Warping (DTW) to detect cloud pixels. We compared qualitatively between AHI cloud property and our results and it showed that AHI cloud property had general tendency toward overestimation and wrongly detected clear as unknown at high SZA. We validated by visual inspection with coincident imagery and it is generally appropriate.

• Journal of Korean Society of Forest Science
• /
• v.96 no.3
• /
• pp.259-268
• /
• 2007

This study was conducted to produce forest thematic maps and estimate forest statistics for Pyeongchang Gun using the kNN technique, which has been applied to produce thematic maps of variables of interest including unobserved plots by combining field plot data, remotely sensed data and other digital map data in forest inventories. The estimation errors for three horizontal reference areas (HRAs), whose radii are 20, 40 and 60 km respectively, were compared. Although the precision for the 40 km radius was lower compared to that for the 60 km radius, the 40 km radius was found to be an efficient HRA because their difference in precision was modest. At a value of k=5 nearest neighbors for the selected HRA, the overall accuracy was high. As a result, using the k=5 neighbors within the HRA of 40 km radius, thematic maps of number of trees, basal area, and growing stock per hectare were generated. As compared to the forest statistics based on field sample plots, the estimated means of each parameter from the produced maps were underestimated.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.24 no.9
• /
• pp.944-950
• /
• 2013

This paper is a study of a rain sensor using an open-ended coaxial cavity resonator which senses the amount of rain drops linearly. It shows that it will be used as a sensor to sense the amount of rain dropped on the windshield of an automobile based on the principle of varied resonant frequency and the loss according to the amount and characteristics of an dielectric lied on the open side of a resonator. The input and output ports are built in the both sides of the resonator and the input and output coupling probes are formed like 'ㄱ' shape. The response of rain drops were simulated by the radius of inner conductor of 2 mm, 5 mm, and 10 mm respectively and it showed that the raindrop was sensed most linearly and sensitively when the radius of inner conductor is 5 mm, We have measured that the resonant frequency have varied from 3.55 GHz to 3 GHz and the Q value have varied from 42.38 to 24.3 according to the variation of rain drop amount on the fabricated resonator. Therefore, it shows that the designed resonator can be applied as a rain sensor that measures the amount of rain drops linearly by using the resonant frequency as a measurement parameter.

• Journal of Korean Society of Forest Science
• /
• v.111 no.4
• /
• pp.603-612
• /
• 2022

The study aims were to evaluate a machine-learning, algorithm-based, forest biomass-estimation model to estimate subnational forest biomass and to comparatively analyze REDD+ forest reference emission levels. Time-series Landsat satellite imagery and ESA Biomass Climate Change Initiative information were used to build a machine-learning-based biomass estimation model. The k-nearest neighbors algorithm (kNN), which is a non-parametric learning model, and the tree-based random forest (RF) model were applied to the machine-learning algorithm, and the estimated biomasses were compared with the forest reference emission levels (FREL) data, which was provided by the Paraguayan government. The root mean square error (RMSE), which was the optimum parameter of the kNN model, was 35.9, and the RMSE of the RF model was lower at 34.41, showing that the RF model was superior. As a result of separately using the FREL, kNN, and RF methods to set the reference emission levels, the gradient was set to approximately -33,000 tons, -253,000 tons, and -92,000 tons, respectively. These results showed that the machine learning-based estimation model was more suitable than the existing methods for setting reference emission levels.