• Korean Journal of Agricultural and Forest Meteorology
• /
• v.20 no.3
• /
• pp.262-276
• /
• 2018

Models to predict Leaf Wetness Duration (LWD) were evaluated using the observed meteorological and dew data at the 11 citrus orchards in Jeju, South Korea from 2016 to 2017. The sensitivity and the prediction accuracy were evaluated with four models (i.e., Number of Hours of Relative Humidity (NHRH), Classification And Regression Tree/Stepwise Linear Discriminant (CART/SLD), Penman-Monteith (PM), Deep-learning Neural Network (DNN)). The sensitivity of models was evaluated with rainfall and seasonal changes. When the data in rainy days were excluded from the whole data set, the LWD models had smaller average error (Root Mean Square Error (RMSE) about 1.5hours). The seasonal error of the DNN model had the similar magnitude (RMSE about 3 hours) among all seasons excluding winter. The other models had the greatest error in summer (RMSE about 9.6 hours) and the lowest error in winter (RMSE about 3.3 hours). These models were also evaluated by the statistical error analysis method and the regression analysis method of mean squared deviation. The DNN model had the best performance by statistical error whereas the CART/SLD model had the worst prediction accuracy. The Mean Square Deviation (MSD) is a method of analyzing the linearity of a model with three components: squared bias (SB), nonunity slope (NU), and lack of correlation (LC). Better model performance was determined by lower SB and LC and higher NU. The results of MSD analysis indicated that the DNN model would provide the best performance and followed by the PM, the NHRH and the CART/SLD in order. This result suggested that the machine learning model would be useful to improve the accuracy of agricultural information using meteorological data.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.9
• /
• pp.351-360
• /
• 2018

The LWR (Locally Weighted Regression) model, which is traditionally a lazy learning model, is designed to obtain the solution of the prediction according to the input variable, the query point, and it is a kind of the regression equation in the short interval obtained as a result of the learning that gives a higher weight value closer to the query point. We study on an incremental ensemble learning approach for LWR, a form of lazy learning and memory-based learning. The proposed incremental ensemble learning method of LWR is to sequentially generate and integrate LWR models over time using a genetic algorithm to obtain a solution of a specific query point. The weaknesses of existing LWR models are that multiple LWR models can be generated based on the indicator function and data sample selection, and the quality of the predictions can also vary depending on this model. However, no research has been conducted to solve the problem of selection or combination of multiple LWR models. In this study, after generating the initial LWR model according to the indicator function and the sample data set, we iterate evolution learning process to obtain the proper indicator function and assess the LWR models applied to the other sample data sets to overcome the data set bias. We adopt Eager learning method to generate and store LWR model gradually when data is generated for all sections. In order to obtain a prediction solution at a specific point in time, an LWR model is generated based on newly generated data within a predetermined interval and then combined with existing LWR models in a section using a genetic algorithm. The proposed method shows better results than the method of selecting multiple LWR models using the simple average method. The results of this study are compared with the predicted results using multiple regression analysis by applying the real data such as the amount of traffic per hour in a specific area and hourly sales of a resting place of the highway, etc.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.44 no.8
• /
• pp.52-59
• /
• 2007

In this paper, we analyzed and measured implant isolation characteristics for a 1.25 Gbps monolithic integrated hi-directional (M-BiDi) optoelectronic system-on-a-chip, which is a key component to constitute gigabit passive optical networks (PONs) for a fiber-to-the-home (FTTH). Also, we derived an equivalent circuit of the implant structure under various DC bias conditions. The 1.25 Gbps M-BiDi transmit-receive SoC consists of a laser diode with a monitor photodiode as a transmitter and a digital photodiode as a digital data receiver on the same InP wafer According to IEEE 802.3ah and ITU-T G.983.3 standards, a receiver sensitivity of the digital receiver has to satisfy under -24 dBm @ BER=10-12. Therefore, the electrical crosstalk levels have to maintain less than -86 dB from DC to 3 GHz. From analysed and measured results of the implant structure, the M-BiDi SoC with the implant area of 20 mm width and more than 200 mm distance between the laser diode and monitor photodiode, and between the monitor photodiode and digital photodiode, satisfies the electrical crosstalk level. These implant characteristics can be used for the design and fabrication of an optoelectronic SoC design, and expended to a mixed-mode SoC field.

• Journal of the Korean earth science society
• /
• v.37 no.1
• /
• pp.29-39
• /
• 2016

In this study, we estimated solar radiation by multiple regression analysis using sunshine duration and precipitation data, which are highly correlated to solar radiation. We found the regression equation using data obtained from GROM (Gangwon Regional Office of Metrology, station 105, 1980-2007) located in Gangneung, South Korea and validated the equation by applying data obtained from new GROM (newly relocated, station 104, 2009-2014) and data obtained from GWNU (Gangneung-Wonju National University, 2013-2014) located between stations 104 and 105. By using sunshine duration data alone, the estimation using data from station 104 resulted in a correlation coefficient of 0.96 and a standard error of $1.16MJm^{-2}$, which was similar to the previous results; the estimation using data from GWNU yielded better results with a correlation coefficient of 0.99 and a standard error of $0.57MJm^{-2}$. By using sunshine duration and precipitation data, the estimation (using data from station 104) yielded a correlation coefficient of 0.96 and a standard error of $0.99MJm^{-2}$, resulting in a lower standard error compared to what was obtained using sunshine duration data alone. The maximum solar radiation bias increased from -26.6% (March 2013) to -31.0% (February 2011) when both sunshine duration and precipitation data were incorporated into the estimation rather than when sunshine duration data alone was incorporated. This was attributed to the concentrated precipitation found during May and July-September, which resulted in negative coefficients of the estimating equation in other months. Therefore, the monthly average solar radiation should be estimated carefully when employing the monthly average precipitation for those places where precipitation is concentrated during summer, such as the Korean peninsula.

• Journal of the Korean Institute of Telematics and Electronics A
• /
• v.33A no.2
• /
• pp.96-113
• /
• 1996

It has been found that the misfit dislocations in heavily boron-doped layers originate from wafer edges. Moreover, the propagation of the misfit dislocation into a heavily boron-doped region can be suppressed by placing a surrounding undoped region. Using a surrounding undoped region the disloction-free heavily boron-deoped silicon membranes have been fabricated. The measured surface roughness, fracture strength, and residual tensile stress of the membrane are 20.angs. peak-to-peak, 1.39${\times}$10$^{10}$ and 2.7${\times}$10$^{9}$dyn/cm$^{2}$, while those of the conventional heavily boron-doped silicon membrane with high density of misfit dislocations are 500 peak-to-peak, 8.27${\times}$10$^{9}$ and 9.3${\times}$10$^{8}$dyn/cm$^{2}$ respectively. The differences between these two membranes are due to the misfit dislocations. Young's modulus has been extracted as 1.45${\times}$10$^{12}$dyn/cm$^{2}$ for both membranes. Also, the effective lattice constant of heavily boron-doped silicon, the in-plane lattice constant of the conventional membrane, and the density of misfit dislocation contained in the conventional membrane have been extracted as density of misfit dislocation contained in the conventional membrane have been extracted as density of misfit dislocation contained in the conventional membrane have been extracted as 5.424.angs. 5.426.angs. and 2.3${\times}$10$^{4}$/cm for the average boron concentration of 1.3${\times}$10$^{20}$/cm$^{-23}$ cm$^{3}$/atom. Without any buffer layers, a disloction-free lightly boron-doped epitaxial layer with good crsytalline quality has been directly grown on the dislocation-free heavily boron-doped silicon layer. X-ray diffraction analysis revealed that the epitaxial silicon has good crystallinity, similar to that grown on lightly doped silicon substrate. The leakage current of the n+/p gated diode fabricated in the epitaxial silicon has been measured to be 0.6nA/cm$^{2}$ at the reverse bias of 5V.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.43 no.11 s.353
• /
• pp.48-57
• /
• 2006

This work describes a 12b 200KHz 0.52mA $0.47mm^2$ algorithmic ADC for sensor applications such as motor controls, 3-phase power controls, and CMOS image sensors simultaneously requiring ultra-low power and small size. The proposed ADC is based on the conventional algorithmic architecture with recycling techniques to optimize sampling rate, resolution, chip area, and power consumption. The input SHA with eight input channels for high integration employs a folded-cascode architecture to achieve a required DC gain and a sufficient phase margin. A signal insensitive 3-D fully symmetrical layout with critical signal lines shielded reduces the capacitor and device mismatch of the MDAC. The improved switched bias power-reduction techniques reduce the power consumption of analog amplifiers. Current and voltage references are integrated on the chip with optional off-chip voltage references for low glitch noise. The employed down-sampling clock signal selects the sampling rate of 200KS/s or 10KS/s with a reduced power depending on applications. The prototype ADC in a 0.18um n-well 1P6M CMOS technology demonstrates the measured DNL and INL within 0.76LSB and 2.47LSB. The ADC shows a maximum SNDR and SFDR of 55dB and 70dB at all sampling frequencies up to 200KS/s, respectively. The active die area is $0.47mm^2$ and the chip consumes 0.94mW at 200KS/s and 0.63mW at 10KS/s at a 1.8V supply.

• Journal of The Korean Astronomical Society
• /
• v.29 no.spc1
• /
• pp.63-64
• /
• 1996

The nature of distant faint blue field galaxies remains a mystery, despite the fact that much attention has been devoted to this subject in the last decade. Galaxy counts, particularly those in the optical and near ultraviolet bandpasses, have been demonstrated to be well in excess of those expected in the 'no-evolution' scenario. This has usually been taken to imply that galaxies were brighter in the past, presumably due to a higher rate of star formation. More recently, redshift surveys of galaxies as faint as B$\~$24 have shown that the mean redshift of faint blue galaxies is lower than that predicted by standard evolutionary models (de-signed to fit the galaxy counts). The galaxy number count data and redshift data suggest that evolutionary effects are most prominent at the faint end of the galaxy luminosity function. While these data constrain the form of evolution of the overall luminosity function, they do not constrain evolution in individual galaxies. We are carrying out a series of observations as part of a long-term program aimed at a better understanding of the nature and amount of luminosity evolution in individual galaxies. Our study uses the luminosity-linewidth relation (Tully-Fisher relation) for disk galaxies as a tool to study luminosity evolution. Several studies of a related nature are being carried out by other groups. A specific experiment to test a 'no-evolution' hypothesis is presented here. We have used the AUTOFIB multifibre spectro-graph on the 4-metre Anglo-Australian Telescope (AAT) and the Rutgers Fabry-Perot imager on the Cerro Tolalo lnteramerican Observatory (CTIO) 4-metre tele-scope to measure the internal kinematics of a representative sample of faint blue field galaxies in the red-shift range z = 0.15-0.4. The emission line profiles of [OII] and [OIII] in a typical sample galaxy are significantly broader than the instrumental resolution (100-120 km $s^{-l}$), and it is possible to make a reliable de-termination of the linewidth. Detailed and realistic simulations based on the properties of nearby, low-luminosity spirals are used to convert the measured linewidth into an estimate of the characteristic rotation speed, making statistical corrections for the effects of inclination, non-uniform distribution of ionized gas, rotation curve shape, finite fibre aperture, etc.. The (corrected) mean characteristic rotation speed for our distant galaxy sample is compared to the mean rotation speed of local galaxies of comparable blue luminosity and colour. The typical galaxy in our distant sample has a B-band luminosity of about 0.25 L$\ast$ and a colour that corresponds to the Sb-Sd/Im range of Hub-ble types. Details of the AUTOFIB fibre spectroscopic study are described by Rix et al. (1996). Follow-up deep near infrared imaging with the 10-metre Keck tele-scope+ NIRC combination and high angular resolution imaging with the Hubble Space Telescope's WFPC2 are being used to determine the structural and orientation parameters of galaxies on an individual basis. This information is being combined with the spatially resolved CTIO Fabry-Perot data to study the internal kinematics of distant galaxies (Ing et al. 1996). The two main questions addressed by these (preliminary studies) are: 1. Do galaxies of a given luminosity and colour have the same characteristic rotation speed in the distant and local Universe? The distant galaxies in our AUTOFIB sample have a mean characteristic rotation speed of $\~$70 km $s^{-l}$ after correction for measurement bias (Fig. 1); this is inconsistent with the characteristic rotation speed of local galaxies of comparable photometric proper-ties (105 km $s^{-l}$) at the > $99\%$ significance level (Fig. 2). A straightforward explanation for this discrepancy is that faint blue galaxies were about 1-1.5 mag brighter (in the B band) at z $\~$ 0.25 than their present-day counterparts. 2. What is the nature of the internal kinematics of faint field galaxies? The linewidths of these faint galaxies appear to be dominated by the global disk rotation. The larger galaxies in our sample are about 2"-.5" in diameter so one can get direct insight into the nature of their internal velocity field from the $\~$ I" seeing CTIO Fabry-Perot data. A montage of Fabry-Perot data is shown in Fig. 3. The linewidths are too large (by. $5\sigma$) to be caused by turbulence in giant HII regions.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.46 no.3
• /
• pp.60-68
• /
• 2009

This work proposes a 10b 200MS/s 65nm CMOS ADC for high-definition video systems such as HDTV requiring high resolution and fast operating speed simultaneously. The proposed ADC employs a four-step pipeline architecture to minimize power consumption and chip area. The input SHA based on four capacitors reduces the output signal range from $1.4V_{p-p}$ to $1.0V_{p-p}$ considering high input signal levels at a low supply voltage of 1.2V. The proposed three-stage amplifiers in the input SHA and MDAC1 overcome the low output resistance problem as commonly observed in a 65nm CMOS process. The proposed multipath frequency-compensation technique enables the conventional RNMC based three-stage amplifiers to achieve a stable operation at a high sampling rate of 200MS/s. The conventional switched-bias power-reduction technique in the sub-ranging flash ADCs further reduces power consumption while the reference generator integrated on chip with optional off-chip reference voltages allows versatile system a locations. The prototype ADC in a 65nm CMOS technology demonstrates a measured DNL and INL within 0.19LSB and 0.61LSB, respectively. The ADC shows a maximum SNDR of 54.BdB and 52.4dB and a maximum SFDR of 72.9dB and 64.8dB at 150MS/S and 200MS/s, respectively. The proposed ADC occupies an active die area of $0.76mm^2$ and consumes 75.6mW at a 1.2V supply voltage.

• Journal of Yeungnam Medical Science
• /
• v.8 no.2
• /
• pp.150-157
• /
• 1991

The purpose of the this study was to assess the morbidity pattern of urban residents in the poor area by health screening for the community diagnosis. The items of health screening were history taking and physical examination by medical doctor and hearing test, check blood pressure, test for hematocrit, liver function(sGOT, sGPT), urine sugar and protein, and chest X-ray. The examinee in health screening were 437 persons and they occupied 16.9% of total residents in the poor area. Male examinee were 129 persons(9.9% of total residents) and female examinee were 308 persons(23.9% of total residents). Age group of above sixty years old, 42.0% of total residents in the poor area were participated, but only 5.9% were participated in age group of 10 to 19 years old. Among the 437 examinee, 191 persons(43.7%) had one or more abnormal findings in health screening. In male 38.7% had abnormal findings, and some what lower than that of female(45.8%). Age group of above sixty years had most high rate of abnormal findings(69.8%), in contrast to age group of 10 to 19 years old (10.9%). Diseases of the digestive system was the most common and which occupies 23.7% of total abnormal findings. And diseases of the circulatory system occupied 19.7%. Low hematocrit(14.6% of total participants of 437 persons) occupies the most common abnormal findings for screening test(hematocrit, blood pressure, hearing test, sGOT/sGPT, urine protein and urine sugar, chest X-ray) and high blood pressure(10.1%) occupied second, third; hearing impairment (5.5%), fourth ; abnormal liver function (4.1%), fifth ; sugar in urine (2.3%), sixth ; protein in urine(1.4%) and lastly abnormal chest X-ray (0.9%). The positive rate of abnormal findings in health screening was very high compared with morbidity rate by health interview. It is supposed that some portion of this high rate is by selection bias in examinee in health screening specially high participating rate in older age, and the other portion is due to the low socioecomic status and bad environment of the residents of the poor area. These findings will be good information for the research and development of health care system in the urban poor area.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.18 no.4
• /
• pp.233-241
• /
• 2016

Experiments were carried out to quantify the topographic effects on attenuation of sunshine in complex terrain and the results are expected to help convert the coarse resolution sunshine duration information provided by the Korea Meteorological Administration (KMA) into a detailed map reflecting the terrain characteristics of mountainous watershed. Hourly shaded relief images for one year, each pixel consisting of 0 to 255 brightness value, were constructed by applying techniques of shadow modeling and skyline analysis to the 3m resolution digital elevation model for an experimental watershed on the southern slope of Mt. Jiri in Korea. By using a bimetal sunshine recorder, sunshine duration was measured at three points with different terrain conditions in the watershed from May 15, 2015 to May 14, 2016. The brightness values of the 3 corresponding pixel points on the shaded relief map were extracted and regressed to the measured sunshine duration, resulting in a brightness-sunshine duration response curve for a clear day. We devised a method to calibrate this curve equation according to sky condition categorized by cloud amount and used it to derive an empirical model for estimating sunshine duration over a complex terrain. When the performance of this model was compared with a conventional scheme for estimating sunshine duration over a horizontal plane, the estimation bias was improved remarkably and the root mean square error for daily sunshine hour was 1.7hr, which is a reduction by 37% from the conventional method. In order to apply this model to a given area, the clear-sky sunshine duration of each pixel should be produced on hourly intervals first, by driving the curve equation with the hourly shaded relief image of the area. Next, the cloud effect is corrected by 3-hourly 'sky condition' of the KMA digital forecast products. Finally, daily sunshine hour can be obtained by accumulating the hourly sunshine duration. A detailed sunshine duration distribution of 3m horizontal resolution was obtained by applying this procedure to the experimental watershed.