• The Journal of the Acoustical Society of Korea
• /
• v.34 no.1
• /
• pp.1-11
• /
• 2015

Salinity affects sound speed in the low salinity environment, in the seas where freshwater from large rivers and flows into the marginal sea area near the Yangtze River and the Niger River. In this paper, SSC (Surface Sound Channel) formed by low salinity water was investigated in the East China Sea and the Gulf of Guinea of rainy season. The data from KODC (Korea Oceanographic Data Center) in the East China Sea and from ARGO (Array for Real-time Geostrophic Oceanography) in the Gulf of Guinea of the tropical area were used for analysis. SSC haline channel was formed 14 times among 32 SSC occurrences when the 90 data from 9 points were analyzed during a decade (2000 ~ 2009) in the East China Sea. In the Gulf of Guinea, haline channel was formed 18 times among 20 SSC occurrences during 3 years (2006 ~ 2009). When the sound speed gradient was analyzed from temperature-salinity gradient diagram, the gradients of both salinity and temperature affect SSC formation in the East China Sea. In contrast, the salinity gradient mostly affects SSC formation due to the least change of temperature in the well-developed mixed layer in the Gulf of Guinea. Their acoustic characteristics show that channel depth is 6.5 m, critical angle is $1.5^{\circ}$ and difference of transmission loss between surface and thermocline is 11.5 dB in the East China Sea, while channel depth is 18 ~ 24 m, critical angle is $4.0{\sim}5.4^{\circ}$ and difference of transmission loss is 21.5 ~ 27.9 dB in the Gulf of Guinea. These results are expected to be used as a basic understanding of the acoustic transmission changes due to low salinity water at the estuaries and the ocean with heavy precipitation.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.64 no.4
• /
• pp.395-405
• /
• 2019

A recent assessment by the Intergovernmental Panel on Climate Change projected that the global average surface temperature will increase by a value 1.5℃ from 2030 to 2052. In this study, we used a temperature gradient chamber that mimicked field conditions to evaluate the effect of increased air temperature on phenology, yield components, protein content, and oil content, to assess soybean growth. In 2017 and 2018, 'Deawonkong', 'Pungsannamulkong', and 'Deapungkong' cultivars were grown in three temperature gradient chambers. Four temperature treatment groups were established by dividing the rows along temperature regimes: ambient temperature + 1℃ (aT+1), ambient temperature + 2℃ (aT+2), ambient temperature + 3℃ (aT+3), ambient temperature + 4℃ (aT+4). Year, cultivar, and temperature treatments significantly affected yield components and seed yield. In 2017, the flowering stage of 'Deawon' and 'Pungsannamul' cultivars in the aT+4 group was delayed compared to the flowering stage of those in the aT+1 group. In 2018, the flowering stage of 'Deawon' and 'Pungsannamul' was delayed at all temperature gradients, owing to high temperature stress, whereas 'Deapung' was regularly flowering in 2017 and 2018. The duration of the grain filling period was six days shorter in 2018 than in 2017 because of high temperature stress. The total number of pods per ㎡ for 'Deawon' and 'Pungsannamul' was 48.8 and 41.5% lower in 2018 than in 2017, respectively, whereas 'Deapung' increased by 6.3%. The 100-seed weight of 'Deawon' and 'Deapung' was 29.2 and 32.1% lower, respectively. However, 'Pungsannamul' decreased by 14.7%. The protein and oil content was lower during the grain filling period in 2018 than in the same period in 2017 because of high temperature stress. In contrast, the oil content in 'Deapung' was higher in 2018 than in 2017. Our results showed that increased temperature during the grain filling period was significantly and negatively correlated with pod number, 100-seed weight, protein content, and oil content.

• Korean Journal of Ecology and Environment
• /
• v.42 no.2
• /
• pp.183-191
• /
• 2009

This study was to evaluate ecological health of Dongjin River in October 2007. The ecological health assessments was based on the Index of Biological Integrity (IBI), Qualitative Habitat Evaluation Index (QHEI), and water chemistry. For the study, the models of IBI and QHEI were modified as 8 and 11 metric attributes, respectively. We also analyzed spatial patterns of chemical water quality over the period of 2005${\sim}$2008, using the water chemistry dataset, obtained from the Ministry of Environment, Korea. In Dongjin River, values of IBI averaged 19 (n=3), which is judged as a "Fair" condition after the criteria of Barbour et al. (1999). There was a distinct spatial variation. IBI score at Site 1 was estimated as 28, indicating a "Good" condition whereas, IBI at Site 2 and Site 3 were as 18 and 12, indicating "Fair" and "Poor" condition, respectively. Habitat analysis showed that QHEI values in the river averaged 117 (n=3), indicating a "Fair${\sim}$Good" condition after the criteria of Barbour et al. (1999). Values of BOD and COD averaged 2.3 mg $L^{-1}$ (scope: 0.1${\sim}$8.9 mg $L^{-1}$) and 5.5 mg $L^{-1}$ (scope: 1.8${\sim}$12.6 mg $L^{-1}$), respectively during the study. Total nitrogen (TN) and total phosphorus (TP) averaged 2.7mg $L^{-1}$ and 0.127mg $L^{-1}$, respectively, and the nutrients showed large longitudinal gradients between the upper and lower reach. Overall, dataset of IBI, QHEI, and water chemistry showed that river health was a gradual decline at upstream to downstream. So, Dongjin River should be protected from habitat disturbance and chemical pollutions.

• Korean Journal of Ecology and Environment
• /
• v.47 no.spc
• /
• pp.48-56
• /
• 2014

Weekly changes of water environments and phytoplankton community with the salinity gradients were investigated at Gyoungpo Lake from April to November in 1998 and 2012. Underwater crossam in Gyoungpo Lake was removed in 2004. Thereafter, average salinity of Gyoungpo lake increased from 7.5 ppt in 1998 to 20 ppt in 2012. A total of 99 and 80 species of phytoplankton was observed from the sampled in 1998 and 2012, respectively. The number of common species during the 2 separate years was 40. Transparency, SS, $NO_3-N$ concentration and N/P ratio in 2012 were lower than those in 1998. During the period of water shortage (April, May) of 2012 transparency decreased due to decreased salinity and increased SS and Chl. a. Correlation coefficients between species and community scores of DCA ordination based on data matrix of the phytoplankton revealed larger variation among sampling seasons in 1998 than in 2012. The increase of seawater influx and conversion rates following the removal of the underwater crossbeam might explain such a differential variation. Gymnodium sp., Peridinium sp., Prorocentrum sp., Nitzschia longissima, Schroederia setigera, Lyngbya sp., Asterococcus limneticus, Asterococcus superbus and Cyclotella meneghiniana were found to well adapt at the high salinities in 2012. Comparatively, Asterrionella formosa, Nitzschia frustulum, Chlorella ellipsoidea, Scenedesmus bijuga and Scenedesmus ellipsoideus were observed at lower salinities in 1998. Two quite contrasting phytoplankton communities were found in the two seasons of a year, spring with limited precipitation and summer, the flood season.

• Journal of Intelligence and Information Systems
• /
• v.22 no.2
• /
• pp.127-142
• /
• 2016

Deep learning model is a kind of neural networks that allows multiple hidden layers. There are various deep learning architectures such as convolutional neural networks, deep belief networks and recurrent neural networks. Those have been applied to fields like computer vision, automatic speech recognition, natural language processing, audio recognition and bioinformatics where they have been shown to produce state-of-the-art results on various tasks. Among those architectures, convolutional neural networks and recurrent neural networks are classified as the supervised learning model. And in recent years, those supervised learning models have gained more popularity than unsupervised learning models such as deep belief networks, because supervised learning models have shown fashionable applications in such fields mentioned above. Deep learning models can be trained with backpropagation algorithm. Backpropagation is an abbreviation for "backward propagation of errors" and a common method of training artificial neural networks used in conjunction with an optimization method such as gradient descent. The method calculates the gradient of an error function with respect to all the weights in the network. The gradient is fed to the optimization method which in turn uses it to update the weights, in an attempt to minimize the error function. Convolutional neural networks use a special architecture which is particularly well-adapted to classify images. Using this architecture makes convolutional networks fast to train. This, in turn, helps us train deep, muti-layer networks, which are very good at classifying images. These days, deep convolutional networks are used in most neural networks for image recognition. Convolutional neural networks use three basic ideas: local receptive fields, shared weights, and pooling. By local receptive fields, we mean that each neuron in the first(or any) hidden layer will be connected to a small region of the input(or previous layer's) neurons. Shared weights mean that we're going to use the same weights and bias for each of the local receptive field. This means that all the neurons in the hidden layer detect exactly the same feature, just at different locations in the input image. In addition to the convolutional layers just described, convolutional neural networks also contain pooling layers. Pooling layers are usually used immediately after convolutional layers. What the pooling layers do is to simplify the information in the output from the convolutional layer. Recent convolutional network architectures have 10 to 20 hidden layers and billions of connections between units. Training deep learning networks has taken weeks several years ago, but thanks to progress in GPU and algorithm enhancement, training time has reduced to several hours. Neural networks with time-varying behavior are known as recurrent neural networks or RNNs. A recurrent neural network is a class of artificial neural network where connections between units form a directed cycle. This creates an internal state of the network which allows it to exhibit dynamic temporal behavior. Unlike feedforward neural networks, RNNs can use their internal memory to process arbitrary sequences of inputs. Early RNN models turned out to be very difficult to train, harder even than deep feedforward networks. The reason is the unstable gradient problem such as vanishing gradient and exploding gradient. The gradient can get smaller and smaller as it is propagated back through layers. This makes learning in early layers extremely slow. The problem actually gets worse in RNNs, since gradients aren't just propagated backward through layers, they're propagated backward through time. If the network runs for a long time, that can make the gradient extremely unstable and hard to learn from. It has been possible to incorporate an idea known as long short-term memory units (LSTMs) into RNNs. LSTMs make it much easier to get good results when training RNNs, and many recent papers make use of LSTMs or related ideas.

• Korean Journal of Environmental Biology
• /
• v.36 no.3
• /
• pp.359-369
• /
• 2018

Harmful algal blooms (HABs) are a serious problem for public health and fisheries industries, thus there exists a need to investigate the possible ways for effective control of HABs. In the present study, we investigated the algicidal effects of a newly developed GreenTD against the HABs (Chattonella marina, Heterosigma akashiwo, Cochlodinium polykriokides, and Heterocapsa circularisquama) and non-HABs (Chaetoceros simplex, Skeletonema sp. and Tetraselmis sp.), which is focused on the different population density and concentration gradients of algicidal substances. The time series viability of target alga was assessed based on the activity of Chl. a photosynthetic efficiency in terms of $F_v/F_m$, and in vivo fluorescence (FSU). Effective control of Raphidophyta, C. marina and H. akashiwo was achieved at a GreenTD concentration of $0.5{\mu}gL^{-1}$ and $0.2{\mu}gL^{-1}$, respectively, and regrowth of both the species was not observed even after 14 days. The inhibitory ratio of the dinoflagellate, C. polykriokides was more than 80% at $0.2{\mu}gL^{-1}$ of GreenTD. H. circularisquama was constantly affected in the presence of $0.2{\mu}gL^{-1}$ of GreenTD in the high- and low-population density experimental groups. On the other hand, diatoms, C. simplex, and Skeletonema sp. were not significantly affected even in the presence of $0.2{\mu}gL^{-1}$ of GreenTD and exhibited re-growth activity with the passage of incubation time. In particular, green alga Tetraselmis sp. remained unaffected even in the presence of the highest concentration of GreenTD ($1.0{\mu}gL^{-1}$), implying that non-HABs were not greatly influenced by the algicidal substances. As a result, the algicidal activity of GreenTD on the harmful and nonharmful algae was as follows: raphidophyte>dinoflagellates>diatoms>green alga. Consequently, our results indicate that inoculation of GreenTD substances into natural blooms at a threshold concentration ($0.2{\mu}gL^{-1}$) can maximize the algicidal activity against HABs species. If we consider the dilution and diffusion rate in the field application, it is hypothesized that GreenTD will demonstrate economic efficiency, thus leading to effective control against the target HABs in the closed bay.

• Land and Housing Review
• /
• v.12 no.3
• /
• pp.11-26
• /
• 2021

With the increasing demand for office space, there have been questions on how office rent distribution produces a change in the urban spatial structure in Seoul. The purpose of this paper is to investigate a relative price gradient and to present a time-series model that can quantitatively explain the dynamic changes in the urban spatial structure. The analysis was dealt with office rent above 3,306 m² for the past 10 years from 1Q 2010 to 4Q 2019 within Seoul. A modified repeat sales model was employed. The main findings are briefly summarized as follows. First, according to the estimates of the office price gradient in the three major urban centers of Seoul, the CBD remained at a certain level with little change, while those in the GBD and the YBD continued to increase. This result reveals that the urban form of Seoul has shifted from monocentric to polycentric. This shows that the spatial distribution of companies has gradually accelerated decentralized concentration implying that the business networks have become significant. Second, contrary to small and medium-sized office buildings that have undertaken no change in the gradient, large office buildings have seen an increase in the gradient. The relative price gradients in small and medium-sized buildings were inversely proportional among the CBD, the GBD, and the YBD, implying their heterogeneous submarkets by office rent movements. Presumably, those differences in the submarkets were attributed to investment attraction, industrial competition, and the credit and preference of tenants. The findings are consistent with the hierarchical system identified in the Seoul 2030 Plan as well as the literature about Seoul's urban form. This research claims that the proposed method, based on the modified repeat sales model, is useful in understanding temporal dynamic changes. Moreover, the findings can provide implications for urban growth strategies under rapidly changing market conditions.

• Korean Journal of Remote Sensing
• /
• v.38 no.5_2
• /
• pp.707-723
• /
• 2022

Although satellite-based sea surface temperature (SST) is advantageous for monitoring large areas, spatiotemporal data gaps frequently occur due to various environmental or mechanical causes. Thus, it is crucial to fill in the gaps to maximize its usability. In this study, daily SST composite fields with a resolution of 4 km were produced through a two-step machine learning approach using polar-orbiting and geostationary satellite SST data. The first step was SST reconstruction based on Data Interpolate Convolutional AutoEncoder (DINCAE) using multi-satellite-derived SST data. The second step improved the reconstructed SST targeting in situ measurements based on light gradient boosting machine (LGBM) to finally produce daily SST composite fields. The DINCAE model was validated using random masks for 50 days, whereas the LGBM model was evaluated using leave-one-year-out cross-validation (LOYOCV). The SST reconstruction accuracy was high, resulting in R² of 0.98, and a root-mean-square-error (RMSE) of 0.97℃. The accuracy increase by the second step was also high when compared to in situ measurements, resulting in an RMSE decrease of 0.21-0.29℃ and an MAE decrease of 0.17-0.24℃. The SST composite fields generated using all in situ data in this study were comparable with the existing data assimilated SST composite fields. In addition, the LGBM model in the second step greatly reduced the overfitting, which was reported as a limitation in the previous study that used random forest. The spatial distribution of the corrected SST was similar to those of existing high resolution SST composite fields, revealing that spatial details of oceanic phenomena such as fronts, eddies and SST gradients were well simulated. This research demonstrated the potential to produce high resolution seamless SST composite fields using multi-satellite data and artificial intelligence.

• Ecology and Resilient Infrastructure
• /
• v.9 no.4
• /
• pp.259-268
• /
• 2022

In this study, seeds of Festuca ovina var. coreana growing in waste coal landfills exposed to heavy metal contamination for a long time were collected, and growth characteristics and heavy metal accumulation capacity were evaluated through greenhouse cultivation experiments with germinated seedlings, and was conducted for the applicability of phytoremediation technology. Concentration gradients of arsenic-treated artificial soil were 25, 62.5, 125, and 250 mg/kg, respectively, lead concentrations were 200, 500, 1000, and 2000 mg/kg, and cadmium concentrations were 15, 30, 60, and 100 mg/kg, respectively In the arsenic, lead, and cadmium-treated experimental groups, the number of leaves of F. ovina var. coreana decreased in all compared to the control group except for the lead-treated groups (200, 500, and 1000 mg/kg). Length growth of the shoot part was increased in all of the arsenic treatment groups compared to the control group, but decreased in all of the root parts. In the 1000 and 2000 mg/kg lead treatment groups, lengths increased compared to the control group, but in the other treatments, they were shorter than the control group. In the case of the cadmium treatment group, all of the shoot parts were increased compared to the control group, and all of the root parts were decreased. In the case of arsenic treatment, the biomass was decreased at all parts and all concentrations compared to the control group. The 200, 500, and 1000 mg/kg lead treatments showed larger biomass than the control group in both shoot and root parts. In the cadmium treatment group, the biomass of both shoot and root parts decreased compared to the control group. As the concentration of heavy metal treatment increased, both the number of leaves and the biomass by plant parts tended to decrease, and the length growth of the shoot part tended to increase slightly, but the root part tended to decrease slightly. The arsenic accumulation concentrations of the shoot and root parts of the 62.5 mg/kg arsenic treatment area were 9.4 mg/kg and 253.3 mg/kg, respectively. While the shoot part of the 250 mg/kg arsenic treatment area withered away, the arsenic accumulation concentration in the root part was analyzed to be 859.1 mg/kg, In the 2,000 mg/kg lead treatment area, the shoot and root parts accumulated 10,308.1 and 11,012.0 mg/kg, which were 1.1 times higher than the root parts. At 100 mg/kg cadmium treatment, the shoot and root parts were 176.0 and 287.2 mg/kg, and the root part accumulated 1.6 times higher than the shoot part. As a result of tolerance evaluation of F. ovina var. coreana, multi-tolerance to three heavy metals was confirmed by maintaining growth without dying in all treatment groups of arsenic, lead, and cadmium. Plant extraction (phytoextraction) of F. ovina var. coreana was verified as a species that can be applied up to 2,000 mg/kg of soil lead contamination.

• Journal of Wetlands Research
• /
• v.25 no.3
• /
• pp.205-212
• /
• 2023

This study was attempted to find out how the ecological niche and interspecies relationship of Quercus aliena and Q. serrata, which are the main constituents of potential natural vegetation along the riverside of mountains in Korea, under climate change conditions. To this end, soil moisture and soil nutrients were treated with 4 grad ients under climate change conditions with elevated CO₂ and temperature, plants we re harvested at the end of the growing season, growth responses of traits were measured, ecological niche breadth and overlap were calculated, and it was compared with that of the control group(ambient condition). In addition, the relationship between the two species was analyzed by principal component analysis using trait values. As a result, the ecological niche breadth of Q. aliena was wider than that of Q. serrata under the moisture environment conditions under climate change. Under nutrient conditions, the ecological niche of the two species were similar. In addition, the ecological overlap for soil moisture of Q. aliena and Q. serrata was wider than the soil nutrient gradient under climate change. The species with traits in which the increase in ecological niche breadth due to climate change occurred more than the decrease was Q. aliena in both water and nutrient gradients. And in the responses of the population level, due to climate change, the adaptability of Q. aliena was higher than that of Q. serrata under the soil moisture condition, but the two species were similar under the nutrient condition. These results mean that the competition between the two species occurs more severely in the water environment under climate change conditions, and at that time, Q. aliena has higher adaptability than Q. serrata.