• Korean Journal of Remote Sensing
• /
• v.37 no.5_3
• /
• pp.1405-1423
• /
• 2021

Precipitation is one of the main factors that affect water and energy cycles, and its estimation plays a very important role in securing water resources and timely responding to water disasters. Satellite-based quantitative precipitation estimation (QPE) has the advantage of covering large areas at high spatiotemporal resolution. In this study, machine learning-based rainfall intensity models were developed using Himawari-8 Advanced Himawari Imager (AHI) water vapor channel (6.7 ㎛), infrared channel (10.8 ㎛), and weather radar Column Max (CMAX) composite data based on random forest (RF). The target variables were weather radar reflectivity (dBZ) and rainfall intensity (mm/hr) converted by the Z-R relationship. The results showed that the model which learned CMAX reflectivity produced the Critical Success Index (CSI) of 0.34 and the Mean-Absolute-Error (MAE) of 4.82 mm/hr. When compared to the GeoKompsat-2 and Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks (PERSIANN)-Cloud Classification System (CCS) rainfall intensity products, the accuracies improved by 21.73% and 10.81% for CSI, and 31.33% and 23.49% for MAE, respectively. The spatial distribution of the estimated rainfall intensity was much more similar to the radar data than the existing products.

• Korean Journal of Remote Sensing
• /
• v.34 no.6_2
• /
• pp.1261-1272
• /
• 2018

It is important to measure the Arctic surface air temperature because it plays a key-role in the exchange of energy between the ocean, sea ice, and the atmosphere. Although in-situ observations provide accurate measurements of air temperature, they are spatially limited to show the distribution of Arctic surface air temperature. In this study, we proposed machine learning-based models to estimate the Arctic surface air temperature in summer based on buoy data and Advanced Microwave Scanning Radiometer 2 (AMSR2)satellite data. Two machine learning approaches-random forest (RF) and support vector machine (SVM)-were used to estimate the air temperature twice a day according to AMSR2 observation time. Both RF and SVM showed $R^2$ of 0.84-0.88 and RMSE of $1.31-1.53^{\circ}C$. The results were compared to the surface air temperature and spatial distribution of the ERA-Interim reanalysis data from the European Center for Medium-Range Weather Forecasts (ECMWF). They tended to underestimate the Barents Sea, the Kara Sea, and the Baffin Bay region where no IABP buoy observations exist. This study showed both possibility and limitations of the empirical estimation of Arctic surface temperature using AMSR2 data.

• Korean Journal of Remote Sensing
• /
• v.37 no.3
• /
• pp.503-514
• /
• 2021

Rice production with adequate level of area is important for decision making of rice supply and demand policy. It is essential to grasp rice cultivation areas in advance for estimating rice production of the year. This study was carried out to classify paddy rice cultivation in Gimje-si using sentinel-1 SAR (synthetic aperture radar) and UAV imagery in early July. Time-series Sentinel-1A and 1B images acquired from early May to early July were processed to convert into sigma naught (dB) images using SNAP (SeNtinel application platform, Version 8.0) toolbox provided by European Space Agency. Farm map and parcel map, which are spatial data of vector polygon, were used to stratify paddy field population for classifying rice paddy cultivation. To distinguish paddy rice from other crops grown in the paddy fields, we used the decision tree method using threshold levels and random forest model. Random forest model, trained by mainly rice cultivation area and rice and soybean cultivation area in UAV image area, showed the best performance as overall accuracy 89.9%, Kappa coefficient 0.774. Through this, we were able to confirm the possibility of early estimation of rice cultivation area in Gimje-si using UAV image.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_3
• /
• pp.891-907
• /
• 2023

Compact Advanced Satellite 500-4 (CAS500-4) is scheduled to be launched to collect high spatial resolution data focusing on vegetation applications. To achieve this goal, accurate surface reflectance retrieval through atmospheric correction is crucial. Therefore, a machine learning-based atmospheric correction algorithm was developed to simulate atmospheric correction from a radiative transfer model using Sentinel-2 data that have similarspectral characteristics as CAS500-4. The algorithm was then evaluated mainly for forest areas. Utilizing the atmospheric correction parameters extracted from Sentinel-2 and GEOKOMPSAT-2A (GK-2A), the atmospheric correction algorithm was developed based on Random Forest and Light Gradient Boosting Machine (LGBM). Between the two machine learning techniques, LGBM performed better when considering both accuracy and efficiency. Except for one station, the results had a correlation coefficient of more than 0.91 and well-reflected temporal variations of the Normalized Difference Vegetation Index (i.e., vegetation phenology). GK-2A provides Aerosol Optical Depth (AOD) and water vapor, which are essential parameters for atmospheric correction, but additional processing should be required in the future to mitigate the problem caused by their many missing values. This study provided the basis for the atmospheric correction of CAS500-4 by developing a machine learning-based atmospheric correction simulation algorithm.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.18 no.4
• /
• pp.81-99
• /
• 2015

The acceleration of global warming has required better understanding of carbon cycles over local and regional areas such as the Korean peninsula. Since forests serve as a carbon sink, which stores a large amount of terrestrial carbon, there has been a demand to accurately estimate such forest carbon sequestration. In Korea, the National Forest Inventory(NFI) has been used to estimate the forest carbon stocks based on the amount of growing stocks per hectare measured at sampled location. However, as such data are based on point(i.e., plot) measurements, it is difficult to identify spatial distribution of forest carbon stocks. This study focuses on urban areas, which have limited number of NFI samples and have shown rapid land cover change, to estimate grid-based forest carbon stocks based on UNFCCC Approach 3 and Tier 3. Land cover change and forest carbon stocks were estimated using Landsat 5 TM data acquired in 1991, 1992, 2010, and 2011, high resolution airborne images, and the 3rd, 5th~6th NFI data. Machine learning techniques(i.e., random forest and support vector machines/regression) were used for land cover change classification and forest carbon stock estimation. Forest carbon stocks were estimated using reflectance, band ratios, vegetation indices, and topographical indices. Results showed that 33.23tonC/ha of carbon was sequestrated on the unchanged forest areas between 1991 and 2010, while 36.83 tonC/ha of carbon was sequestrated on the areas changed from other land-use types to forests. A total of 7.35 tonC/ha of carbon was released on the areas changed from forests to other land-use types. This study was a good chance to understand the quantitative forest carbon stock change according to the land cover change. Moreover the result of this study can contribute to the effective forest management.

• Korean Journal of Remote Sensing
• /
• v.39 no.6_2
• /
• pp.1651-1669
• /
• 2023

Sea ice currently covers approximately 7% of the world's ocean area, primarily concentrated in polar and high-altitude regions, subject to seasonal and annual variations. It is very important to analyze the area and type classification of sea ice through time series monitoring because sea ice is formed in various types on a large spatial scale, and oil and gas exploration and other marine activities are rapidly increasing. Currently, research on the type and area of sea ice is being conducted based on high-resolution satellite images and field measurement data, but there is a limit to sea ice monitoring by acquiring field measurement data. High-resolution optical satellite images can visually detect and identify types of sea ice in a wide range and can compensate for gaps in sea ice monitoring using Geostationary Ocean Color Imager-II (GOCI-II), an ocean satellite with short time resolution. This study tried to find out the possibility of utilizing sea ice monitoring by training a rule-based machine learning model based on learning data produced using high-resolution optical satellite images and performing detection on GOCI-II images. Learning materials were extracted from Liaodong Bay in the Bohai Sea from 2021 to 2022, and a Random Forest (RF) model using GOCI-II was constructed to compare qualitative and quantitative with sea ice areas obtained from existing normalized difference snow index (NDSI) based and high-resolution satellite images. Unlike NDSI index-based results, which underestimated the sea ice area, this study detected relatively detailed sea ice areas and confirmed that sea ice can be classified by type, enabling sea ice monitoring. If the accuracy of the detection model is improved through the construction of continuous learning materials and influencing factors on sea ice formation in the future, it is expected that it can be used in the field of sea ice monitoring in high-altitude ocean areas.

• Journal of Digital Convergence
• /
• v.18 no.4
• /
• pp.243-250
• /
• 2020

Systematic farming can be planned and managed if long-term agricultural weather information of the plantation is available. Because the greatest risk factor for crop cultivation is the weather. In this study, a method for long-term predicting of agricultural weather using the GloSea5 and machine learning is presented for the cultivation of Chinese cabbage. The GloSea5 is a long-term weather forecast that is available up to 240 days. The deep neural networks and the spatial randomforest were considered as the method of machine learning. The longterm prediction performance of the deep neural networks was slightly better than the spatial randomforest in the sense of root mean squared error and mean absolute error. However, the spatial randomforest has the advantage of predicting temperatures with a global model, which reduces the computation time.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.23 no.6
• /
• pp.545-550
• /
• 2013

In pattern recognition field, data classification is an essential process for extracting meaningful information from data. Adaptive boosting algorithm, known as AdaBoost algorithm, is a kind of improved boosting algorithm for applying to real data analysis. It consists of weak classifiers, such as random guessing or random forest, which performance is slightly more than 50% and weights for combining the classifiers. And a strong classifier is created with the weak classifiers and the weights. In this paper, a research is performed using AdaBoost algorithm for detecting chaff echo which has similar characteristics to precipitation echo and interrupts weather forecasting. The entire process for implementing chaff echo classifier starts spatial and temporal clustering based on similarity with weather radar data. With them, learning data set is prepared that separated chaff echo and non-chaff echo, and the AdaBoost classifier is generated as a result. For verifying the classifier, actual chaff echo appearance case is applied, and it is confirmed that the classifier can distinguish chaff echo efficiently.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.37 no.6
• /
• pp.405-416
• /
• 2019

Due to the development of information and communication technology, the production and processing speed of data is getting faster. To classify objects using machine learning, which is a field of artificial intelligence, data required for training can be easily collected due to the development of internet and geospatial information technology. In the field of geospatial information, machine learning is also being applied to classify or recognize objects using images and point clouds. In this study, the problem of manually constructing training data using existing digital map version 1.0 was improved, and the technique of classifying roads, buildings and vegetation using image and point clouds were proposed. Through experiments, it was possible to classify roads, buildings, and vegetation that could clearly distinguish colors when using true ortho-image with only RGB (Red, Green, Blue) bands. However, if the colors of the objects to be classified are similar, it was possible to identify the limitations of poor classification of the objects. To improve the limitations, random forest and support vector machine techniques were applied after band fusion of true ortho-image and normalized digital surface model, and roads, buildings, and vegetation were classified with more than 85% accuracy.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.38 no.1
• /
• pp.23-33
• /
• 2020

Recently, high-resolution images can be easily acquired using UAV (Unmanned Aerial Vehicle), so that it is possible to produce small area observation and spatial information at low cost. In particular, research on the generation of cover maps in crop production areas is being actively conducted for monitoring the agricultural environment. As a result of comparing classification performance by applying RF(Random Forest), SVM(Support Vector Machine) and CNN(Convolutional Neural Network), deep learning classification method has many advantages in image classification. In particular, land cover classification using satellite images has the advantage of accuracy and time of classification using satellite image data set and pre-trained parameters. However, UAV images have different characteristics such as satellite images and spatial resolution, which makes it difficult to apply them. In order to solve this problem, we conducted a study on the application of deep learning algorithms that can be used for analyzing agricultural lands where UAV data sets and small-scale composite cover exist in Korea. In this study, we applied DeepLab V3 +, FC-DenseNet (Fully Convolutional DenseNets) and FRRN-B (Full-Resolution Residual Networks), the semantic image classification of the state-of-art algorithm, to UAV data set. As a result, DeepLab V3 + and FC-DenseNet have an overall accuracy of 97% and a Kappa coefficient of 0.92, which is higher than the conventional classification. The applicability of the cover classification using UAV images of small areas is shown.