머신러닝 기반 산림 토양 pH의 디지털 토양도 작성 연구

심우진,안유순,정관용 국토지리학회 2024 국토지리학회지 Vol.58 No.1

한반도는 지질과 기후의 영향으로 산성 토양 비율이 높으며, 최근 대기오염물질에 의한 산성비의 유입으로 산림 토양의 산성화가 심해지고 있다. 본 연구에서는 산림 토양 특성의 지리적 분포를 예측할 수 있는 디지털 토양도 작성 기법을 적용하여 남한 전역의 산림 토양 pH 디지털 토양도를 구축하였다. 다양한 머신러닝 기법을 적용한 결과, 랜덤 포레스트 알고리즘이 가장 정확도가 높았다. 산림 토양 pH의 분포와 관련된 주요 환경 요인으로 지리변수, 지형변수, 모재변수가 도출되었다. 이 결과를 바탕으로 남한 일대의 산림 토양 pH 분포를 확인한 결과, 석회암과 현무암 등 염기성 모재가 분포한 지역에서 pH가 높게, 주로 편마암 모재가 주로 분포하는 지역에서 토양 pH가 낮게 나타났다. 이 중 강산성(pH 4.5 이하) 토양은 수도권과 충청권 서남부, 남해안 등에 주로 분포했다. 강산성 토양은 식생의 생육 환경과 탄소 격리 능력에 부정적인 영향을 미칠 수 있으므로 산성화를 가속화하는 인위적 요인을 저감하고 주기적인 모니터링을 통해 지속적으로 관리해야 할 것이다. The geology and climate of the Korean Peninsula contribute to a high prevalence of acidic soil, and this situation is exacerbated by the recent increase in acid rain resulting from air pollution, further worsening the acidification of forest soils. In this research, through digital soil mapping to predict the spatial distribution of forest soil properties, we developed digital soil maps of forest soil pH across South Korea. Upon employing various machine learning methods, the random forest algorithm emerged as the most accurate. Key environmental factors influencing the distribution of forest soil pH were identified, including geographic, topographic, and parent material variables. The spatial distribution of forest soil pH revealed higher pH values in areas with alkaline bedrock (limestone and basalt) and lower pH values in areas dominated by gneiss parent material. Notably, highly acidic soils (pH < 4.5) were found in the metropolitan area, southwestern Chungcheong region, and the southern coast. Strongly acidic soil can negatively affect vegetation growth environment and carbon sequestration capacity, so anthropogenic factors that accelerate acidification should be reduced and continuously managed through periodic monitoring.

토양정보를 이용한 우리나라 물 보유능 추정

홍석영 ( Suk-young Hong ),부디만미나스니 ( Budiman Minasny ),한경화,김이현 ( Yi-hyun Kim ),이경도 ( Kyung-do Lee ) 한국농공학회 2013 한국농공학회 학술대회초록집 Vol.2013 No.-

The knowledge on the spatial distribution of soil available water capacity at a regional or national extent is essential, as soil water capacity is a component of the water and energy balances in the terrestrial ecosystem. It controls the evapotranspiration rate, and has a major impact on climate. This paper demonstrates a protocol for mapping soil available water capacity in South Korea at a fine scale using data available from surveys. The procedures combined digital soil mapping technology with the available soil map of 1 : 25,000. We used the modal profile data from the Taxonomical Classification of Korean Soils. The data consist of profile description along with physical and chemical analysis for the modal profiles of the 380 soil series.However not all soil samples have measured bulk density and water content at -10 and -1500 kPa. Thus they need to be predicted using pedotransfer functions. Furthermore, water content at -10 kPa was measured using ground samples. Thus a correction factor is derived to take into account the effect of bulk density. Results showed that Andisols has the highest mean water storage capacity, followed by Entisols and Inceptisols which have loamy texture. The lowest water retention is Entisols which are dominated by sandy materials. Profile available water capacity to a depth of 1mwas calculated and mapped for Korea. The western part of the country shows higher available water capacity than the eastern part which is mountainous and has shallower soils. The highest water storage capacity soils are the Ultisols and Alfisols (mean of 206 and 205 mm, respectively). Validation of the maps showed promising results. The map produced can be used as an indication of soil physical quality of Korean soils.

스플라인 모형을 이용한 우리나라의 층위별 토성 지도 작성

홍석영 ( Sukyoung Hong ),( Budiman Minasny ),송관철 ( Kwancheol Song ),현병근 ( Byungkeun Hyun ),김이현 ( Yihyun Kim ),이경도 ( Kyungdo Lee ),이덕배 ( Deogbae Lee ) 한국농공학회 2012 한국농공학회 학술대회초록집 Vol.2012 No.-

Quantitative soil information is important, soil properties at different depths are very useful information to model water quantity & quality and crop growth. This paper estimates soil texture such as clay, sand, and silt content at the 6 standard depths using the equal-area smoothing spline functions. A Korean soil database of 380 soil series was compiled, which includes chemical and physical properties such as sand and clay content for the profiles of each soil series. We applied the equal-area spline depth functions for sand, silt and clay content. Derived mean values of sand, silt and clay content from the fitted spline at standard depths were mapped by corresponding soil series to produce rasterized spatial and depth-wise soil property maps. Mean values of clay content were 12.2% at 0-5 cm, 12.4% at 5-15 cm, 12.8% at 15-30 cm, 12.7% at 30-60 cm, and 11.2% at 60-100 cm, respectively. Clay content evenly distributed at different depths. Standard deviation from 0-5 cm to 30-60 cm showed similar values and increase abruptly at deeper than 60 cm. We also compared the results to recently collected soil samples to estimate the accuracy of the maps. Scatter plot of recently measured and predicted clay content showed good agreement.

Post-Agrogenic Self-Restoration of Soddy Podzolic Soils: Results and Methods N.P. Sorokina, D.N. Kozlov, I.B. Kuznetsova

Natalia Sorokina,Daniil Kozlov,Inna Kuznetsova 한국토양비료학회 2014 한국토양비료학회 학술발표회 초록집 Vol.2014 No.6

KOREAN SOILS AND INFORMATION

Hong, S.Y,Y.H. Kim,B.K. Hyun,K.C. Song,Y.J. Lee,M.S. Kim,J.S. Lee,Y.S. Zhang,K.H. Han,S.K. Ha,S.G. Yoon 한국토양비료학회 2012 한국토양비료학회 세미나 Vol.2012 No.6

Soil survey and information is of great importance for the use and conservation of soil resources that are essential for human welfare and ecosystem sustainability. This paper introduces soil survey and soil inventory of Korea focusing on national soil database, soil and environment information system, and use of soil database for natural resources management focusing on agricultural fields. Various scales of soil maps and soil testing data were established through a series of intensive National Soil Survey Projects and Soil Testing Projects conducted by Rural Development Administration (RDA) since early 1960s. An internet-based information system for soil and environment (http://soil.rda.go.kr) was developed on the basis of ‘National Soil Survey Projects’ and ‘Agro-environmental Change Monitoring Project’, which monitors spatial and temporal changes of agricultural environment. Soils data has a great potential of further application in estimation of soil carbon storage, water capacity, and soil loss. Digital mapping of soil and environment using state-of-the-art and emerging technologies with a pedometrics concept will lead to future direction.

Statistical Estimation of Soil Carbon Stocks in Chungcheong Province through Digital Soil Mapping and Multiple Linear Regression

강윤구,오택근,이재한,이준영 한국토양비료학회 2023 한국토양비료학회지 Vol.56 No.3

Digital soil mapping (DSM) is a statistical technique that utilizes soil characteristics and environmental factors to create spatial distribution maps representing soil properties. The SCORPAN model, consisting of soil (S), climate (C), organisms (O), relief (R), parent materials (P), age (A) and space (N), describes the environmental factors used in DSM techniques. The objectives of this study were to assess the spatial distribution map of soil carbon stocks in Chungcheong province and predict soil carbon stocks within the 0 - 30 cm depth using DSM technique. The minimum and maximum predicted carbon stocks were 25.11 ton C ha-1 and 183.55 ton C ha-1, respectively, with a mean of 46.92 ± 13.66 ton C ha-1. The spatial distribution map of soil carbon stocks revealed higher carbon stock in Chungcheongbuk-do, particularly in Danyang-gun, while lower carbon stocks were observed in the coastal areas of Chungcheongnam-do. The estimated economic value of soil carbon stocks in Chungcheong province was 406.3 billion won, based on the average soil carbon stock, agricultural land area and carbon offset trading price. The validation outcomes of the DSM are summarized as follows: the model achieved a coefficient of determination (R2) of 0.15, indicating the 15% confidence levels to the validation data. The mean absolute error (MAE) was 20.78, and the root mean square error (RMSE) was 29.51, respectively. The scatter plot between observed and predicted soil carbon stocks revealed that the predicted values were lower than the observed values, indicating a need for improvement in the model’s predictive performance. Therefore, the estimated soil carbon stocks and its spatial distribution map in this study can serve as fundamental information for assessing the potential carbon sequestration capacity of agricultural soils and contributing to climate change mitigation and carbon neutrality efforts.

농업 경관에서 전자토양도 기법을 이용한 토양탄소 저장량 예측 및 경제적 가치평가

정관용 국토지리학회 2018 국토지리학회지 Vol.52 No.3

Among soil ecosystem services, there is growing interest in soil carbon storage that provides climate regulation functions. This study estimated the amount of soil carbon stocks using digital soil mapping and the 1:25,000 soil map in an agricultural landscape and conducted economic evaluation of them. The random forest model (0.91kg/m2) showed better results than the method using the soil map (1.77kg/m2) based on the mean squared error. In estimating soil carbon stocks using random forest, important variables such as distance to the coast, elevation, silt content, distance to river, clay content, sand content and curvature were selected in order. The amount of soil carbon stocks in the cropland of Jindo was estimated to be about 370,000 tons (random forest) and about 490,000 tons (soil map). The value of soil carbon stocks in the cropland of Jindo considering the carbon market price was estimated to be about 8.1 billion won. This study is expected to be used as a basic study for decision making related to various soil environmental policies such as climate change and sustainable soil management. 토양 생태계서비스 중에서 기후변화를 완화할 수 있는 기후조절 기능을 제공하는 토양탄소 저장에 대한 관심이 높아지고 있다. 본 연구는 전자토양도 기법(랜덤포레스트)과 1:25,000 토양도를 이용하여 농경지의 토양탄소 저장량을 추정하고, 이에 대한 경제성 평가를 실시하였다. 모형의 검증결과에서 평균 제곱 오차를 기준으로 했을 때 랜덤포레스트 모형(0.91kg/m2)이 토양도를 이용한 예측방식(1.77kg/m2)보다 나은 결과를 보였다. 랜덤포레스트를 이용한 토양탄소 저장량 예측에서 해안과의 거리, 고도, 실트함량, 하천과의 거리, 점토함량, 모래함량, 곡면률 순으로 중요한 변수가 선택되었다. 진도군 농경지의 토양탄소 저장량은 두 방식을 적용했을 때 각각 약 37만 톤(랜덤포레스트)과 약 49만 톤(토양도)으로 추정된다. 탄소시장 가격을 고려한 진도군 농경지의 토양탄소 저장량의 가치는 약 81억 원으로 추정된다. 본 연구는 기후변화와 지속가능한 토양관리 등 다양한 토양환경 정책과 관련된 의사결정에 필요한 기초연구로 활용될 수 있을 것으로 기대된다.

Estimation of Soil Organic Carbon (SOC) Stock in South Korea Using Digital Soil Mapping Technique

Seong-Jin Park,Soon-Ik Kwon,Seong-Heon Kim,Jaehong Shim,Yun-Hae Lee,Taek-Keun Oh 한국토양비료학회 2021 한국토양비료학회지 Vol.54 No.2

Estimation of soil organic carbon (SOC) stocks is highly relevant considering that SOCs is the central driver in soil fertility and climate change mitigation. This study aims to (i) evaluate the SOC stock in the first 0 - 30 ㎝ and 0 - 100 ㎝ soil layer on a national scale from spatially explicit explanatory environmental variables and a legacy soil database and (ii) the spatial distribution of SOCs at national scale through digital mapping technique. A spatial model was established using Cubist, a decision tree algorithm and based on soil data (s factor), climatic (c factor), topographic (r factor). Results showed that soil texture, soil parent, mean annual precipitatio and elevation were the most important predictors of SOCs. The Cubist prediction model had a Root Mean Squared Error (RMSE) equal to 19.5 at 0 - 30 ㎝, 68.7 at 0 - 100 ㎝. The predicted mean SOC stock from fitted models was 35 ton C ㏊<SUP>-1</SUP> for 0 - 30 ㎝ depth, 87 ton C ㏊<SUP>-1</SUP> for 0 - 100 ㎝ soil depth. In total, soil stored approximately 330 Mt C for 0 - 30 ㎝ depth and 842 Mt C for 0 - 100 ㎝ depth.

현장 측정된 토양미생물 종 다양성과 생태자연도 등급 자료 간의 통계적 상관관계 평가와 토양생태의 질 산정방안 제안

기동원(Ki Dongwon),박준홍(Park Joonhong),이재진(Lee Jaejin),노백호(Rho Paikho) 대한토목학회 2007 대한토목학회논문집 B Vol.27 No.6B

국토개발사업의 사전 계획 과정에서 개발할 것인지 보전할 것인지에 대한 의사결정은 인간 활동에 영향을 줄 수 있는 편리함과 이익을 고려해야 할 뿐만 아니라, 자연환경생태에 미칠 수 있는 영향을 종합적으로 예측하고 평가할 수 있는 자료기반 및 통합기법을 요구한다. 그 과정에서 사전 평가가 신뢰성 있고 객관적으로 이루어지기 위해서는 자연생태의 질에 대한 과학적이고 정량적인 정보가 필요하다. 자연생태의 주요 구성요소 중 토양생태계는 생태 전반적인 중요성에도 불구하고 동식물생태에 비해 기초지료기반이 부족하며 토양생태질의 등급화에 대한 개념 설정도 미흡하다. 본 연구에서는 현재 환경영향평가 시 고려되고 있는 생태자연도의 자연생태 등급 자료가 토양생태의 질을 반영하고 있는지의 여부를 평가하였다. 이를 위해서 대상지역의 토양시료에 대한 토양미생물 종 다양성을 측정하고 기존의 생태자연도, 식생우수성, 생물다양성 등급자료와의 관계를 통계적으로 비교분석하였다. 또한 토양생태의 질을 정량적으로 평가할 수 있는 방법과 산간지역의 토양시료 채취 방법에 대하여 논의하였다. 그리고 본 연구의 결과를 토대로 기존의 국토환경생태관련 지형정보를 이용하여 미지의 지역에 대한 토양생태등급을 산정하고, 이를 이용해 토양생태수치지도를 작성하는 방법에 대하여 제안하였다. For improving sustainability of mega-construction projects, their impact on ecological consequences has to be predicted in a planning stage. For more reliable ecological impact evaluation, ecological quality has to be quantitatively classified first, and then an integrated information system with multiple-layered ecological and other environmental information databases is needed for efficiently handling and utilizing a huge amount of complex ecological and environmental information, Soil ecology is well known to have significant roles in maintaining the whole ecological systems in the Earth. However, its quality classification and assessment methodology has been poorly established in comparison to those for animal and plant ecosystems. In this study, a method for quantitatively classifying the quality of soil ecology using soil microbial diversity measurement was developed for mountain areas. When examining whether the soil ecology quality data have any correlation with the current ecological information including overall ecology quality index, animal/plant biodiversity, and forest health index, the current ecological information that are typically used in environmental impact assessment in our country was found to have a low power to reflect soil ecology quality. It may due to the fact that the current ecological information system does not include soil ecology quality information. This clearly suggests that an inclusion of soil ecology quality information is needed for more reliable environmental impact assessment. Based upon the findings from this work, an algorithm of estimating soil ecological quality using the existing nation-scale environmental and ecological geo-spatial information and its mapping method were proposed.

지형분석을 이용한 산지토양 탄소의 분포 예측과 불확실성

정관용 ( Gwanyong Jeong ) 한국지형학회 2016 한국지형학회지 Vol.23 No.3

Soil carbon(C) is an essential property for characterizing soil quality. Understanding spatial patterns of soil C is particularly limited for mountain areas. This study aims to predict the spatial pattern of soil C using terrain analysis in a steep mountainous area. Specifically, model performances and prediction uncertainties were investigated based on the number of resampling repetitions. Further, important predictors for soil C were also identified. Finally, the spatial distribution of uncertainty was analyzed. A total of 91 soil samples were collected via conditioned latin hypercube sampling and a digital soil C map was developed using support vector regression which is one of the powerful machine learning methods. Results showed that there were no distinct differences of model performances depending on the number of repetitions except for 10-fold cross validation. For soil C, elevation and surface curvature were selected as important predictors by recursive feature elimination. Soil C showed higher values in higher elevation and concave slopes. The spatial pattern of soil C might possibly reflect lateral movement of water and materials along the surface configuration of the study area. The higher values of uncertainty in higher elevation and concave slopes might be related to geomorphological characteristics of the research area and the sampling design. This study is believed to provide a better understanding of the relationship between geomorphology and soil C in the mountainous ecosystem.