Remote sensing, numerical models, and machine learning have been widely used for investigating environmental risks under climate change. It is known that they tend to do an excellent job in mapping, simulating, and projecting the long-term changes in average conditions. However, damages associated with extreme weathers by droughts, floods, forest fires, heat-related mortality, and crop yield loss are often more devastating than those caused by gradual climate changes. How remote sensing, numerical models, and machine learning can be used for assessing the impacts of extreme weathers on the natural and human systems remains uncertain.
This session aims to summarize current progress in assessing the ability of remote sensing, numerical models, and machine learning for quantifying climate risks in multiple sectors, such as water, agriculture, and human health.
We especially welcome investigations focusing on the inter-comparison of methodologies, as well as multi-sectoral, cross-sectoral, and integrated assessments.

Co-organized by CL2/ESSI1/NH6
Convener: Guoyong LengECSECS | Co-conveners: Jian Peng, Shengzhi Huang, Zheng DuanECSECS, Shiqiang Zhang
| Attendance Mon, 04 May, 14:00–15:45 (CEST)

Files for download

Download all presentations (8MB)

Chat time: Monday, 4 May 2020, 14:00–15:45

Chairperson: Guoyong Leng
D795 |
Sara Santamaria Aguilar and Thomas Wahl

Future changes in the wind wave climate due to atmospheric changes can intensify present erosion and flood risk. Knowledge on both mean and extreme wave climate is necessary for understanding changes in sediment dynamics and flood events at the coastline. In order to assess potential wave changes, ensemble nearshore wave projections are required for covering   the entire range of wave conditions and also the large uncertainties related to future climate states. However, nearshore wave projections are not available for most coastal regions due to the excessive computational effort required for dynamically downscaling ensemble offshore wave data. As a result, the large relative contribution of waves to coastal flooding and erosion is commonly omitted in the assessment of those hazards. In this context, machine learning models can be an efficient tool for downscaling ensemble global wave projections if they are able to accurately simulate the non-linear processes of wave propagation due to their low computational requirements. Here, we analyse the performance of three machine learning methods, namely random forest, multivariate adaptive regression splines and artificial neural networks, for downscaling the wave climate along the coast of Florida. We further compare the performance of these three models to the multiple linear regression, which is a statistical model frequently used, although it does not account for the non-linearities associated with wave propagation processes. We find that the three machine learning models perform better than the multiple linear regression for all wave parameters (significant wave height, peak and mean periods, direction) along the entire coastline of Florida, which highlights the ability of these models to reproduce the non-linear wave propagation processes. Specifically, random forest shows the best performance and the lowest computational training times. In addition, this model shows a remarkably good performance in simulating the wave extreme events compared to the other models. By following a tree bagging approach, random forest can also provide confidence intervals and reduce the tuning process. The latter is one of the main disadvantages of the artificial neural networks, which also show a high performance for wave downscaling but require more training and tuning effort. Although the significant wave height and the periods can be simulated with very high accuracy (R2 higher than 0.9 and 0.8 respectively), the wave direction is poorly simulated by all models due to its circular behaviour. We find that a transformation of the direction into sine and cosine can improve the model performance. Finally, we downscale an ensemble of global wave projections along the coast of Florida and assess potential changes in the wave climate of this region.   

How to cite: Santamaria Aguilar, S. and Wahl, T.: Wave downscaling using machine learning, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-553, https://doi.org/10.5194/egusphere-egu2020-553, 2020.

D796 |
| Highlight
Shengzhi Huang, Jing Zhao, and Kang Ren

The Budyko curve is an effective tool for estimating how precipitation (P) partition into evapotranspiration (E) and streamflow (Q). Controlling the shape of the Budyko curve, the Budyko parameter represents the superimposed impact of various periodic factors (including climatic factors, catchment characteristics, teleconnection factors and anthropogenic activities) on the watershed water-energy balance dynamics, and such superimposed impact is not conducive to identifying the driving factors of the dynamic change of Budyko parameter at different time scales, and thus affect the parameter estimation. Here we obtain the dynamic change of Budyko parameter for the Wei River Basin (WRB)-a typical Loess Plateau region in China based on a 11-years moving window, and then adopt the Empirical Mode Decomposition (EMD) method to reveal the relationships between influencing factors and Budyko parameter series at multiple time scales by considering the interplay among different influencing factors. Results indicate that (1) Budyko parameter series are decomposed into 4-, 12-, 20-, exceeding 20-year time scale oscillations and a residual component with an significantly increasing trend in the upstream of the WRB (UWR) and the middle and lower reaches of the WRB (MDWR), a non-significantly decreasing trend in the Jing River Basin (JRB) and Beiluo River Basin (BLRB); (2) by analyzing the residual trend component, evaporation ratio (E/P), soil moisture (SM) and effective irrigated area (EIA) are found to induce the significant increase of parameter in the UWR, whereas that in the MDWR is dominated by baseflow (BF) and Niño 3.4; (3) parameter dynamics at the 4-year time scale is dominated by E/P, aridity index (EP/P), BF and SM; BF, PDO and sunspots attribute to the dynamics at 12-year time scale; all the factors except BF and SM contribute to the dynamics at 20- or exceeding 20-year time scales. The results of this study will help identify the connection between watershed water-energy balance dynamics and changing environment at multiple time scales, and also be beneficial for guiding water resources management and ecological development planning on the Loess Plateau region.

How to cite: Huang, S., Zhao, J., and Ren, K.: Watershed water-energy balance dynamics and their association with diverse influencing factors at multiple time scales, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-12162, https://doi.org/10.5194/egusphere-egu2020-12162, 2020.

D797 |
Gohar Ghazaryan, Sergii Skakun, Simon König, Ehsan Eyshi Rezaei, Stefan Siebert, and Olena Dubovyk

Timely monitoring of agricultural production and early yield predictions are essential for food security. Crop growth conditions and yield are related to climate variability and extreme events. Remotely sensed time-series can be used to study the variability in crop growth and agricultural production. However, the choice of remotely sensed data and methods is still an issue, as different datasets have different spatiotemporal characteristics. Thus, our primary goal was to study the impact of applying different remotely sensed time series on yield estimation in U.S. at the county and field scale. Furthermore, the impact of crop growth conditions on yield variability was assessed. For county-level analysis, MODIS-based surface reflectance, Land Surface Temperature, and Evapotranspiration time series were used as input datasets. Whereas field-level analysis was carried out using NASA’s Harmonized Landsat Sentinel-2 (HLS) product. 3D convolutional neural network (CNN) and CNN followed by long-short term memory (LSTM) were used. For county-level analysis, the CNN-LSTM model had the highest accuracy, with a mean percentage error of 10.3% for maize and 9.6% for soybean. This model presented robust results for the year 2012, which is considered a drought year. In the case of field-level analysis, all models achieved accurate results with R2 exceeding 0.8 when data from mid growing season were used. The results highlight the potential of yield estimation at different management scales.

How to cite: Ghazaryan, G., Skakun, S., König, S., Eyshi Rezaei, E., Siebert, S., and Dubovyk, O.: Crop Yield Estimation Using Multi-source Satellite Image Series and Deep Learning, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-13957, https://doi.org/10.5194/egusphere-egu2020-13957, 2020.

D798 |
| Highlight
Su-Bin Cho and Yang-Won Lee

Climate change is an important factor in crop growth, and it is significant to understand the relationship between climate change and rice yield. This study used annual rice yield from the USDA(United States Department of Agriculture) for each of China’s 16 administrative regions from 1979 to 2009, as well as average climate data from July to August, which were meteorological observations collected from the CRU(Climate Research Unit). The relationship between selected rice yield and climate change was nonlinear and modelled using a deep neural network to train even rows and verify odd rows of data. This study is expected to contribute to better food self-sufficiency and forecast future grain yields in China.

How to cite: Cho, S.-B. and Lee, Y.-W.: Analysis of Rice Yield in CHINA by Climate Change using Deep Neural Network, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-15701, https://doi.org/10.5194/egusphere-egu2020-15701, 2020.

D799 |
Robert Behling, Sigrid Roessner, and Saskia Foerster

One of the consequences of global climate change is the more frequent occurrence of extreme weather conditions. Semi-arid regions are especially vulnerable since evapotranspiration significantly exceeds precipitation for most of the year and rainfall occurrence is dominantly sporadic and highly variable in amount and spatial extent. Consequently, these regions suffer from droughts of increasing duration and severity, occasionally interrupted by strong rainfall events generating high surface runoff and in part highly destructive floods. In semi-arid regions water retention capability is often further reduced by changes of the original vegetation cover due to conversion into farmland and intensification of land use. The result is widespread land degradation by a decrease in permanent vegetation cover and an increase in soil erosion. Under such conditions sustainable water resources management is of key importance, however, reliable long-term observations describing the water cycle and the resulting water budget are missing for many regions of the world. This situation requires new approaches in improving seasonal forecast for relevant water resources parameters as well as spatiotemporally explicit understanding the of influence of water and land use management on the long-term development of water availability and land surface conditions. 
The German collaborative research project ‘Seasonal water resources management in semi-arid regions: Transfer of regionalized global information to practice’ (SaWaM) aims at the development of methods allowing the use of global data for deriving information needed for regional water resources management in semi-arid regions by integrating meteorological, hydrological and ecosystem sciences and supported by satellite remote sensing analysis. The performance, practical applicability and transferability of the developed methods are assessed in several semi-arid regions including Brazil, Iran and Sudan. Here, we present our work on the analysis of the seasonal and long-term vegetation dynamics at different spatial and temporal scales using satellite time series data of different spatial and temporal resolution (MODIS and Sentinel-2).  Our goal is linking the derived vegetation dynamics to changes in meteorological conditions, water availability and land use. In this context we put emphasis on the spatiotemporal analysis of bioproductivity related to different land use types and climatic conditions to identify and characterize hotspots of water usage in form of irrigated agriculture as a basis for further evaluation of the underlying water management practices.
We perform time series analysis of satellite-derived vegetation indices (VI) using various statistical aggregates, such as maximum, mean and temporal duration related to variable time periods (hydrological year, dry and wet season, growing patterns) as well as additive time series decomposition. Thus, we analyze long-term trends, seasonal deviations from long-term average conditions, and break points in the time series related to land use and water management changes. Moreover, we compare the derived spatiotemporal VI dynamics against the dynamics of hydrometeorological conditions (e.g. precipitation, evapotranspiration, temperature) as well as land use patterns in order to evaluate the impact of hydrometeorological drought conditions on different land use types and water management practices.  In conclusion, we present prototypes for information products supporting decision making of the local experts in the target regions.

How to cite: Behling, R., Roessner, S., and Foerster, S.: Satellite time series analysis of vegetation dynamics for water resources management in semi-arid regions, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-16520, https://doi.org/10.5194/egusphere-egu2020-16520, 2020.

D800 |
Tran Vu La, Christophe Messager, Rémi Sahl, and Marc Honnorat

Thanks to the geostationary meteorological satellites of METEOSAT (Europe), GOES (USA), and Himawari (Japan), the nowcast of convective systems (CS) can be performed in most of the world with a 5-15-minute observation time sampling and about 2.8-km spatial resolution (up to about 1-km for the new-generation satellites).

However, the CS forecast, including the prediction of their effects on the surface, is still a challenge due to the lack of high-resolution radar data and a deep understanding of this topic. Indeed, for now, most numerical weather prediction (NWP) models cannot deliver an accurate time and space estimation of surface wind patterns and wind gusts associated with the CS.

In the meantime, Synthetic Aperture Radar (SAR) and ASCAT (scatterometers) may be used for the detection of surface wind patterns potentially associated with the deep convective clouds that may be identified on METEOSAT images. Additionally, the intensity of wind patterns may be estimated from SAR and ASCAT data. Based on this result, Deep Learning (or Machine Learning) is proposed in an ongoing study for improving the predictions of wind gusts, based on the combination of several data sources such as SAR, ASCAT, METEOSAT. The obtained results of this step will be used to integrate into the current NWP models.

How to cite: La, T. V., Messager, C., Sahl, R., and Honnorat, M.: Artificial Intelligence for Improvement of Convective System Tracking and Its Surface Effect Prediction, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-17825, https://doi.org/10.5194/egusphere-egu2020-17825, 2020.

D801 |
Anita Bayer, Christine Mihalyfi-Dean, Robert Behling, Christof Lorenz, Saskia Foerster, Sigrid Roessner, and Almut Arneth

Semi-arid areas suffer from small amounts and a large variability in rainfall combined with an increasing risk of droughts under climate change. These long and short-term changes in water availability directly affecting regional livelihoods are depicted in the condition of the rather sparse vegetation. In this study, seasonal and long-term trends in indicators of the vegetation condition related to water availability and droughts (NDVI vs. fAPAR, NPP, soil water content, excess water) are identified from remote sensing data (MODIS) and a process-based dynamic vegetation model (LPJ-GUESS) for at least two semi-arid river basins. Identified trends of both methods are compared and evaluated based on the underlying processes and related to knowledge of past drought events. Finally, we answer the question, which methods and indicators are suitable to identify changes in the vegetation condition preceding a drought and during drought phases considering the methods and indicators as above plus simple precipitation-based drought indicators (e.g. standardized precipitation index, SPI) and enhanced drought indicators applying multiple indicators theirselves (e.g. combined drought indicator, CDI). The study is imbedded in the SaWaM project (Seasonal Water Management for semi-arid areas) and contributes to improved water management in the project regions by the integrated analysis of remote sensing and ecosystem modelling results that are made available to regional stakeholders tasked with water management in an online tool .

How to cite: Bayer, A., Mihalyfi-Dean, C., Behling, R., Lorenz, C., Foerster, S., Roessner, S., and Arneth, A.: Identification of droughts from monitored and modelled vegetation condition for improved water management in semi-arid areas, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-18580, https://doi.org/10.5194/egusphere-egu2020-18580, 2020.

D802 |
| Highlight
Jian Peng, Simon Dadson, Feyera Hirpa, Ellen Dyer, Thomas Lees, Diego G. Miralles, Sergio M. Vicente-Serrano, and Chris Funk

Droughts in Africa cause severe impacts such as crop failure, food shortages, famine, epidemics and even mass migration. To minimize the effects of drought on water and food security over Africa, a high-resolution drought dataset is essential to establish robust drought hazard probabilities and to assess drought vulnerability considering a multi- and cross-sectorial perspective that includes crops, hydrological systems, rangeland, and environmental systems. Such assessments are essential for policy makers, their advisors, and other stakeholders to respond to the pressing humanitarian issues caused by these environmental hazards.  In this study, a high spatial resolution Standardized Precipitation-Evapotranspiration Index (SPEI) drought dataset is presented to support these assessments. We compute historical SPEI data based on Climate Hazards group InfraRed Precipitation with Station data (CHIRPS) precipitation estimates and Global Land Evaporation Amsterdam Model (GLEAM) potential evaporation estimates. The high resolution SPEI dataset (SPEI-HR) presented here spans from 1981 to 2016 (36 years) with 5 km spatial resolution over the continent of Africa. To facilitate the diagnosis of droughts of different duration, accumulation periods from 1 to 48 months are provided. The quality of the resulting dataset was compared with coarse-resolution SPEI based on Climatic Research Unit (CRU) Time-Series (TS) datasets, and Normalized Difference Vegetation Index (NDVI) calculated from the Global Inventory Monitoring and Modeling System (GIMMS) project, as well as with root zone soil moisture modelled by GLEAM. Agreement between the coarse resolution SPEI from CRU TS (SPEI-CRU) and the developed SPEI-HR provides confidence in the estimation of temporal and spatial variability of droughts in Africa with SPEI-HR. In addition, agreement of SPEI-HR versus NDVI and root zone soil moisture – with average correlation coefficient (R) of 0.54 and 0.77, respectively – further suggests that SPEI-HR can provide valuable information to study drought-related ecological and societal impacts at sub-basin and district scales in Africa. The dataset is archived in Centre for Environmental Data Analysis (CEDA), with link: http://dx.doi.org/10.5285/bbdfd09a04304158b366777eba0d2aeb (Peng et al., 2019). 


Peng, J.; Dadson, S.; Hirpa, F.; Dyer, E.; Lees, T.; Miralles, D.G.; Vicente-Serrano, S.M.V.-S.; Funk, C. (2019): High resolution Standardized Precipitation Evapotranspiration Index (SPEI) dataset for Africa. Centre for Environmental Data Analysis, 05 August 2019. doi:10.5285/bbdfd09a04304158b366777eba0d2aeb.

How to cite: Peng, J., Dadson, S., Hirpa, F., Dyer, E., Lees, T., Miralles, D. G., Vicente-Serrano, S. M., and Funk, C.: A pan-African high-resolution drought index dataset, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-18591, https://doi.org/10.5194/egusphere-egu2020-18591, 2020.

D803 |
José Vaz, Célia M. Gouveia, and Isabel F. Trigo

Understanding climate variability and change and its impacts on natural systems is becoming more and more important as changes in earth surface condition near surface air temperature and precipitation. Over Portugal, the observed warming trends have been found to be asymmetric with respect to seasonal and diurnal cycles, with greatest warming occurring for the minimum temperature and during winter and spring. These observed trends exert strong influences on agriculture systems, affecting production viability through changes in winter hardening, frost occurrence, growing season lengths and heat accumulation for ripening potential.

Remote sensing technology has been developing steadily and its products can provide many applications in agriculture, namely crop identification, crop growth monitoring and yield prediction. Recently the LSA SAF team set up a strategy to generate long term data records from Meteosat Second Generation satellite series (2004 to present), releasing Land Surface Temperature (LST), Reference Evapotranspiration (ETREF) and Vegetation parameters (FAPAR, LAI and FVC) using a stable set of input data and algorithm, which would be suitable for climate variability and change detection studies. On the other hand, a new product to characterize the ecosystem processes, the Gross Primary Production (GPP), is under production since 2018.

In this work we propose to computed Water Use Efficiency (WUE), as the ratio between Gross Primary Production (GPP) and Reference Evapotranspiration (ETREF), using LSA-SAF Products. WUE translates the exchanges of carbon and water gross fluxes, between natural ecosystem and the atmosphere, allowing to monitor the adaptability of the ecosystems to climate change. The role played by Evapotranspiration and Water Use Efficiency for different crops in Portugal is evaluated, namely on Wine Production for Douro Region. Results for 2018 and 2019 highlights the vulnerability of the different sectors of Douro Region to dry and wet conditions, namely helping to analyze the impact of droughts on Douro wine production.

Acknowledgements: This study was performed within the framework of the LSA-SAF, co-funded by EUMETSAT This work was partially supported by national funds through FCT (Fundação para a Ciência e a Tecnologia, Portugal) under projects CLMALERT (ERA4CS/0005/2016).

How to cite: Vaz, J., Gouveia, C. M., and Trigo, I. F.: The role of Evapotranspiration and Water Use Efficiency in Agriculture in Portugal , EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-19248, https://doi.org/10.5194/egusphere-egu2020-19248, 2020.

D804 |
| Highlight
Elisavet Parselia, Charalambos Kontoes, Ioannis Kioutsioukis, Spiros Mourelatos, Christos Hadjichristodoulou, and Nikolaos I. Stilianakis

The aim of this study is the development of an operational Early Warning System (EWS) that will utilize new and enhanced satellite Earth Observation (EO) sensors with the purpose of forecasting and risk mapping the West Nile Virus (WNV) outbreaks. Satellite EO data were leveraged to estimate environmental variables that influence the transmission cycle of the pathogen that leads to WNV, a mosquito-borne disease (MBD). The system was trained with epidemiological and entomological data from the region of Central Macedonia, the most epidemic-prone region in Greece regarding the WNV. The satellite derived environmental parameters of the Normalized Difference Vegetation Index (NDVI), the Normalized Difference Water Index (NDWI), the Land Surface Temperature (LST), precipitation data as well as proximity to water bodies were coupled with meteorological data and were used as explanatory variables for the models. The management and analysis of the big satellite data was conducted with the Open Data Cube (ODC), providing an open and freely accessible exploitation architecture. Statistical and machine learning algorithms were used for short-term forecast, while dynamical models were utilized for the seasonal forecast.The system explores the analysis of big satellite data and proves its scalability by replicating the same models in different geographic regions; e.g the northeastern Italian region of Veneto. This EWS will be used as a tool for helping local decision-makers to improve health system responses, take preventive measures in order to curtail the spread of WNV in Europe and address the relevant priorities of the Sustainable Development Goals (SDGs) such as good health and well-being (SDG 3) and climate action (SDG 13).

How to cite: Parselia, E., Kontoes, C., Kioutsioukis, I., Mourelatos, S., Hadjichristodoulou, C., and Stilianakis, N. I.: Early Warning System for West Nile Virus outbreaks based on Satellite Earth Observation Data , EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20612, https://doi.org/10.5194/egusphere-egu2020-20612, 2020.

D805 |
Muqi Xiong

Water-driven soil erosion is the most widespread form of soil degradation worldwide, which threatens to the sustainability of agriculture. Climate change may aggravate the threat of erosion. On the basis of the Revised Universal Soil Loss Equation, combined with Geographic Information Systems (GIS), we assessed spatiotemporal variances in global water erosion risk trends during the period 1992–2015 using the linear regression model. The research objective was to explore the spatial pattern of global water erosion risk change in recent decades and to identify the driving factors. The results show that the global water erosion risk increased over 54% of the surface during 1992–2015, with an average rate of 0.17 t·ha-1·yr-2. The lands with significant increasing trends (p < 0.05) accounted for 12% of global lands, with an average rate of 0.27 t·ha-1·yr-2. In which, over 75% regions with significant increasing trends were croplands and forest lands in the cold climate zone as the rainfall intensity increased. However, the increasing rates of soil erosion risk on bare lands and croplands were extremely larger than that on lands with natural vegetation, which means that water erosion on natural lands had much lower sensitive to rainfall changes. These results suggest that improving vegetation conditions in the region with sensitive climate change could reduce the erosion threat.

How to cite: Xiong, M.: Global water soil erosion risk associated with climate change, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20942, https://doi.org/10.5194/egusphere-egu2020-20942, 2020.

D806 |
| Highlight
Xiaomeng Yin and Guoyong Leng

Understanding historical crop yield response to climate change is critical for projecting future climate change impacts on yields. Previous assessments rely on statistical or process-based crop models, but each has its own strength and weakness. A comprehensive comparison of climate impacts on yield between the two approaches allows for evaluation of the uncertainties in future yield projections. Here we assess the impacts of historical climate change on global maize yield for the period 1980-2010 using both statistical and process-based models, with a focus on comparing the performances between the two approaches. To allow for reasonable comparability, we develop an emulator which shares the same structure with the statistical model to mimic the behaviors of process-based models. Results show that the simulated maize yields in most of the top 10 producing countries are overestimated, when compared against FAO observations. Overall, GEPIC, EPIC-IIASA and EPIC-Boku show better performance than other models in reproducing the observed yield variations at the global scale. Climate variability explains 42.00% of yield variations in observation-based statistical model, while large discrepancy is found in crop models. Regionally, climate variability is associated with 55.0% and 52.20% of yield variations in Argentina and USA, respectively. Further analysis based on process-based model emulator shows that climate change has led to a yield loss by 1.51%-3.80% during the period 1980-1990, consistent with the estimations using the observation-based statistical model. As for the period 1991-2000, however, the observed yield loss induced by climate change is only captured by GEPIC and pDSSAT. In contrast to the observed positive climate impact for the period 2001-2010, CLM-Crop, EPIC-IIASA, GEPIC, pAPSIM, pDSSAT and PEGASUS simulated negative climate effects. The results point to the discrepancy between process-based and statistical crop models in simulating climate change impacts on maize yield, which depends on not only the regions, but also the specific time period. We suggest that more targeted efforts are required for constraining the uncertainties of both statistical and process-based crop models for future yield predictions. 

How to cite: Yin, X. and Leng, G.: Global contribution of climate variability and trends to maize yield change in observations and crop models during 1980-2010, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-21042, https://doi.org/10.5194/egusphere-egu2020-21042, 2020.

D807 |
Seoyeon kim and Yangwon lee

Land surface temperature is crucial in many field of study, such as Earth's surface water cycle, energy balances, energy exchange of ecosystem and  global climate change. As the role of LST is important, it should be accurately obtained on a global scale. However, it is still difficult to calculate LST from satellite because there are constraints of Atmospheric correction, cloud effect, verification representatively. Therefore, the goal is to improve and optimize the accuracy of LST estimates in satellite-based measurement by mixing various data such as multi-spectral thermal infrared image, hyperspectral thermal infrared image, microwave satellite image, etc., or comparing and applying many LST calculation methods and algorithms.

How to cite: kim, S. and lee, Y.: Improvement of satellite-based land surface temperature estimation, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-21240, https://doi.org/10.5194/egusphere-egu2020-21240, 2020.

D808 |
You jeong yoon and Yang won lee

Recently in Korea, the fish catch of offshore waters recorded less than 1 million tons in 44 years due to climate change and drastic changes in the fishing environment. Therefore, it is essential to produce and provide accurate fishing forecast information, such as the location of fishing fields and the amount of fish production, that varies in time and space according to fishing conditions to enhance the competitiveness of the fishing industry. Since the factors affecting the fish catch have various and nonlinear relationships, so this study predicted the catch based on deep learning. The study was selected as the three major fish species of the Korean coast -- anchovy, mackerel and squid. The research area was selected as four fishing area. (One fishing area is 14 km * 14 km). In order to produce accurate forecasted fishing information, it is necessary to identify major marine weather and biological factors affecting the fish catch by fish species and artificial intelligence modeling using marine and weather satellite images. The satellite data used in the study are from the Korea Meteorological Administration (KMA). So far, research on the relationship between two or more factors and fish catches has been insufficient in the previous research, so this study may contribute to the prediction of fishing trends.

How to cite: yoon, Y. J. and lee, Y. W.: Deep learning-based prediction of fish catch for the offshore waters in south korea, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-21243, https://doi.org/10.5194/egusphere-egu2020-21243, 2020.