Accurate measurements of various hydrological cycle components (e.g. precipitation, evapotranspiration, soil moisture and water storage changes) are essential for understanding the hydrological processes and further for sustainable water resources management. Hydrological cycle components are characterized by significant variability in time and space. The conventional in-situ measurements from gauges are generally considered to be the most accurate measurements, but scientific communities are often encountered with the limited availability and capability of in-situ measurements. Specifically, the network of gauge stations is often sparse and overall the number of stations is still on decreasing trend over the globe. The point-based feature makes gauge measurements insufficient to capture spatial and temporal variability of hydrological cycle components. Therefore, alternative data sources should be investigated to fill the data gaps.

Satellite remote sensing has been shown great capability of estimating various hydrological cycle components at different temporal and spatial scales. Various communities have recognized the importance of satellite remote sensing, but they have been stressing the need for improvements in accuracy and particularly the spatial resolution because the spatial resolution of remotely sensed products is still often too coarse for many applications. To this regard, a specific topic “spatial downscaling” has emerged; over last decades, considerable efforts have been made to develop various spatial downscaling algorithms to improve the spatial resolution of remotely sensed estimates.

Machine learning and geostatistical methods have been innovatively utilized to advance the spatial downscaling in satellite remote sensing community. Together with the algorithms development in spatial downscaling, further pertinent research question arise: how to accurately evaluate the skill of downscaled remote sensing products? All current approaches for evaluation contain known limitations and, hence, there is a clear need for the development of novel procedures for fair evaluation particularly considering the limitations (e.g. representativeness and availability) of ground measurements form gauge stations.

The aim of this session is to present and discuss novel procedures in spatial downscaling of remotely sensed hydrological cycle components with emphasis on algorithms development, innovative evaluation and application of downscaled estimates.

Convener: Zheng DuanECSECS | Co-conveners: Jianzhi DongECSECS, Jian PengECSECS, Hongkai Gao
| Attendance Tue, 05 May, 16:15–18:00 (CEST)

Files for download

Download all presentations (6MB)

Chat time: Tuesday, 5 May 2020, 16:15–18:00

Chairperson: Zheng Duan
D278 |
Oleg Yu. Khachay, Andrey Yu. Khachay, and Olga Hachay

In the enormous and still poorly mastered gap between the macro level, where the well-developed continuum theories of continuous media and engineering methods of calculation and design operate, and the atomic, subordinate to the laws of quantum mechanics, there is an extensive meso-hierarchical level of the structure of matter. At this level unprecedented previously products and technologies can be artificially created. Nano technology is a qualitatively new strategy in technology: it creates objects in exactly the opposite way - large objects are created from small ones [1]. We have developed a new method for modeling acoustic monitoring of a layered-block elastic medium with several inclusions of various physical and mechanical hierarchical structures [2]. An iterative process is developed for solving the direct problem for the case of three hierarchical inclusions of l, m, s-th ranks based on the use of 2D integro-differential equations. The degree of hierarchy of inclusions is determined by the values ​​of their ranks, which can be different while the first rank is associated with the atomic structure, the following ranks are associated with increasing geometric dimensions, which contain inclusions of lower ranks and sizes. Hierarchical inclusions are located in different layers one above the other: the upper one is abnormally stressed, the second is abnormally elastic and the third is abnormally dense. The degree of filling with inclusions of each rank for all three hierarchical inclusions is different. Modeling is carried out from smaller sizes to large inclusions; as a result, it becomes possible to determine the necessary parameters of the formed material from acoustic monitoring data.
[1] Nanotechnology in the coming decade. Forecast of the direction of research. (2002). World, Moscow - 292 p.
[2] Hachay, O. A., Khachay, A. Yu. and Khachay O. Yu. (2018). Modeling algorithm of acoustic waves penetrating through a medium with composite hierarchical inclusions.// AIP Conference Proceedings 2053, 030023; https://doi.org/10.1063/1.5084384.

How to cite: Khachay, O. Yu., Khachay, A. Yu., and Hachay, O.: Mathematical Modeling Algorithms for Obtaining New Materials with Desired Properties Using Nano-hierarchical Structures., EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-1323, https://doi.org/10.5194/egusphere-egu2020-1323, 2019

D279 |
Luca Zappa, Matthias Forkel, Angelika Xaver, and Wouter Dorigo

Remotely sensed data from microwave sensors have been successfully used to retrieve soil moisture on a global scale. In particular, passive and active microwave sensors with large footprints can observe the same location with a (sub-)daily frequency, but typically are characterized by spatial resolutions in the order of tens of km. Therefore, such coarse scale products can accurately capture the temporal dynamics of soil moisture but are inadequate in providing spatial details. However, several agricultural and hydrological applications could greatly benefit from soil moisture observations with a sub-kilometer spatial resolution while preserving a daily revisit time.

Here, we present a framework for downscaling coarse resolution satellite soil moisture products (ASCAT and SMAP) to high spatial resolution. In particular, we build robust relationships between remotely sensed soil moisture and ancillary variables on soil texture, topography, and vegetation cover. Such relationship is built through Random Forest regressions, trained against in-situ measurements of soil moisture. The proposed approach is developed and tested in an agricultural catchment equipped with a high-density network of in-situ sensors. Our results show a strong consistency between the downscaled and the observed spatio-temporal patterns of soil moisture. Furthermore, including a proxy of vegetation cover in the Random Forest regressions results in considerable improvements of the downscaling performance. Finally, if only limited training data can be used, priority should be given to increase the number of sensor locations to adequately cover the spatial heterogeneity, rather than expanding the duration of the measurements. 

Future research will focus on including additional ancillary variables as model predictors, e.g. Land Surface Temperature or backscatter, and on applying the downscaling framework to other regions with similar environmental and climatic conditions.

How to cite: Zappa, L., Forkel, M., Xaver, A., and Dorigo, W.: Estimation of high-resolution soil moisture using machine learning, satellite observations and ground measurements. A case study in a hilly agricultural region, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-8338, https://doi.org/10.5194/egusphere-egu2020-8338, 2020

D280 |
Zheng Duan, Cheng Chen, Hongkai Gao, and Jian Peng

Precipitation is an important component of the water cycle. Precipitation is characterized with high temporal and spatial variability. Accurate measurements of precipitation at high spatiotemporal resolution are essential for many applications in the fields of hydrology, meteorology and ecology. The traditional rain gauge stations provide direct measurements of rainfall at the surface but at a limited scale; rain gauge measurements are often considered as point-based measurements that are insufficient to represent the spatial variability of rainfall over a certain region, especially in the case of sparse rain gauge network. Satellite remote sensing has been developing with great ability of being used for estimating various water cycle components at different temporal and spatial scales. Considerable efforts have been made to develop satellite precipitation products at different spatial and temporal resolutions over the global or quasi-global scale. The majority of global/quasi-global precipitation products are at the spatial resolution of 0.25° (~25 km) with very few products at 0.05°-0.10° resolution. The usefulness of satellite precipitation products has been increasingly recognized but the relative coarse spatial resolution is still a limitation for many applications such as hydrological modelling at basin scales that generally require precipitation data at a desirable higher spatial resolution (e.g. 1 km). Over recent years, numerous spatial downscaling procedures/methods have been proposed to obtain precipitation products at higher spatial resolution. The relationships between precipitation and various auxiliary land-surface variables were explored and incorporated into spatial downscaling procedures using a large range of regression algorithms. Advanced machine learning and geostatistical methods have also been innovatively used to develop spatial downscaling procedures.


The aim of this study is to present a comprehensive review of studies on spatial downscaling of satellite precipitation products over the recent years. We will summarize the proposed spatial downscaling methods, investigated auxiliary land-surface variables and the evaluation strategy. The performance of spatial downscaling methods in studied regions and their applications will be compared and discussed in terms of advantages and limitations. Finally, we will conclude this paper with outlook on future research needs and associated challenges about spatial downscaling of satellite precipitation products.

How to cite: Duan, Z., Chen, C., Gao, H., and Peng, J.: A review of spatial downscaling of satellite precipitation products, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-11164, https://doi.org/10.5194/egusphere-egu2020-11164, 2020

D281 |
Stefan Mayr, Igor Klein, Claudia Künzer, and Martin Rutzinger

Large-scale remote sensing products offer opportunities to address global society relevant questions. One of the most vital resources of our planet is fresh water. To monitor dynamics, the application of water surface time-series has proven to be an effective tool, but to access reliable information, validation efforts are essential. Furthermore, increased utilization of remote sensing time-series products can be seen in modelling applications. In this process, uncertainty estimation of input datasets is typically required. Especially for large-scale remote sensing products with high temporal resolution, common validation approaches as comparison to in situ data or intercomparison to similar products is hardly viable. Here we propose the use of supervised- and unsupervised outlier detection methods to yield pixel-wise uncertainty estimates in an internal validation. Therefore, several algorithms are applied on a global, MODIS (Moderate Resolution Imaging Spectroradiometer) based daily accessible water surface product (DLR Global WaterPack). Two main sources have been identified to introduce uncertainty to the binary classification of cloud free observations. As mixed pixels (water/non-water) and water impurities contribute to changes in the RED-NIR profile, we evaluate their effects by utilizing classified Landsat 8 images to determine water subpixel fractions and identify turbid water. Results are analyzed and compared in initial test regions across the globe.

How to cite: Mayr, S., Klein, I., Künzer, C., and Rutzinger, M.: Uncertainty Estimation of a global Remote Sensing Water Surface Time-Series: the DLR Global WaterPack, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-16598, https://doi.org/10.5194/egusphere-egu2020-16598, 2020

How to cite: Mayr, S., Klein, I., Künzer, C., and Rutzinger, M.: Uncertainty Estimation of a global Remote Sensing Water Surface Time-Series: the DLR Global WaterPack, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-16598, https://doi.org/10.5194/egusphere-egu2020-16598, 2020

How to cite: Mayr, S., Klein, I., Künzer, C., and Rutzinger, M.: Uncertainty Estimation of a global Remote Sensing Water Surface Time-Series: the DLR Global WaterPack, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-16598, https://doi.org/10.5194/egusphere-egu2020-16598, 2020

D282 |
Vivien Georgiana Stefan, Olivier Merlin, Beatriz Molero, Maria-Jose Escorihuela, and Salah Er-Raki

High resolution (HR) soil moisture estimates are needed by a range of agricultural and hydrological applications, considering it’s one of the drivers of evaporation, infiltration and runoff. Since the resolution of current remote sensing estimates (tens of kilometres) is insufficient for the majority of these applications, different downscaling techniques are used to improve the resolution. Amongst the existing methodologies, DISPATCH (DISaggregation based on a Physical And Theoretical scale CHange) has been proven to accurately improve the resolution of SMOS (Soil Moisture Ocean Salinity) soil moisture data, by using a soil evaporative efficiency (SEE) model. SEE can be derived from remotely sensed land surface temperature (LST) and normalized difference vegetation index (NDVI) data. DISPATCH uses two different SEE models: a temperature-based LST-driven model, and a soil moisture-based SMOS-driven model. This study aims at improving the robustness of the soil moisture-based SEE model, by testing different calibrations and models. Two SM-based SEE models, one linear and one nonlinear, are tested, each being calibrated from remote sensing data on a daily and on a multi-date basis. The approaches were implemented over two mixed dry and irrigated areas in Catalonia, Spain, and over a dry area in Morocco.  When looking at the two models in the daily calibration mode, the linear model performs better. Over the two areas in Spain, the correlation coefficients obtained with the linear model are 0.63 and 0.18 as opposed to 0.13 and -0.08, respectively. In Morocco, the correlation coefficients are roughly similar, 0.32 (linear mode) and 0.31 (nonlinear mode). The slopes of linear regression are also improved in the linear case, 0.44 and 0.88, as opposed to -0.14 and 0.11, for the Spanish sites. However, the best results were obtained in the case of the nonlinear model with an annual calibration. When comparing the linear and nonlinear models in the annual calibration mode, correlation coefficients are improved when using the nonlinear mode, from 0.13 and -0.08 to 0.78 and to 0.47 (Spanish site), and from 0.25 to 0.33 (Moroccan site). The slopes of linear regression are also improved, from 0.11 to 0.88, -0.14 to 1.15 (Spain) and from 0.53 to 0.74 (Morocco). The root mean square difference is generally low, ranging from 0.03 to 0.17 m3/m3. Considering several studies that report a strong nonlinear behaviour of the SEE with respect to SM, the nonlinear SM-based model in DISPATCH, combined with a multi-date calibration, is proven to give a significantly better performance, enhancing the robustness of the derived HR SM products.

How to cite: Stefan, V. G., Merlin, O., Molero, B., Escorihuela, M.-J., and Er-Raki, S.: On the calibration of an evaporation-based disaggregation method of SMOS soil moisture data, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-17670, https://doi.org/10.5194/egusphere-egu2020-17670, 2020

D283 |
Yong-Tak Kim, Carlos H R Lima, and Hyun-Han Kwon

Rainfall simulation by climate model is generally provided at coarse grids and bias correction is routinely needed for the hydrological applications. This study aims to explore an alternative approach to downscale daily rainfall simulated by the regional climate model (RCM) at any desired grid resolution along with bias correction using a Kriging model, which better represents spatial dependencies of distribution parameters across the watershed. The Kringing model also aims to reproduce the spatial variability observed in the ground rainfall gauge. The proposed model is validated through the entire weather stations in South Korea and climate change scenarios simulated by the five different RCMs informed by two GCMs. The results confirmed that the proposed spatial downscaling model could reproduce the observed rainfall statistics and spatial variability of rainfall. The proposed model further applied to the climate change scenario. A discussion of the potential uses of the mode is offered.

KEYWORDS: Climate Change Scenario, Global Climate Models, Regional Climate Models, Statistical Downscaling, Spatial-Temporal Bias



This work was funded by the Korea Meteorological Administration Research and Development Program under Grant KMI(KMI2018-01215)

How to cite: Kim, Y.-T., Lima, C. H. R., and Kwon, H.-H.: Kriging Approach to Quantile Delta Mapping (QDM) for Spatial Downscaling of Climate Change Scenario, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20829, https://doi.org/10.5194/egusphere-egu2020-20829, 2020

D284 |
Ahmad Al Bitar, Nitu Ojha, Chiara Corbari, Olivier Merlin, Yann Kerr, and Marco Mancini

Downscaling of L-Band microwave using Sentinel-3 land surface temperature

A large number of agricultural and water management applications require sub-kilometric frequent revisit surface Soil Moisture (SM) observations. L-band passive radiometer acquisitions are especially suited for soil moisture retrieval since they are less susceptible to attenuation by vegetation than active methods and are less sensitive to surface roughness than C or X – bands. However, while providing a 3 days global coverage for ascending and descending orbits with the currently available missions (SMOS/SMAP) the spatial resolution of the space-borne L-band radiometers is of ~40 km. Downscaling technics have been extensively used to increase the resolution of the SM products by combining data from optical (Merlin et al. 2012) and SAR sensors (Tomer et al. 2015). Here, we use land surface temperature data from the Sentinel-3 sensors to disaggregate the SMOS SM product into the DISPATCH algorithm. DISPATCH is based on the link between the evaporative efficiency and the SM (Merlin et al. 2010). The exercices is applied over Italy and compared to in-situ SM observations and model outputs over two sites in Northern and southrn Italy (Chiese and Capitanata). The algorithm is run using  MODIS and the Sentinel-3 data for a comparative results. The potential of the combined use of Sentienl-3/MODIS and SMOS/SMAP is also investigate. The current study extends the application of an existing algorithm to new operational data from the Copernicus programe while accessing the advantages and ceavates. 

How to cite: Al Bitar, A., Ojha, N., Corbari, C., Merlin, O., Kerr, Y., and Mancini, M.: Downscaling of L-Band microwave using Sentinel-3 land surface temperature, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-21822, https://doi.org/10.5194/egusphere-egu2020-21822, 2020