Search

A Deep Neural Network (DNN) is used to estimate the Advanced Scatterometer (ASCAT) C-band microwave normalized backscatter (σ 40 o ), slope (σ ′) and curvature (σ ″) over France. The Interactions between Soil, Biosphere and Atmosphere (ISBA) land surface model (LSM) is used to produce land surface variables (LSVs) that are input to the DNN. The DNN is trained to simulate σ 40 o , σ ′ and σ ″ from 2007 to 2016. The predictive skill of the DNN is evaluated during an independent validation period from 2017 to 2019. Normalized sensitivity coefficients (NSCs) are computed to study the sensitivity of ASCAT observables to changes in LSVs as a function of time and space. Model performance yields a near-zeros bias in σ 40 o and σ ′. The domain-averaged values of ρ are 0.84 and 0.85 for σ 40 o and σ ′, compared to 0.58 for σ ″. The domain-averaged unbiased RMSE is 8.6% of the dynamic range for σ 40 o and 13% for σ ′, with land cover having some impact on model performance. NSC results show that the DNN-based model could reproduce the physical response of ASCAT observables to changes in LSVs. Results indicated that σ 40 o is sensitive to surface soil moisture and LAI and that these sensitivities vary with time, and are highly dependent on land cover type. The σ ′ was shown to be sensitive to LAI, but also to root zone soil moisture due to the dependence of vegetation water content on soil moisture. The DNN could potentially serve as an observation operator in data assimilation to constrain soil and vegetation water dynamics in LSMs. • A Deep Neural Network was trained to predict ASCAT backscatter, slope and curvature. • Input variables describing the soil and vegetation were obtained from the ISBA LSM. • Excellent agreement was achieved between observed and predicted ASCAT observables. • Sensitivity of ASCAT observables to inputs is consistent with microwave theory. • The DNN can be used as a measurement operator to assimilate ASCAT data into ISBA. [ABSTRACT FROM AUTHOR]

Summary: Important damages occur in small headwater catchments when they are hit by severe storms with complex spatio-temporal structure, sometimes resulting in flash floods. As these catchments are mostly not covered by sensor networks, it is difficult to forecast these floods. This is particularly true for road submersions, representing major concerns for flood event managers. The use of Quantitative Precipitation Estimates and Forecasts (QPE/QPF) especially based on radar measurements could particularly be adequate to evaluate rainfall-induced risks. Although their characteristic time and space scales would make them suitable for flash flood modelling, the impact of their uncertainties remain uncertain and have to be evaluated. The Gard region (France) has been chosen as case study. This area is frequently affected by severe flash floods, and an application devoted to the road network has also been recently developed for the North part of this region. This warning system combines distributed hydro-meteorological modelling and susceptibility analysis to provide warnings of road inundations. The warning system has been tested on the specific storm of the 29–30 September 2007. During this event, around 200mm dropped on the South part of the Gard and many roads were submerged. Radar-based QPE and QPF have been used to forecast the exact location of road submersions and the results have been compared to the effective road submersions actually occurred during the event as listed by the emergency services. Used on an area it has not been calibrated, the results confirm that the road submersion warning system represents a promising tool for anticipating and quantifying the consequences of storm events at ground. It rates the submersion risk with an acceptable level of accuracy and demonstrates also the quality of high spatial and temporal resolution radar rainfall data in real time, and the possibility to use them despite their uncertainties. However because of the quality of rainfall forecasts falls drastically with time, it is not often sufficient to provide valuable information for lead times exceeding 1h. [ABSTRACT FROM AUTHOR]

Summary: A method to estimate antecedent soil moisture conditions is proposed to improve the accuracy of flash flood forecasts at ungauged locations. The method combines two indexes: a ‘climatic’ temporal index calculated in each cell using an uncalibrated soil moisture accounting (SMA) scheme and a spatial ‘statistical’ index giving the “average saturation state” usually encountered before a flood. The proposed method was carried out on two different models: an event-based distributed model, and a REGRESSION model. Simulation results were analysed for 562 individual events, issued from 160 catchments located in the South of France. The presented method improved the efficiency of both models. This indicates that the both proposed indexes give relevant information for estimating antecedent soil moisture conditions at ungauged locations. Furthermore, a contingency statistical analysis based on a 2-year flood threshold showed that alerts were better forecasted by the distributed model. These encouraging results suggest that further efforts should be made for the development of the method and its operational application. [ABSTRACT FROM AUTHOR]

Summary: In the general context of field experiment design, this paper presents a modeling study that quantifies the respective impact of rainfall estimation and soil variability on the simulated discharge for an extreme event in southern France. The CVN distributed hydrological model, built within the LIQUID® modeling platform is used. The method is illustrated for two medium sized catchments, Saumane (99km2) and Uzès (88km2) using raingauges and two radar estimates. The soil properties are extracted from an existing soil database provided for the whole region. The model parameter specification uses available observation and a priori hydrological knowledge. No parameter adjustment is performed. For model evaluation on the regional scale, simulated maximum peak discharges are compared with post-flood estimations for 32 catchments. The area of these catchments ranges from 2.5 to 99km2 and model results are satisfactory. Then, the study focuses on the Saumane and Uzès catchments. A sensitivity analysis highlights the role of the Manning roughness coefficient on the simulated hydrographs dynamics. The impact of the bottom boundary condition of the infiltration and water redistribution module is also shown for the gauged Saumane catchment. Then the impact of rainfall input and soil spatial variability is presented. The results show that (i) the use of radar data is necessary to properly simulate the flood dynamics; (ii) although radar volume-scanning strategy has been shown to give more accurate results on a pixel/gauge comparison of the rainfall estimations, it is not necessarily the case when catchment averaged amounts are considered, especially for catchments in mountainous areas; (iii) the impact of the variability in soil properties on the simulated discharges is of the same order of magnitude as the impact of differences in rainfall estimation; (iv) the flood dynamics presents two phases: the first one, mainly controlled by the soil properties and the second one, since the soils are saturated, controlled by the rainfall variability. Therefore, uncertainties on both observations need to be mitigated in order to improve flash-flood understanding. [Copyright &y& Elsevier]

Summary: Characterisation of the error structure of radar quantitative precipitation estimation (QPE) is a major issue for applications of radar technology in hydrological modelling. Due to the variety of sources of error in radar QPE and the impact of correction algorithms, the problem can only be addressed practically by comparison of radar QPEs with reference values derived from ground-based measurements. Using the radar and raingauge datasets of the Bollène-2002 experiment, a preliminary investigation of this subject has been carried out within a geostatistical framework in the context of the Cévennes–Vivarais Mediterranean Hydrometeorological Observatory. First, raingauge measurements were critically analysed using variograms to detect erroneous squared differences between pairs of raingauge values. The anisotropic block kriging technique was then used to compute and select the most reliable reference values. The statistical distribution and the spatial and temporal structure of the residuals between radar and reference values was established and analysed. The error variance separation concept was also tested to estimate the variance of the residuals between the radar estimates and the true unknown rainfall. A preliminary version of the radar QPE error model was established for 1-km2 domains and time steps ranging from 1 to 12h using the limited data sample of the Bollène-2002 experiment. The error model is dependent on the time scales considered and needs to be conditioned on the rain rate and rainfall type as well as on the radar range. [ABSTRACT FROM AUTHOR]

• Contributions and complementarities of the different frequencies for detecting 4 types of land use (and two states of bare soils. • The use of one mono-frequency images provides limited overall results. • A significant improvement in the classifications by fusing images delivered by optical and radar sensors, for most of the classes studied. • The complementary nature of the contributions made by L-band and optical waves. The monitoring of different crops (cultivated plots) and types of surface (bare soils, etc.) is a crucial economic and environmental issue for the management of resources and human activity. In this context, the objective of this study is to evaluate the contribution of multispectral satellite imagery (optical and radar) to land use and land cover classification. Object-oriented supervised classifications, based on a Random Forest algorithm, and majority zoning post-processing are used. This study emerges from the experiment on multi-sensor crop monitoring (MCM'10, Baup et al., 2012) conducted in 2010 on a mixed farming area in the southwest of France, near Toulouse. This experiment enabled the regular and quasi-synchronous collection of multi-sensor satellite data and in situ observations, which are used in this study. 211 plots with contrasting characteristics (different slopes, soil types, aspects, farming practices, shapes and surface areas) were monitored to represent the variability of the study area. They can be grouped into four classes of land cover: 39 grassland areas, 100 plots of wheat, 13 plots of barley, 20 plots of rapeseed, and 2 classes of bare soil: 23 plots of small roughness and 16 plots of medium roughness. Satellite radar images in the X-, C- and L-bands (HH polarization) were acquired between 14 and 18 April 2010. Optical images delivered by Formosat-2 and corresponding field data were acquired on 14 April 2010. The results show that combining images acquired in the L-band (Alos) and the optical range (Formosat-2) improves the classification performance (overall accuracy = 0.85, kappa = 0.81) compared to the use of radar or optical data alone. The results obtained for the various types of land cover show performance levels and confusions related to the phenological stage of the species studied, with the geometry of the cover, the roughness states of the surfaces, etc. Performance is also related to the wavelength and penetration depth of the signal providing the images. Thus, the results show that the quality of the classification often increases with increasing wavelength of the images used. [ABSTRACT FROM AUTHOR]