Assimilation of satellite soil moisture data into rainfall-runoff modelling for several catchments worldwide (original) (raw)
Related papers
Data Assimilation of Satellite Soil Moisture into Rainfall-Runoff Modelling: A Complex Recipe?
Remote Sensing, 2015
Data assimilation (DA) of satellite soil moisture (SM) observations represents a great opportunity for improving the ability of rainfall-runoff models in predicting river discharges. Many studies have been carried out so far demonstrating the possibility to reduce model prediction uncertainty by incorporating satellite SM observations. However, large discrepancies can be perceived between these studies with the result that successful DA is not only related to the quality of the satellite observations but can be significantly controlled by many methodological and morphoclimatic factors. In this article, through an experimental study carried out on the Tiber River basin in Central Italy, we explore how the catchment area, soil type, climatology, rescaling technique, observation and model error selection may affect the results of the assimilation and can be the causes of the apparent discrepancies obtained in the literature. The results show that: (i) DA of SM generally improves discharge predictions (with a mean efficiency of about 30%); (ii) unlike catchment area, the soil type and the catchment specific characteristics might have a remarkable influence on the results; (iii) simple rescaling techniques may perform equally well to more complex ones; (iv) an accurate quantification of the model error is paramount for a correct choice of the observation error and, (v) SM temporal variability has a stronger influence than the season itself. On this basis, we advise that DA of SM may be not a simple task and one should carefully test the optimality of the assimilation experiment prior to drawing any general conclusions.
Water Resources Research, 2011
Recently, Crow et al. (2009) developed an algorithm for enhancing satellite-based land rainfall products via the assimilation of remotely sensed surface soil moisture retrievals into a water balance model. As a follow-up, this paper describes the benefits of modifying their approach to incorporate more complex data assimilation and land surface modeling methodologies. Specific modifications improving rainfall estimates are assembled into the Soil Moisture Analysis Rainfall Tool (SMART), and the resulting algorithm is applied outside the contiguous United States for the first time, with an emphasis on West African sites instrumented as part of the African Monsoon Multidisciplinary Analysis experiment. Results demonstrate that the SMART algorithm is superior to the Crow et al. baseline approach and is capable of broadly improving coarse-scale rainfall accumulations measurements with low risk of degradation. Comparisons with existing multisensor, satellite-based precipitation data products suggest that the introduction of soil moisture information from the Advanced Microwave Scanning Radiometer via SMART provides as much coarse-scale (3 day, 1) rainfall accumulation information as thermal infrared satellite observations and more information than monthly rain gauge observations in poorly instrumented regions.
Water Resources Research, 2016
This paper explores the use of active and passive microwave satellite soil moisture products for improving streamflow prediction within four large (>5000km 2) semiarid catchments in Australia. We use the probability distributed model (PDM) under a data-scarce scenario and aim at correcting two key controlling factors in the streamflow generation: the rainfall forcing data and the catchment wetness condition. The soil moisture analysis rainfall tool (SMART) is used to correct a near real-time satellite rainfall product (forcing correction scheme) and an ensemble Kalman filter is used to correct the PDM soil moisture state (state correction scheme). These two schemes are combined in a dual correction scheme and we assess the relative improvements of each. Our results demonstrate that the quality of the satellite rainfall product is improved by SMART during moderate-to-high daily rainfall events, which in turn leads to improved streamflow prediction during high flows. When employed individually, the soil moisture state correction scheme generally outperforms the rainfall correction scheme, especially for low flows. Overall, the combined dual correction scheme further improves the streamflow predictions (reduction in root mean square error and false alarm ratio, and increase in correlation coefficient and Nash-Sutcliffe efficiency). Our results provide new evidence of the value of satellite soil moisture observations within data-scarce regions. We also identify a number of challenges and limitations within the schemes.
Soil as a natural rain gauge: Estimating global rainfall from satellite soil moisture data
Journal of Geophysical Research: Atmospheres, 2014
Measuring precipitation intensity is not straightforward; and over many areas, ground observations are lacking and satellite observations are used to fill this gap. The most common way of retrieving rainfall is by addressing the problem "top-down" by inverting the atmospheric signals reflected or radiated by atmospheric hydrometeors. However, most applications are interested in how much water reaches the ground, a problem that is notoriously difficult to solve from a top-down perspective. In this study, a novel "bottom-up" approach is proposed that, by doing "hydrology backward," uses variations in soil moisture (SM) sensed by microwave satellite sensors to infer preceding rainfall amounts. In other words, the soil is used as a natural rain gauge. Three different satellite SM data sets from the Advanced SCATterometer (ASCAT), the Advanced Microwave Scanning Radiometer (AMSR-E), and the Microwave Imaging Radiometer with Aperture Synthesis are used to obtain three new daily global rainfall products. The "First Guess Daily" product of the Global Precipitation Climatology Centre (GPCC) is employed as main benchmark in the validation period 2010-2011 for determining the continuous and categorical performance of the SM-derived rainfall products by considering the 5 day accumulated values. The real-time version of the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis product, i.e., the TRMM-3B42RT, is adopted as a state-of-the-art satellite rainfall product. The SM-derived rainfall products show good Pearson correlation values (R) with the GPCC data set, mainly in areas where SM retrievals are found to be accurate. The global median R values (in the latitude band ±50°) are equal to 0.54, 0.28, and 0.31 for ASCAT-, AMSR-E-, and SMOS-derived products, respectively. For comparison, the median R for the TRMM-3B42RT product is equal to 0.53. Interestingly, the SM-derived products are found to outperform TRMM-3B42RT in terms of average global root-mean-square error statistics and in terms of detection of rainfall events. The regions for which the SM-derived products perform very well are Australia, Spain, South and North Africa, India, China, the Eastern part of South America, and the central part of the United States. The SM-derived products are found to estimate accurately the rainfall accumulated over a 5 day period, an aspect particularly important for their use for hydrological applications, and that address the difficulties of estimating light rainfall from TRMM-3B42RT.
In the Ensemble Kalman Filter (EnKF)-based data assimilation, the background prediction of a model is updated using observations and relative weights based on the model prediction and observation uncertainties. In practice, both model and observation uncertainties are difficult to quantify thus have been often assumed to be spatially and temporally independent Gaussian random variables. Nevertheless, it has been shown that incorrect assumptions regarding the structure of these errors can degrade the performance of the stochastic data assimilation. This work investigates the autocorrelation structure of the microwave satellite soil moisture retrievals and explores how assumed observation error structure affects streamflow prediction skill when assimilating these observations into a rainfall-runoff model. An AMSR-E soil moisture product and the Probability Distribution Model (PDM) are used for this purpose. Satellite soil moisture data is transformed with an exponential filter to make it comparable to the root zone soil moisture state of the model. The exponential filter formulation explicitly incorporates an autocorrelation component in the rescaled observation, however, the error structure of this operator has been treated until now as an independent Gaussian process. In this work, the variance of the rescaled observation error is estimated based on the residuals from the rescaled satellite soil moisture and the calibrated model soil moisture state. Next, the observation error structure is treated as a Gaussian independent process with time-variant variance; a weakly autocorrelated random process (with autocorrelation coefficient of 0.2) and a strongly autocorrelated random process (with autocorrelation coefficient of 0.8). These experiments are compared with a control case which corresponds to the commonly used assumption of Gaussian independent observation error with time-fixed variance. Model error is represented by perturbing rainfall forcing data and soil moisture state. These perturbations are assumed to represent all forcing and model structural/parameter errors. Error parameters are calibrated by applying two discharge ensemble verification criteria. Assimilation results are compared and the impacts of the observation error structure assumptions are assessed. The study area is the semi-arid 42,870 km 2 Warrego at Wyandra River catchment, located in Queensland, Australia. This catchment is chosen for its flooding history, along with having geographical and climatological conditions that enable soil moisture satellite retrievals to have higher accuracy than in other areas. These conditions include large area, semi-arid climate and low vegetation cover. Moreover, the catchment is poorly instrumented, thus satellite data provides valuable information. Results show a consistent improvement of the model forecast accuracy of the control case and in all experiments. However, given that a stochastic assimilation is designed to correct stochastic errors, the systematic errors in model prediction (probably due to the inaccurate forcing data within the catchment) are not addressed by these experiments. The assumed observation error structures tested in the different experiments do not exhibit significant effect in the assimilation results. This case study provides useful insight into the assimilation of satellite soil moisture retrievals in poorly instrumented semi-arid catchments.
Hydrological applications of satellite data: 1. Rainfall estimation
Journal of Geophysical Research, 1996
In this study we investigate the ability of satellite visible and infrared data to produce reliable rainfall amount estimates that could be used by hydrological models to predict streamflow for large basins. Rainfall estimates are obtained by (1) classification of clouds to raining and nonraining clouds and (2) applying a multivariate statistical model between rainfall and indices derived from the satellite observations. Satellite data corresponding to 180 randomly selected days in the period May-September 1982-1988 are used in this study that focuses on the estimation of daily rainfall. The Des Moines River basin in the midwestern United States is the application area. The correlation coefficient between model-predicted and rain gauge-observed mean areal precipitation over areas of order 10,000 km 2 is found to be about 0.85. In an example application the satellite rainfall estimates are used to force the operational National Weather Service hydrologic forecast model for a subbasin of the Des Moines River basin. The model has been calibrated with rain gauge data. The results show that differences between rain gauge and satellite rainfall input generate differences in flow forecasts and upper soil water model estimates, which are a function of the antecedent soil water conditions. A companion paper [Guetter et al., this issue] quantifies the effects that the differences between rain gauge and satellite rainfall estimates have on flow and upper soil water model predictions for various spatial scales and for hydrologic models calibrated with and without satellite data. Wyss, J., E.R. Williams, and R.L. Bras, Hydrologic modeling of New England river basins using radar rainfall data,
Journal of the Indian Society of Remote Sensing, 2019
Accurate knowledge of the spatio-temporal variation in soil moisture provides insight into larger-scale hydrological processes and can, therefore, help in improving hydrological predictions. The strength of remote sensing for mapping surface soil moisture is well proven. In addition, data assimilation offers the opportunity to combine the advantages of modelling with those of remote sensing data to achieve higher accuracy and continuous improvement in hydrological forecasts. In this study, Advanced Microwave Scanning Radiometer for Earth observation science soil moisture product was assimilated into Variable Infiltration Capacity (VIC) hydrological model using Kalman filter data assimilation technique. Further, the updated multilayer spatio-temporal soil moisture distributions across the Mahanadi Basin, India, were simulated using the hydrological model. The VIC model was set up and parameterized using field-observed and remote sensing-derived data. Based on the sensitivity analysis of the model, the 'four-parameter' (Tmax, Tmin, Prec, and WS) meteorological forcing scenario was selected as the operational scenario. The output fluxes obtained from VIC were routed to simulate discharge at five stations for the calibration and validation of the model. With R 2 and model efficiency values close to 0.95 and 0.99, respectively, the model was proven to be suitable for simulating the hydrological responses of the basin. Soil moisture was assimilated in the top soil layer of the model using the Kalman filter approach, and the multilayer soil moisture regime was generated using the modelling approach. The validation of soil moisture (assimilated) products proves that these products are better than remote sensing and traditionally modelled soil moisture products, in both spatial and temporal domains in terms of availability and accuracy.
Validation of remote sensing soil moisture products with a distributed continuous hydrological model
2014 IEEE Geoscience and Remote Sensing Symposium, 2014
The reliable estimation of soil moisture in space and time is of fundamental importance in operational hydrology to improve the forecast of the rainfall-runoff response of catchments and, consequently, flood predictions. Nowadays several satellite-derived soil moisture products are available and can offer a chance to improve hydrological model performances especially in environments with scarce ground based data. The goal of this work is to test the effects of the assimilation of different satellite soil moisture products in a distributed physically based hydrological model. Among the currently available different satellite platforms, four soil moisture products, from both the ASCAT scatterometer and the SMOS radiometer, have been assimilated using a Nudging scheme. The model has been applied to a test basin (area about 800 km 2) located in Northern Italy for the period July 2012-June 2013.
Assimilation of observed soil moisture data in storm rainfall-runoff modeling
Journal of Hydrologic Engineering, 2009
Estimation of antecedent wetness conditions is one of the most important aspects of storm rainfall-runoff modeling. This study investigated the use of observations of near-surface soil moisture carried out in a small experimental plot to estimate wetness conditions of five nested catchments, from 13 to 137 km 2 in area, in Central Italy, including the plot itself. In particular, the relationship between the observed degree of saturation, e , and the potential maximum retention parameter, S, of the soil conservation service-curve number ͑SCS-CN͒ method for abstraction was investigated using 15 rainfall-runoff events ͑ten for calibration and five for verification͒ that occurred in the period 2002-2005. Two antecedent precipitation indices ͑API͒ and one base flow index ͑BFI͒ were also considered for the estimation of wetness conditions. When interpreting S as the mean soil water deficit of the catchment, an inverse linear relationship with e was found with the coefficient of determination decreasing with catchment area, but still significant for the largest catchment. On the contrary, the reliability of regression increased with catchment area when BFI was employed. Both API indices led to poor results for all investigated catchments. The accuracy of the modified SCS-CN method, i.e., incorporating e for the estimation of S, coupled with a geomorphological unit hydrograph transfer function, was tested in simulating the catchment response. Assimilating the observed soil moisture in the rainfall-runoff model, both the runoff volume and the peak discharge were well predicted with average Nash-Sutcliffe efficiency greater than 90% in the verification phase.