Luis A. Bastidas

Parameter estimation of a land surface scheme using multicriteria methods

Attempts to create models of surfaceߚ;atmosphere interactions with greater physical realism have resulted in land surface schemes (LSS) with large numbers of parameters. The hope has been that these parameters can be assigned typical values by inspecting the literature. The potential for using the various observational data sets that are now available to extract plot-scale estimates for the parameters of a complex LSS via advanced parameter estimation methods developed for hydrological models is explored in this paper. Results are reported for two case studies using data sets of typical quality but very different location and climatological regime (ARM-CART and Tucson). The traditional single-criterion methods were found to be of limited value. However, a multicriteria approach was found to be effective in constraining the parameter estimates into physically plausible ranges when observations on at least one appropriate heat flux and one properly selected state variable are available.

Sensitivity analysis of a land surface scheme using multicriteria methods

Attempts to model surface-atmosphere interactions with greater physical realism have resulted in complex land surface schemes (LSS) with large numbers of parameters. A companion paper describes a multicriteria calibration procedure for extracting plot-scale estimates of the preferred ranges of these parameters from the various observational data sets that are now available. A complementary procedure is presented in this paper that provides an objective determination of the multicriteria sensitivity of the modeled variables to the parameters, thereby allowing the number of calibration parameters and hence the computational effort to be reduced. Two case studies are reported for the BATS model using data sets of typical quality but very different location and climatological regime (ARM-CART and Tucson). The sensitivity results were found to be consistent with the physical properties of the different environments, thereby supporting the reasonableness of the model formulation. Further, when the insensitive parameters are omitted from the calibration process, there is little degradation in the quality of the model description and little change in the preferred range of the remaining parameters.

Evaluation and Transferability of the Noah Land Surface Model in Semiarid Environments

Abstract This paper investigates the performance of the National Centers for Environmental Prediction (NCEP) Noah land surface model at two semiarid sites in southern Arizona. The goal is to evaluate the transferability of calibrated parameters (i.e., direct application of a parameter set to a “similar” site) between the sites and to analyze model performance under the various climatic conditions that can occur in this region. A multicriteria, systematic evaluation scheme is developed to meet these goals. Results indicate that the Noah model is able to simulate sensible heat, ground heat, and ground temperature observations with a high degree of accuracy, using the optimized parameter sets. However, there is a large influx of moist air into Arizona during the monsoon period, and significant latent heat flux errors are observed in model simulations during these periods. The use of proxy site parameters (transferred parameter set), as well as traditional default parameters, results in diminished model perfo...

Constraining Land Surface and Atmospheric Parameters of a Locally Coupled Model Using Observational Data

Abstract In coupled land surface–atmosphere modeling, the possibility and benefits of constraining model parameters using observational data bear investigation. Using the locally coupled NCAR Single-column Community Climate Model (NCAR SCCM), this study demonstrates some feasible, effective approaches to constrain parameter estimates for coupled land–atmosphere models and explores the effects of including both land surface and atmospheric parameters and fluxes/variables in the parameter estimation process, as well as the value of conducting the process in a stepwise manner. The results indicate that the use of both land surface and atmospheric flux variables to construct error criteria can lead to better-constrained parameter sets. The model with “optimal” parameters generally performs better than when a priori parameters are used, especially when some atmospheric parameters are included in the parameter estimation process. The overall conclusion is that, to achieve balanced, reasonable model performance ...

Downscaling and Assimilation of Surface Soil Moisture Using Ground Truth Measurements

Methods for reconciliation of spatial and temporal scales of data have become increasingly important as remote sensing data become more readily available and as the science of hydrology moves more heavily toward distributed modeling. The purpose of this paper is to develop a method to disaggregate coarse-resolution remote sensing data to finer scale resolutions that are more appropriate for use in hydrologic studies and water management. This disaggregation is done with the help of point measurements on the ground. The downscaling of remote sensing data is achieved by three main steps: initialization, spatial pattern mimicking, and assimilation. The first two steps are part of the main algorithm, and the last step, assimilation, is included for fine-tuning and to ensure further compatibility between the coarse-scale and fine-scale images. The assimilation step also incorporates the information coming from the point measurements. The approach has been applied and validated by downscaling images for two cases. In the first case, a synthetically generated random field is reproduced at fine and coarse resolutions. The downscaled image has been shown to match the spatial properties of the true image according to the variogram test as well as the magnitude of values according to the various univariate goodness-of-fit measures R2 = 0.91. In the second case, a soil moisture field from the Southern Great Plains (SGP 97) experiments is downscaled from a resolution of 800 m X 800 m to a resolution of 50 m X 50 m.

Downscaling and Forecasting of Evapotranspiration Using a Synthetic Model of Wavelets and Support Vector Machines

Providing reliable forecasts of evapotranspiration (ET) at farm level is a key element toward efficient water management in irrigated basins. This paper presents an algorithm that provides a means to downscale and forecast dependent variables such as ET images. Using the discrete wavelet transform (DWT) and support vector machines (SVMs), the algorithm finds multiple relationships between inputs and outputs at all different spatial scales and uses these relationships to predict the output at the finest resolution. Decomposing and reconstructing processes are done by using 2-D DWT with basis functions that suit the physics of the property in question. Two-dimensional DWT for one level will result in one datum image (low-low-pass filter image) and three detail images (low-high, high-low, and high-high). The underlying relationship between the input variables and the output are learned by training an SVM on the datum images at the resolution of the output. The SVM is then applied on the detailed images to produce the detailed images of the output, which are needed to help downscale the output image to a higher resolution. In addition to being downscaled, the output image can be shifted ahead in time, providing a means for the algorithm to be used for forecasting. The algorithm has been applied on two case studies, one in Bondville, IL, where the results have been validated against AmeriFlux observations, and another in the Sevier River Basin, UT.

Sensitivity analysis using mass flux and concentration

Sensitivity analysis for hydrochemical models requires consideration of the multivariate nature of watershed response. A robust multiobjective generalized sensitivity analysis (MOGSA) procedure, recently developed at the University of Arizona, was used to fully investigate the single objective parameter sensitivity of the Alpine Hydrochemical Model (AHM). A total of 20000 simulations for a two-year period were conducted for the Emerald Lake watershed in Sequoia National Park, California. For each simulation 21 objective functions were evaluated: they were discharge and both concentration and mass flux for ten chemical species. The MOGSA procedure revealed that only 2000 simulations were necessary to establish the parameters sensitive to mass flux or concentration. We found significant differences in parameter sensitivity for concentration versus mass flux objective functions. For example, a snowpack elution parameter and a number of hydrologic parameters were sensitive for Cl - concentration, while only the snowpack elution parameter was sensitive for Cl- mass flux. By using mass flux instead of concentration fewer mineral weathering parameters and more soil exchange parameters were sensitive. Mass flux calculations emphasize the spring snowmelt and peak discharge events of the early summer. Our results indicate that using mass instead of concentration permits better identification of the model parameters that most affect stream conditions during peak springtime flows and that some combination of mass flux and concentration objectives should be used in evaluating model performance.

Impacts of a Parameterization Deficiency on Offline and Coupled Land Surface Model Simulations

Surface water and energy balance plays an important role in land surface models, especially in coupled land surface‐atmospheric models due to the complicated interactions between land surfaces and the overlying atmosphere. The primary purpose of this paper is to demonstrate the significant negative impacts that a minor deficiency in the parameterization of canopy evaporation may have on offline and coupled land surface model simulations. In this research, using the offline NCAR Land Surface Model (LSM) and the locally coupled NCAR Single-column Community Climate Model (SCCM) as examples, intensive effort has been focused on the exploration of the mechanisms involved in the activation of unrealistically high canopy evaporation and thus unreasonable surface energy partitions because of a minor deficiency in the parameterization of canopy evaporation. The main causes responsible for exacerbating the impacts of the deficiency of the land surface model through the coupling of the two components are analyzed, along with possible impacts of land surface parameters in triggering the problems. Results from experimental runs show that, for a large number of randomly generated physically realistic land surface parameter sets, this model deficiency has caused the occurrences of negative canopy water with a significantly high frequency for both the offline NCAR LSM and the coupled NCAR SCCM, suggesting that land surface parameters are not the only important factors in triggering the problems associated with the model deficiency. In addition, the concurrence of intense solar radiation and enough precipitation is identified to be mainly responsible for exacerbating the negative impacts of the parameterization deficiency. Finally, a simple adjustment has been made in this study to effectively prevent the occurrences of negative canopy water storages, leading to significantly improved model performances.

The challenge of predicting flash floods from thunderstorm rainfall

A major characteristic of the hydrometeorology of semi–arid regions is the occurrence of intense thunderstorms that develop very rapidly and cause severe flooding. In summer, monsoon air mass is often of subtropical origin and is characterized by convective instability. The existing observational network has major deficiencies for those regions in providing information that is important to run–off generation. Further, because of the complex interactions between the land surface and the atmosphere, mesoscale atmospheric models are currently able to reproduce only general features of the initiation and development of convective systems. In our research, several interrelated components including the use of satellite data to monitor precipitation, data assimilation of a mesoscale regional atmospheric model, modification of the land component of the mesoscale model to better represent the semi–arid region surface processes that control run–off generation, and the use of ensemble forecasting techniques to improve forecasts of precipitation and run–off potential are investigated. This presentation discusses our ongoing research in this area; preliminary results including an investigation related to the unprecedented flash floods that occurred across the Las Vegas valley (Nevada, USA) in July of 1999 are discussed.

A Parsimonious Hydrological Model for a Data Scarce Dryland Region

Inapplicability of state of the art hydrological models due to scarce data motivates the need for a modeling approach that can be well constrained to available data and still model the dominant processes. Such an approach requires embedded model relationships to be simple and parsimonious in parameters for robust model selection. Simplicity in functional relationship is also important from water management point of view if these models are to be coupled with economic system models for meaningful policy assessment. We propose a similar approach, but rather than selecting (through calibration) processes from a set of processes predefined in terms of functionalities or modules, we model already known dominant processes in dryland areas (evaporation, Hortonian overland flows, transmission loses and subsurface flows) in a simple manner by explicitly programming them as constraints and obtain parameters by minimizing a performance based objective function. Such use of mathematical programming allows flexible model calibration and simulation in terms of available data and constraints. The model results of the approach are however not perfect given its infancy. Nonetheless its imperfections can guide us to further improvements, in particular with regards to model structure improvement.