H12A-01 10:20h
Brightness Temperature Versus Surface Soil Moisture Assimilation
Accurate initialisation of land surface soil moisture and temperature is crucial for improved weather and climate prediction in coupled land-ocean-atmosphere models. Remote sensing methods provide the necessary measurements for constraining off-line land surface model predictions to be used in coupled-model initialisation. Low frequency passive microwave is the remote sensor of choice as it has all-weather capability with 1 to 3 day repeat coverage, and is sensitive to both near-surface soil moisture and temperature. However, to be useful for prediction studies, deeper soil moisture and temperature must be inferred from the near-surface observations using a data assimilation framework. The current practice of data assimilation uses a derived near-surface soil moisture product rather than the passive microwave measurements of brightness temperatures directly. The potential advantages for assimilating the brightness temperature are i) a capability to constrain soil temperature in addition to soil moisture prediction, and ii) that soil temperature estimates used in deriving the soil moisture product may be poor, yielding a poor near-surface soil moisture retrieval. However, it is currently unclear if it is better to assimilate the derived soil moisture product or the raw brightness temperature observations. Therefore, this study explores the potential improvement in soil moisture prediction accuracy achieved by direct assimilation of brightness temperature data relative to a retrieved soil moisture product. This study uses C-band data from the Scanning Multifrequency Microwave Radiometer (SMMR) and the Catchment Land Surface Model for all of Australia.
H12A-02 10:35h
VALIDATION OF A COUPLED HYDROLOGICAL AND METEOROLOGICAL MODEL SYSTEM FOR INVESTIGATING FEEDBACK EFFECTS
Understanding of the interaction within the land surface hydrological processes is the key to determining the effect of land-use change and climate change on the hydrological systems. Traditionally, the hydrological impacts of climate change have been based on driving hydrological models with the output of region climate models. These climate models often operate at spatial and temporal scales that are much larger than the scales required to analyse the effects on the hydrological system. This is in part because of computational limitations and in part because of the physics of the regional models do not justify much higher resolution. Furthermore there is an inherent contradiction in this approach since these climate models include their own hydrological model component. Similarly in analysing the hydrological effects of land-use change the feedback to the meteorological system is often neglected. To address these issues a coupled hydrological and meteorological model system for evaluating interactions at hydrological (catchment) scales has been developed. A comprehensive hydrological modelling system describing the terrestrial component of the hydrological cycle has been modified to allow coupling to a local scale meteorological models. The hydrological mode includes both catchment rainfall-runoff processes and routing and hydraulic processes in the river system. As simulations can be run with and without coupling to the meteorological model, it is possible to evaluate the impact of feedbacks between the two systems on hydrological predictions. The uncoupled system is first validated against remote sensing and eddy correlation measurements at the field and landscape scale describing the hydrological and energy fluxes on the land-surface. The coupled system is then validated against field data describing both the atmosphere and hydrological system. Finally, a sensitivity analysis is carried out to examine the sensitivity of hydrological predictions to atmospheric feedbacks.
H12A-03 10:50h
Further Test of the Hypothesis of Maximum Evaporation Using New Field Observations
It was recently proposed (Wang et al, 2004, WRR) that there exists an extremum principle that governs the evaporation processes over land surfaces. The hypothesis of maximum evaporation was tested using observations of surface fluxes and surface temperature and soil moisture from earlier field experiments. To further test the theory, a new field experiment was conducted this Spring in Iowa for a more rigorous test of the hypothesis. Two observing towers were set up to measure turbulent fluxes of latent, sensible, and radiative fluxes with high sampling frequency. Soil heat flux and soil temperature at and below the land surface were measured using state-of-the-art devices. Soil moisture (volumetric water content and soil water potential) was also sampled with comparable temporal resolution, which is important for the hypothesis test. Continuous measurements over a period of two months have been obtained. The preliminary results, based on the new data, also support of the hypothesis of maximum evaporation.
H12A-04 11:05h
Constraining root-zone soil moisture estimates under dense vegetation using multi-frequency remote sensing observations.
Operational monitoring of surface soil moisture via spaceborne microwave radiometry should become a reality within the next decade. Unfortunately, the vertical support of these measurements is too shallow (top 2 to 5 cm of soil column) and the horizontal resolution too coarse (less than 10 km) for many agricultural and water resource applications. The most viable solution for the lack of vertical measurement support is the use of data assimilation systems and multi-layer hydrologic modeling to estimate root-zone soil moisture based on sufficiently frequent surface soil moisture observations. While such inversion are theoretically possible using data assimilation systems, it is unclear how robust surface soil moisture data assimilation procedures will be over agricultural crops where root-zone soil water loses are dominated by root uptake of soil water at depths far greater than the measurement depth of the radiometer. Consequently, the most robust strategies for operationally monitoring root-zone soil moisture in agricultural areas are likely to be based on integrating both microwave surface soil moisture retrievals and surface energy balance predictions obtained from thermal surface radiometric temperature observations into a multi-layer hydrologic model. This research explores competing strategies for combining microwave soil moisture retrievals and radiometric surface temperature observations within a hydrologic modeling framework to improve the model's representation of the root-zone soil water balance. Remote sensing observations will be used to constrain key hydrologic fluxes into (and out of) the soil column root zone. Results will demonstrate circumstances under which the assimilation of surface soil moisture alone will be inadequate to fully constrain root-zone soil moisture estimates beneath heavily vegetated canopies and explore the potential for surface energy flux estimates from diagnostic remote sensing models to provide additional constraints.
H12A-05 11:20h
On the Selection of Soil Hydraulic Properties in Land Surface Models Based on Soil Texture
One of the largest problems facing land surface modelers today is the lack of adequate parameter estimates. In particular, there is a lack of solid information on the spatial distribution of Soil Hydraulic Properties (SHPs). This study focuses on the effect of SHP selection on model outputs. We focus on modeled surface fluxes following a typical rain storm in a semi-arid environment. SHPs are often defined based on a soil texture classification, but soil texture class alone does not adequately specify SHPs. We show that a land surface model run with the range of SHPs from a large soils database produces more variability within a soil texture class than between classes. We take 1306 soils from the SHP database of Schaap and Leij (1998) and run the Noah land surface model with each soil. The only soil texture class which appears distinct is the sand class, and even for sands there is significant overlap with all other classes. Within most soil texture classes, the outputs have a range of 400$\frac{W}{m^2}$ for latent and sensible heat fluxes, 60$\frac{W}{m^2}$ for ground heat flux, and a range of almost 10K in surface temperature, in contrast the average difference between mean values of different soil texture classes is only 50$\frac{W}{m^2}$ for latent and sensible heat. With the exception of extremely sandy soils, the use of soil texture class as a predictor variable explains only 5% of the variance in model outputs, thus soil texture class should not be used to determine SHPs for land surface modeling.
H12A-06 11:35h
Modelling of Sub-daily Hydrological Processes Using Daily Time-Step Models: A Distribution Function Approach to Temporal Scaling
Mismatches in scale between the fundamental processes, the model and supporting data are a major limitation in hydrologic modelling. Surface runoff generation via infiltration excess and the process of soil erosion are fundamentally short time-scale phenomena and their average behaviour is mostly determined by the short time-scale peak intensities of rainfall. Ideally, these processes should be simulated using time-steps of the order of minutes to appropriately resolve the effect of rainfall intensity variations. However, sub-daily data support is often inadequate and the processes are usually simulated by calibrating daily (or even coarser) time-step models. Generally process descriptions are not modified but rather effective parameter values are used to account for the effect of temporal lumping, assuming that the effect of the scale mismatch can be counterbalanced by tuning the parameter values at the model time-step of interest. Often this results in parameter values that are difficult to interpret physically. A similar approach is often taken spatially. This is problematic as these processes generally operate or interact non-linearly. This indicates a need for better techniques to simulate sub-daily processes using daily time-step models while still using widely available daily information. A new method applicable to many rainfall-runoff-erosion models is presented. The method is based on temporal scaling using statistical distributions of rainfall intensity to represent sub-daily intensity variations in a daily time-step model. This allows the effect of short time-scale nonlinear processes to be captured while modelling at a daily time-step, which is often attractive due to the wide availability of daily forcing data. The approach relies on characterising the rainfall intensity variation within a day using a cumulative distribution function (cdf). This cdf is then modified by various linear and nonlinear processes typically represented in hydrological and erosion models. The statistical description of sub-daily variability is thus propagated through the model, allowing the effects of variability to be captured in the simulations. This results in cdfs of various fluxes, the integration of which over a day gives respective daily totals. Using 42-plot-years of surface runoff and soil erosion data from field studies in different environments from Australia and Nepal, simulation results from this cdf approach are compared with the sub-hourly (2-minute for Nepal and 6-minute for Australia) and daily models having similar process descriptions. Significant improvements in the simulation of surface runoff and erosion are achieved, compared with a daily model that uses average daily rainfall intensities. The cdf model compares well with a sub-hourly time-step model. This suggests that the approach captures the important effects of sub-daily variability while utilizing commonly available daily information. It is also found that the model parameters are more robustly defined using the cdf approach compared with the effective values obtained at the daily scale. This suggests that the cdf approach may offer improved model transferability spatially (to other areas) and temporally (to other periods).
H12A-07 11:50h
Sensitivity of the diurnal and seasonal course of modeled runoff to three different land surface model soil moisture parameterizations
Land surface models (LSMs) used in climate modeling include detailed above-ground biophysics but usually lack a good representation of soil hydrological processes. While evapotranspiration can be modeled and measured at a wide range of scales, runoff is a local scale process linked to topography and can only be measured at the catchment-scale. Both processes are closely linked through soil moisture, which is treated as a subgrid-scale process in climate modeling. To explore this connection, catchment-scale LSM simulations are performed with the use of three different soil moisture parameterizations over the Rhone catchment for the years 1986-1988. Results show that the use of a multilayer soil in comparison to the widely used 3-layer soil allows a better reproduction of the seasonal dynamics of runoff. Including lateral soil moisture flow significantly enhances monthly runoff performance and provides an effective means to recover from the dry soil moisture conditions at the end of summer. Snowmelt runoff in the Alpine part of the catchment is sensitive to upscaling and none of the used parameterizations can account for this process. Runoff in the continental part of the Rhone performs well at larger scales without using lateral soil moisture flow. Overall, accuracy in timing and magnitude of simulated runoff is substantially increased by the use of lateral soil moisture flow, especially at the daily time-scale. However evapotranspiration is not sensitive to the different parameterizations of soil moisture processes in the Rhone catchment.
H12A-08 12:05h
The potential of assimilating remotely sensed soil moisture into land surface models
Assimilation of remote sensing data into hydrological modeling has the potential to improve forecasting accuracy; with space-borne, low frequency microwave observations being especially interesting because of its sensitivity to surface soil moisture and its change. However, in conveying the soil moisture information to land surface models, both the brightness temperature and the retrieved soil moisture product suffer from errors introduced by the sensor, atmospheric conditions, and retrieval parameterization. More importantly, under a variety of conditions (e.g. very dry surfaces, heavily vegetated surfaces, or rain conditions during sensor observations) the remote sensing data is not informative. Understanding the statistical dynamics of the errors in both the brightness temperatures and retrieved soil moisture, and the statistical relationships between the surface wetness that influences the microwave signal and the surface land surface modeling layer is fundamental in developing data assimilation procedures that can incorporate space-based radiometric measurements. The work uses statistical methods based on Copula probability distributions to relate space-based soil moisture estimates to either in-situ measurements or land surface model states. The Copula based joint distributions are used to generate ensembles, which are then assimilated into the upper 10-cm soil layer of the Variable Infiltration Capacity land surface model using an Ensemble Kalman Filter (EnKF). The remotely sensed soil moisture product is from a five-year, daily retrieval using measurements from the TRMM 10.7 GHz Microwave Imager (TMI).