Agricultural Engineering and Agronomy

Specific catchment area (SCA) patterns are commonly computed on grids using flow direction algorithms that treat the flow as coming from a point source at the pixel centre. These algorithms are all ambiguous in the definition of the flow width to be associated with a pixel when computing the SCA. Different methods for computing the flow width have been suggested, without giving an objective reason. In the few cases where this issue has been specifically discussed, the flow width is derived from subjective analysis and incorrect conceptualizations. This paper evaluates alternative approaches for defining the flow width when computing SCA patterns using the D∞ and D8 algorithms, by comparing theoretical and computed SCA patterns on sloping planes, inward and outward cones. Two new methods of defining the flow width are also analysed for both the D∞ and D8 algorithms. The performances of the different methods are discussed in relation to two dimensionless parameters: (1) the global resolution, defined as the ratio of a characteristic length of the study area to the grid size and (2) the upslope area resolution, defined as the ratio of the theoretical SCA to the grid size. The optimal methods are identified by specific threshold values of these dimensionless parameters. We conclude that assuming the flow width invariant and equal to the grid size is generally the best approach in most practical circumstances, both for the D∞ and D8 algorithm. Copyright © 2005 John Wiley & Sons, Ltd.

The temporal dynamics of the soil water content in the unsaturated zone is the dominant control on the groundwater recharge and on the partitioning of the rainfall in infiltration and runoff. A numerical solution of Richards’ equation can offer a detailed simulation of the water dynamics in a soil profile, when detailed data on soil hydraulic properties, climate and boundary conditions are known. Such detailed data are seldom available even at the plot scale. Moreover Richards-based models become intractable when the water budget analysis is extended from the plot scale to the catchment scale and therefore information on the spatial variability in soil hydraulic properties and boundary conditions need to be incorporated. To overcome these limitations several models have been proposed for resolving the water budget at the plot scale with a simplified approach. The performance of these water budget models have been tested against field data only in a few cases that are not comprehensive for all the climatic and soil-vegetation systems. These issues motivated the collection of detailed field data in an experimental plot (15 m x 40 m) in the Agri River Basin of the Basilicata Region in Southern Italy. The plot has a slope of 14% along the longer side and it is hydraulically isolated on lateral boundaries. The plot is covered by short grass. Hydraulic properties have been evaluated from soil cores collected at three different depths (0-15, 20-40 and 40-60 cm) in ten different sites of the plot. Within the same range of depths soil moisture has been monitored with TDR probes in six locations and with a neutron probe in three locations, during the span of one year. Climatic conditions have been monitored with a complete meteorological station. Runoff occurrences have been also observed. This data set has been used to parameterize and initialize both a Richards based model and simplified semi-analytical solutions of the soil water budget proposed in literature. The performance of the different water budget models has been evaluated by comparison of the simulated and measured data at different temporal scales.

Distributed hydrological models employ digital terrain analysis algorithms to provide the spatial discretisation of the terrain in a network of elemental units as well as to define the element terrain attributes and the lateral connectivity among the elements. There are quite a few terrain analysis techniques proposed in the literature. The most commonly applied are contour-based and grid-based techniques. Distributed hydrological models are generally applied with one specific type of terrain analysis technique, although there is no clear evidence that one technique offers better performance than others. We propose that all the terrain-based networks can be characterised within the same general framework and we develop a process based distributed model able to implement any terrain-based network within this framework. We first applied this model to a 1 km^2 catchment in the North Island of New Zealand, implementing a contour-based element network (element area ranging between 50 m^2 and 2250 m^2, with an average value of 300 m^2). We simulated the catchment flow response and soil moisture spatial and temporal dynamics over a period of two years and we tested the model performance against observed flow and soil moisture patterns. We then applied the same model with the same set of parameters, implementing a 25-m grid-based element network, with a multiple-flow-direction algorithm. The simulated flow results are not significantly different, compared to the overall model performance with respect to the observed flow. Larger differences occur in the simulated soil moisture patterns, particularly in the valley areas, where lateral flow is dominant in controlling the local soil moisture. Catchment average soil moisture is significantly higher for the grid approach because the slope computed in the grid-based network is generally lower than in the contour-based network. Simulated soil moisture variance is also much higher for the grid approach because of the greater fragmentation of the flow-paths in the grid-based element network than in the contour-based network. The conclusion is that the type of discretisation can have important effects on the performance of a distributed hydrological model.

This paper provides a general analytical framework for assessing the dependence existing between spatial rainfall organisation, basin morphology and runoff response. The analytical framework builds upon a set of spatial rainfall statistics (termed "spatial moments of catchment rainfall") which describe the spatial rainfall organisation in terms of concentration and dispersion statistics as a function of the distance measured along the flow routing coordinate. The introduction of these statistics permits derivation of a simple relationship for the quantification of storm velocity at the catchment scale. The paper illustrates the development of the analytical framework and explains the conceptual meaning of the statistics by means of application to five extreme flash floods occurred in various European regions in the period 2002-2007. High resolution radar rainfall fields and a distributed hydrologic model are employed to examine how effective are these statistics in describing the degree of spatial rainfall organisation which is important for runoff modelling. This is obtained by quantifying the effects of neglecting the spatial rainfall variability on flood modelling, with a focus on runoff timing. The size of the study catchments ranges between 36 to 982 km2. The analysis reported here shows that the spatial moments of catchment rainfall can be effectively employed to isolate and describe the features of rainfall spatial organization which have significant impact on runoff simulation. These statistics provide essential information on what space-time scales rainfall has to be monitored, given certain catchment and flood characteristics, and what are the effects of space-time aggregation on flood response modeling.

If predictions of pedotransfer functions (PTFs) are distributed in space, then they require a spatial evaluation. Three aspects of performance can be considered in the evaluation of a spatially distributed PTF: (i) the correlation of observed and predicted quantities across different spatial scales; (ii) the reproduction of observed variance across different spatial scales; and (iii) the spatial pattern of the model error. Further, when there is more than one PTF available, we must be able to choose which is the most appropriate for a particular spatial scale. In this study, we observed soil hydraulic properties (water retention and saturated hydraulic conductivity) at 100 regularly spaced locations on a 5000-m transect in southern Italy. Four PTFs (referred to as Model V, Model S, Model W, and Model L) were then used to predict the hydraulic properties at the sampled locations. We used wavelet analysis to examine how the variances and correlations of the observed and predicted properties varied with spatial scale and location in the landscape. This spatial analysis revealed aspects of variability that could not be investigated under assumptions of stationarity (e.g. by geostatistics). These results included: the underestimation of observed variances at particular spatial scales; the failure of the predictions to correlate adequately with the observed variables at particular spatial scales; and, also, evidence of scale- and location-dependent correlations, which imply that non-spatial correlation coefficients do not suffice to describe the joint spatial variation of the observations and predictions. We also found significant correlations between model errors and auxiliary variables, which indicated how the site-specific predictions of the PTFs might be improved. Finally, we proposed a wavelet concordance correlation to rank the performance of each PTF, thereby enabling the choice of the best PTF at a particular spatial scale. In general terms, we found that a locally calibrated PTF (Model L) yielded the best results.

The quasi-dynamic wetness index, in its original development, was computed by calculating the travel time along all the possible upslope flow paths on a contour-based terrain network. In more recent applications the same approach has been extended to gridded digital elevation models with single-flow-direction algorithms. Multiple-flow-direction algorithms, although more effective in representing flow paths, have not been used because they are not practicable with the established methodology. We propose an alternative method for computing the quasi-dynamic wetness index based on the numerical integration of the linear-kinematic wave equation. This method can be applied to any of the terrain-based flow-direction algorithms currently published. The method is robust and efficient.

A analytical framework is used to characterise five flood events of different type in the Kamp area in Austria: one long-rain event, two short-rain events, one rain-on-snow event and one snowmelt event. Specifically, the framework quantifies the contributions of the space-time variability of rainfall/snowmelt, runoff coefficient, hillslope and channel routing to the flood runoff volume and the delay and spread of the resulting hydrograph. The results indicate that the components obtained by the framework clearly reflect the individual processes which characterise the event types. For the short-rain events, temporal, spatial and movement components can all be important in runoff generation and routing, which would be expected because of their local nature in time and, particularly, in space. For the long-rain event, the temporal components tend to be more important for runoff generation, because of the more uniform spatial coverage of rainfall, while for routing the spatial distribution of the produced runoff, which is not uniform, is also important. For the rain-on-snow and snowmelt events, the spatio-temporal variability terms typically do not play much role in runoff generation and the spread of the hydrograph is mainly due to the duration of the event. As an outcome of the framework, a dimensionless response number is proposed that represents the joint effect of runoff coefficient and hydrograph peakedness and captures the absolute magnitudes of the observed flood peaks.

Retrieving soil moisture profiles from surface data assimilation in soil hydrological models has to cope with the uncertainty in the soil hydraulic parameters. Some studies have proved that, in case of poorly identified soil hydraulic parameters, model predictions may even be worse than model estimates without data assimilation. Therefore, the sequential estimation of both states and parameters from noisy signals is a fundamental prerequisite for suitably addressing the underlying challenge. The dual Extended Kalman Filter (EKF) has proved, at least theoretically, to be an efficient technique for generating approximate maximum-likelihood estimates of both states and parameters of a discrete-time nonlinear dynamic system. The objective of this work is to evaluate the accuracy of the dual EKF for retrieving soil moisture profiles and soil hydraulic parameters simultaneously by the assimilation of moisture-related surface data in a one-dimensional simulation hydrologic model. Parameters under scrutiny were the saturated hydraulic conductivity, Ks, as well as parameters "α" and "n" describing the shape of the van Genuchten water retention relation. The present study provides useful insights on the implementation of the dual EKF into computer models applied to the vadose zone hydrology. The sensitivity of dual EKF results to several computational aspects is evaluated with numerical experiments. An effective procedure is proposed for getting efficient predictions even for high errors of the initial state and parameter values. The parameter convergence is affected more than the depth of the observations within the soil profile than by the initial state and parameter values.

The assessment of rainfall erosivity, as defined by the RUSLE R-factor, depends on availability of sub-hourly rainfall intensity time-series which are seldom available for a significant temporal extent. Even in areas where significant sub-hourly time-series have been recorded, their spatial distribution is too coarse to be interpolated with an acceptable level of uncertainty. We present a four-parameter model for estimating monthly values of the R-factor (Ri) from mean values of the maximum daily rainfall (di) and accumulated rainfall (mi) in each i-th month of the year. The model has been calibrated and validated by using sub-hourly rainfall data from a set of 110 automatic raingauges which have been operating for more than 8 years in Campania Region (Southern Italy). Ri values are estimated by power functions of di and mi, scaled by a two-parameter periodic function. This periodic function has been specifically designed to account for the meteorological conditions affecting the temporal variability of Ri across the year in Mediterranean regions. Ri is mainly controlled by the occurrence of single rainfall events of high intensity and short duration within each month and is highly correlated with the average value of the maximum hourly rainfall of the month (hi). Both Ri and hi exhibit marked peak values in September, while di and mi are relatively more uniformly distributed across the rainy season, from September to April. Meteorological conditions in late summer and early autumn enhance the occurrence of convective storms, characterized by high intensity and short duration, which drop almost the entire rainfall depth observed in one month. In the rest of the rainy season rainfall events are characterized by longer duration and lower average rainfall intensity. Thus the functional relation between Ri and (di, mi) changes from one month to the other. The model simulates this inter-monthly variability by the periodic scaling function. Tests have been performed in the calibration and validation stages, to verify the model efficiency not only at monthly scale, but also at seasonal and annual scales.

An intensive field monitoring programme was conducted in 1998 and 1999 in an 84 ha catchment located on the North Island of New Zealand. The data collected includes six soil moisture patterns, 12 soil moisture time-series, flow at the outlets of two subcatchments of 56 ha and 28 ha, rainfall and other meteorological data. This data set was used in a downward approach to constrain the conceptualisations and the parameters of a terrain-based distributed model, aiming to simulate the spatial and temporal variability of the soil moisture and the flow response observed in the two subcatchments. The principal mechanism producing runoff was assessed by a preliminary data analysis, involving rainfall, flow and soil moisture time-series as well as the simulation of infiltration processes at the point scale. Runoff was identified as being mainly produced by saturation excess across the entire monitoring period, despite the high intensity rainfall observed in that area. The model soil-water-retention parameters were determined from the soil moisture patterns. The other soil parameters controlling the soil transmissivity were determined by calibration against the observed flow in 1999 in the 56 ha subcatchment, accumulated at the daily scale. The analysis of the flow data at the hourly scale illustrated the need for a more complex subsurface transmissivity function in order to produce lateral storm flow with a larger range of celerity. A simple solution was to modify the decay of the lateral transmissivity with the soil moisture content by adding a second component activated only for soil moisture close to saturation. The additional parameters were calibrated against the observed hourly flow in 1999 in the 56 ha subcatchment. The remaining data were used for validation purposes. This data-driven, downward approach to identifying the model conceptualisation and parameters resulted in a model capable of reproducing the observed catchment behaviour while minimising model complexity.

Specific catchment area (SCA) patterns are commonly computed on grids using flow direction algorithms that treat the flow as coming from a point source at the pixel centre. These algorithms are all ambiguous in the definition of the flow width to be associated with a pixel when computing the SCA. Different methods for computing the flow width have been suggested, without giving an objective reason. In the few cases where this issue has been specifically discussed, the flow width is derived from subjective analysis and incorrect conceptualizations. This paper evaluates alternative approaches for defining the flow width when computing SCA patterns using the D and D8 algorithms, by comparing theoretical and computed SCA patterns on sloping planes, inward and outward cones. Two new methods of defining the flow width are also analysed for both the D and D8 algorithms. The performances of the different methods are discussed in relation to two dimensionless parameters: (1) the global resolution, defined as the ratio of a characteristic length of the study area to the grid size and (2) the upslope area resolution, defined as the ratio of the theoretical SCA to the grid size. The optimal methods are identified by specific threshold values of these dimensionless parameters. We conclude that assuming the flow width invariant and equal to the grid size is generally the best approach in most practical circumstances, both for the D and D8 algorithm.