Can satellite land surface temperature data be used similarly to river discharge measurements for distributed hydrological model calibration?

Abstract

A new methodology is proposed for the calibration of distributed hydrological models at the basin scale by constraining an internal model variable using satellite data of land surface temperature (LST). The model algorithm solves the system of energy and mass balances in terms of a representative equilibrium temperature that governs the fluxes of energy and mass over the basin domain. This equilibrium surface temperature, which is a critical model state variable, is compared to operational satellite LST, while calibrating soil hydraulic parameters and vegetation variables differently in each pixel, minimizing the errors. This procedure is compared to the traditional calibration using only discharge measurements. The distributed energy water balance model, Flash-flood Event-based Spatially-distributed rainfall–runoff Transformation – Energy Water Balance model (FEST-EWB), is used to test this approach. This methodology is applied to the Upper Yangtze River basin (China) using MODIS LST retrieved from satellite data in the framework of the NRSCC-ESA DRAGON-2 Programme. The calibration procedure based on LST seems to outperform the calibration based on discharge, with lower relative error and higher Nash-Sutcliffe efficiency index on cumulated volume.

Editor D. Koutsoyiannis; Associate editor C. Perrin     

On the importance of soil moisture in calibration of rainfall–runoff models: two case studies

Streamflow modelling results from the GR4H and PDM hydrological models were evaluated in two Australian sub-catchments, using (1) calibration to streamflow and (2) joint-calibration to streamflow and soil moisture. Soil moisture storage in the models was evaluated against soil moisture observations from field measurements. The PDM had the best performance in terms of both streamflow and soil moisture estimations during the calibration period, but was outperformed by GR4H during validation. It was also shown that the soil moisture estimation was improved significantly by joint-calibration for the case where streamflow and soil moisture estimations were poor. In other cases, addition of the soil moisture constraint did not degrade the results. Consequently, it is recommended that GR4H be used, in preference to the PDM, in the foothills of the Murrumbidgee catchment or other Australian catchments with semi-arid to sub-humid climate, and that soil moisture data be used in the calibration process.     

How well do indicator variograms capture the spatial connectivity of soil moisture?

Indicators are binary transforms of a variable and are 1 or 0, depending on whether the variable is above or below a threshold. Indicator variograms can be used for a similar range of geostatistical estimation techniques as standard variograms. However, they are more flexible as they allow different ranges for small and large values of a hydrological variable. Indicator geostatistics are also sometimes used to represent the connectivity of high values in spatial fields. Examples of connectivity are connected high values of hydraulic conductivity in aquifers, leading to preferential flow, and connected band-shaped saturation zones in catchments. However, to the authors' knowledge the ability of the indicator approach to capture connectivity has never been shown conclusively. Here we analyse indicator variograms of soil moisture in a small south-east Australian catchment and examine how well they can represent connectivity. The indicator variograms are derived from 13 soil moisture patterns, each consisting of 500–2000 point TDR (time domain reflectometry) measurements. Winter patterns are topographically organized with long, thin, highly connected lines of high soil moisture in the drainage lines. In summer the patterns are more random and there is no connectivity of high soil moisture values. The ranges of the 50^th and 90^th percentile indicator semivariograms are approximately 110 and 75 m, respectively, during winter, and 100 and 50 m, respectively, during summer. These ranges indicate that, compared with standard semivariograms, the indicator semivariograms provide additional information about the spatial pattern. However, since the ranges are similar in winter and in summer, the indicator semivariograms were not able to distinguish between connected and unconnected patterns. It is suggested that new statistical measures are needed for capturing connectivity explicitly. © 1998 John Wiley & Sons, Ltd.     

Are spatial patterns of soil moisture at plot scales generalisable across catchments,climates, and other characteristics? A synthesis of synoptic soil moisture across the Mid-Atlantic

The spatial variation of soil moisture over very small areas (<100 m²) can have nonlinear impacts on cycling and flux rates resulting in bias if it is not considered, but measuring this variation is difficult over extensive temporal and spatial scales. Most studies examining spatial variation of soil moisture were conducted at hillslope (0.01 km²) to multi-catchment spatial scales (1000 km²). They found the greatest variation at mid wetness levels and the smallest variation at wet and dry wetness levels forming a concave down relationship. There is growing evidence that concave down relationships formed between spatial variation of soil moisture and average soil moisture are consistent across spatial scales spanning several orders of magnitude, but more research is needed at very small, plot scales (<100 m²). The goal of this study was to characterise spatial variation in shallow soil moisture at the plot scale by relating the mean of measurements collected in a plot to the standard deviation (SD). We combined data from a previous study with thousands of new soil moisture measurements from 212 plots in eight catchments distributed across the US Mid-Atlantic Region to (1) test for a generalisable mean–SD relationship at plot scales, (2) characterise how landcover, land use, season, and hillslope position contribute to differences in mean–SD relationships, and (3) use these generalised mean–SD relationships to quantify their impacts on catchment scale nitrification and denitrification potential. Our study found that 98% of all measurements formed a generalised mean–SD relationship like those observed at hillslope and catchment spatial scales. The remaining 2% of data comprised a mean–SD relationship with greater spatial variation that originated from two riparian plots reported in a previous study. Incorporating the generalised mean–SD relationship into estimates of nitrification and denitrification potential revealed strong bias that was even greater when incorporating mean–SD observations from the two riparian plots with significantly greater spatial variation.     

Is the soil moisture precipitation feedback enhanced by heterogeneity and dry soils? A comparative study

The interaction between the land surface and the atmosphere is a crucial driver of atmospheric processes. Soil moisture and precipitation are key components in this feedback. Both variables are intertwined in a cycle, that is, the soil moisture – precipitation feedback for which involved processes and interactions are still discussed. In this study the soil moisture – precipitation feedback is compared for the sempiternal humid Ammer catchment in Southern Germany and for the semiarid to subhumid Sissili catchment in West Africa during the warm season, using precipitation datasets from the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS), from the German Weather Service (REGNIE) and simulation datasets from the Weather Research and Forecasting (WRF) model and the hydrologically enhanced WRF-Hydro model. WRF and WRF-Hydro differ by their representation of terrestrial water flow. With this setup we want to investigate the strength, sign and variables involved in the soil moisture – precipitation feedback for these two regions. The normalized model spread between the two simulation results shows linkages between precipitation variability and diagnostic variables surface fluxes, moisture flux convergence above the surface and convective available potential energy in both study regions. The soil moisture – precipitation feedback is evaluated with a classification of soil moisture spatial heterogeneity based on the strength of the soil moisture gradients. This allows us to assess the impact of soil moisture anomalies on surface fluxes, moisture flux convergence, convective available potential energy and precipitation. In both regions the amount of precipitation generally increases with soil moisture spatial heterogeneity. For the Ammer region the soil moisture – precipitation feedback has a weak negative sign with more rain near drier patches while it has a positive signal for the Sissili region with more rain over wetter patches. At least for the observed moderate soil moisture values and the spatial scale of the Ammer region, the spatial variability of soil moisture is more important for surface-atmosphere interactions than the actual soil moisture content. Overall, we found that soil moisture heterogeneity can greatly affect the soil moisture – precipitation feedback.     

Could CMIP6 climate models reproduce the early-2000s global warming slowdown?

The unexpected global warming slowdown during 1998–2013 challenges the existing scientific understanding of global temperature change mechanisms, and thus the simulation and prediction ability of state-of-the-art climate models since most models participating in phase 5 of the Coupled Model Intercomparison Project(CMIP5) cannot simulate it. Here, we examine whether the new-generation climate models in CMIP6 can reproduce the recent global warming slowdown, and further evaluate their capacities for simulating key-scale natural variabilities which are the most likely causes of the slowdown. The results show that although the CMIP6 models present some encouraging improvements when compared with CMIP5, most of them still fail to reproduce the warming slowdown. They considerably overestimate the warming rate observed in 1998–2013,exhibiting an obvious warming acceleration rather than the observed deceleration. This is probably associated with their deficiencies in simulating the distinct temperature change signals from the human-induced long-term warming trend and/or the three crucial natural variabilities at interannual, interdecadal, and multidecadal scales. In contrast, the 4 models that can successfully reproduce the slowdown show relatively high skills in simulating the long-term warming trend and the three keyscale natural variabilities. Our work may provide important insight for the simulation and prediction of near-term climate changes.     

Copula models for frequency analysis what can be learned from a Bayesian perspective?

Large spring floods in the Québec region exhibit correlated peakflow, duration and volume. Consequently, traditional univariate hydrological frequency analyses must be complemented by multivariate probabilistic assessment to provide a meaningful design flood level as requested in hydrological engineering (based on return period evaluation of a single quantity of interest). In this paper we study 47 years of a peak/volume dataset for the Romaine River with a parametric copula model. The margins are modeled with a normal or gamma distribution and the dependence is depicted through a parametric family of copulas (Arch 12 or Arch 14). Parameter joint inference and model selection are performed under the Bayesian paradigm. This approach enlightens specific features of interest for hydrological engineering: (i) cross correlation between margin parameters are stronger than expected , (ii) marginal distributions cannot be forgotten in the model selection process and (iii) special attention must be addressed to model validation as far as extreme values are of concern.     

How much complexity is needed to simulate watershed streamflow and water quality? A test combining time series and hydrological models

Modelled hydrologic processes are represented in a set of numerical equations; the complexity of which can be measured by the total number of variables needed. A single dominant hydrologic process could control the hydrologic response of a watershed, and so the identification of the corresponding dominant variable(s) would aid in identifying a parsimonious model and in collecting more reliable data. By accounting for both model complexity and serial correlation in the variables, a model is used to identify the dominant variables for representing watershed scale streamflow, sediment transport and phosphorus yields. Long‐term water quantity and quality data were used to show that rainfall and non‐linear soil water storage were the dominant variables for weekly streamflow, suspended sediment and particulate phosphorus. Model accuracy did not consistently improve when other statistically significant variables were included. The results suggest that improved model performance may not justify the added model complexity. As such, identification of dominant variables would be the priority for developing parsimonious hydrologic models, especially at watershed scales. Copyright © 2013 John Wiley & Sons, Ltd.     

Can the tropical storms originated from the Bay of Bengal impact the precipitation and soil moisture over the Tibetan Plateau?

This study investigates the impacts of tropical storms originated from the Bay of Bengal(BOBTSs) on the precipitation and soil moisture over the Tibetan Plateau(TP) in April–June(AMJ) and September–December(SOND) during 1981–2011 based on the best track dataset provided by Joint Typhoon Warning Centre(JTWC). Results indicate that there are about 1.35 BOBTSs influence the TP in each year and most of them occurred in May and October, and the BOBTSs in AMJ influence the TP with larger extension and higher latitudes than those in SOND. The maximum regional precipitation induced by the BOBTSs accounts for more than 50% for the total precipitation in the corresponding month and about 20% for the season. Further analysis reveals that the surface soil moisture anomalies induced by the BOBTSs can persist only 20–25 days in AMJ, and the case is also true for the snow depth in SOND. Numerical simulations by using the regional climate model of Weather Research and Forecasting(WRF) suggest that the soil moisture anomalies in the sub-surface can last 2 months whereas for the surface it can persist only about 20 days, which agrees well with the observation analysis. Overall, the effect of the preceding BOBTSs on the snow depth and soil moisture anomalies over the TP cannot maintain to summer, and there is no robust connection between the BOBTSs and summer precipitation anomalies in East China. Moreover, since the mid-1990 s, the spring rainfall induced by the BOBTSs over the TP seems to be enhanced to a certain degree because of the intensified BOBTSs.     

Could tsunami risk be underestimated using core‐based reconstructions? Lessons from ground penetrating radar

Where should we take cores for palaeotsunami research? It is generally considered that local depressions with low energy environments such as wetlands are one of the best places. However, it is also recognized that the presence or absence of palaeotsunami deposits (and their relative thickness) is highly dependent upon subsoil microtopography. In the beach ridge system of Ishinomaki Plain, Japan, several palaeotsunami deposits linked to past Japan Trench earthquakes have been reported. However, the number of palaeotsunami deposits reported at individual sites varies considerably. This study used ground penetrating radar (GPR) combined with geological evidence to better understand the relationship between palaeotopography and palaeotsunami deposit characteristics. The subsurface topography of the ~3000–4000 bp beach ridge was reconstructed using GPR data coupled with core surveys of the underlying sediments. We noted that the number (and thickness) of the palaeotsunami deposits inferred from the cores was controlled by the palaeotopography. Namely, a larger number of events and thicker palaeotsunami deposits were observed in depressions in the subsurface microtopography. We noted a total of three palaeotsunami deposits dated to between 1700 and 3000 cal bp , but they were only observed together in 11% of the core sites. This result is important for tsunami risk assessments that use the sedimentary evidence of past events because we may well be underestimating the number of tsunamis that have occurred. We suggest that GPR is an efficient and invaluable tool to help researchers identify the most appropriate places to carry out geological fieldwork in order to provide a more comprehensive understanding of past tsunami activity. Copyright © 2017 John Wiley & Sons, Ltd.     

Could surface roughness be a poor proxy for landslide age? Results from the Swabian Alb,Germany

The potential of surface roughness to quantify geomorphological landforms and processes has been enhanced with the availability of high‐resolution digital terrain models (DTM). Recent studies that attempt to identify landslide features with surface roughness have suggested that this measure of topographic heterogeneity may also be applied to estimate the relative age of landslides. This is a provisional study that explores the potential of this relationship by assessing the ability of surface roughness to act as a proxy for relative landslide age. The surface roughness for a set of 12 dated landslides in the Swabian Alb that occurred between 1789 and 1985 was calculated from a 1 m² spatial resolution LiDAR DTM with three algorithms: root‐mean‐square‐height (RMSH), standard deviation of slope (SDS), and direction cosine eigenvalue ratios (DCE). Scale‐dependence was analysed by calculating surface roughness for a range of moving window sizes (3 × 3, 5 × 5, 9 × 9 and 15 × 15), and surface roughness for each landslide was summarized by the median and upper quartile. Only weak correlations (best Spearman's rho 0.58) were present between landslide age and surface roughness. This correlation becomes weaker with increasing moving window size. Given weak observed associations and discussed challenges pertaining to the complexities of landslide morphology change over time, we currently find that surface roughness alone may not be justifiable to act as a proxy for landslide age for our study region. Furthermore, we recommend future studies should focus on addressing possible natural and anthropogenic factors such as land use change that may alter surface roughness. These studies may focus on one of the three roughness measures used here as they are strongly correlated. Copyright © 2014 John Wiley & Sons, Ltd.     

Do French macroseismic intensity observations agree with expectations from the European Seismic Hazard Model 2013?

Probabilistic seismic hazard assessments are the basis of modern seismic design codes. To test fully a seismic hazard curve at the return periods of interest for engineering would require many thousands of years’ worth of ground-motion recordings. Because strong-motion networks are often only a few decades old (e.g. in mainland France the first accelerometric network dates from the mid-1990s), data from such sensors can be used to test hazard estimates only at very short return periods. In this article, several hundreds of years of macroseismic intensity observations for mainland France are interpolated using a robust kriging-with-a-trend technique to establish the earthquake history of every French mainland municipality. At 24 selected cities representative of the French seismic context, the number of exceedances of intensities IV, V and VI is determined over time windows considered complete. After converting these intensities to peak ground accelerations using the global conversion equation of Caprio et al. (Ground motion to intensity conversion equations (GMICEs): a global relationship and evaluation of regional dependency, Bulletin of the Seismological Society of America 105:1476–1490, 2015), these exceedances are compared with those predicted by the European Seismic Hazard Model 2013 (ESHM13). In half of the cities, the number of observed exceedances for low intensities (IV and V) is within the range of predictions of ESHM13. In the other half of the cities, the number of observed exceedances is higher than the predictions of ESHM13. For intensity VI, the match is closer, but the comparison is less meaningful due to a scarcity of data. According to this study, the ESHM13 underestimates hazard in roughly half of France, even when taking into account the uncertainty in the conversion from intensity to acceleration. However, these results are valid only for the acceleration range tested in this study (0.01 to 0.09 g).     

Crossing the rural–urban boundary in hydrological modelling: How do conceptual rainfall–runoff models handle the specificities of urbanized catchments?

Landscape differences induced by urbanization have prompted hydrologists to define a fuzzy boundary between rural- and urban-specific hydrological models. We addressed the validity of establishing this boundary, by testing two rural models on a large sample of 175 French and United States (US) urbanized catchments, and their 175 rural neighbours. The impact of urbanization on the hydrological behaviour was checked using four metrics. Using a split-sample test, we have compared the performances, parameter distributions, and internal fluxes of GR4H and IHACRES, two conceptual and continuous models running at the hourly time step. Both model structures are based on soil moisture accounting reservoirs (infiltration, runoff, and actual evapotranspiration) and quick flow/slow flow routing components, with no consideration of any specific feature related to urbanization. Results showed: (a) Except for the ratio of streamflow flashiness to precipitation flashiness, the range of hydrological signature metrics in rural catchments encompassed the specificities of urbanized ones. Overall, the urbanized catchments showed higher ratios of mean streamflow to mean precipitation (median values: 0.39 vs. 0.27) and streamflow flashiness to precipitation flashiness (0.13 vs. 0.03), besides lower baseflow index (0.42 vs. 0.62) and shorter characteristic response time (3 vs. 14 hr). (b) The performances of GR4H revealed no significant distinction between rural and urbanized catchments in terms of Kling–Gupta Efficiency (KGE), whereas IHACRES better simulated urbanized catchments, especially during summer. (c) With respect to differences in urbanization level, the GR4H and IHACRES parameters showed different distributions. The differences in parameters were consistent with the differences in hydrological behaviour, which is promising for a model-based assessment of the impact of urbanization. (d) The models agreed less in reproducing the internal fluxes over the urbanized catchments than over the rural ones. These results demonstrate the flexibility of conceptual models to handle the specificities of urbanized catchments.     

Could ocean currents be responsible for the west to east spread of aquatic invasive species in Maritime Canadian waters?

The circulation in the shelf seas of Maritime Canada is predominantly in the northeast–southwest direction. Despite the mean northeast–southwest flow, a number of AIS invasions have been observed to proceed in the opposite direction – from the Gulf of Maine, around Nova Scotia, and into the southern Gulf of St. Lawrence. Flow fields from a numerical circulation model are used to investigate whether these invasions could be due to drift in ocean currents. Particle tracking experiments are performed and probability density functions (PDFs) derived that describe the probability of drifting a given upstream distance in a given drift time. Analysis of these PDFs revealed that for invasions that took 20–40 y to occur, propagule drift in ocean currents could be responsible for the upstream spread, while this was not the case for short timescale invasions (<10 y). Rafting could be responsible for both short and long timescale invasions.     

Should average shear-wave velocity in the top 30 m of soil be used to describe seismic amplification?

The average velocity of shear waves in the top 30 m of soil, ν_L, has become the parameter used by many engineering design codes and most recently by published empirical-scaling equations to estimate the amplitudes of strong ground motion. Yet there are few studies to determine whether this is a meaningful parameter to use—and whether estimates that do use it are reliable. In 1995, the authors studied this problem and concluded that ν_L should not be used. We reported then that an older site characterization in terms of the soil site parameter proposed by Seed et al. [1], s_L, worked better because it included a measure of the thickness of the soil layers to considerably greater depths. Our report, however, made no difference; numerous papers continued to be published based only on scaling in terms of ν_L, and worse, they also ignored the geological site conditions. The purpose of this paper is to emphasize that the average shear-wave velocity in the top 30 m of soil should not be the only site parameter used to scale strong-motion amplitudes. While the search continues for the more meaningful site parameters to use in empirical scaling of strong earthquake ground motion, it is better to use s_L to describe the amplification of seismic waves by soil deposits near the surface.     

Do mixing models with different input requirement yield similar streamflow source contributions? Case study: A tropical montane catchment

Hydrogeochemical based mixing models have been successfully used to investigate the composition and source identification of streamflow. The applicability of these models is limited due to the high costs associated with data collection and the hydrogeochemical analysis of water samples. Fortunately, a variety of mixing models exist, requiting different amount of data as input, and in data scarce regions it is likely that preference will be given to models with the lowest requirement of input data. An unanswered question is if models with high or low input requirement are equally accurate. To this end, the performance of two mixing models with different input requirement, the mixing model analysis (MMA) and the end-member mixing analysis (EMMA), were verified on a tropical montane headwater catchment (21.7 km²) in the Ecuadorian Andes. Nineteen hydrogeochemical tracers were measured on water samples collected weekly during 3 years in streamflow and eight potential water sources or end-members (precipitation, lake water, soil water from different horizons and springs). Results based on 6 conservative tracers, revealed that EMMA (using all tracers) and MMA (using pair-combinations out of the 6 conservative ones), identified the same end-members: rainfall, soil water and spring water., as well as, similar contribution fractions to streamflow from rainfall 21.9% and 21.4%, soil water 52.7% and 52.3%, and spring water 26.1% and 28.7%, respectively. Our findings show that a hydrogeochemical mixing model requiring a few tracers can provide similar outcomes than models demanding more tracers as input data. This underlines the value of a preliminary detailed hydrogeochemical characterization as basis to derive the most cost-efficient monitoring strategy.     

What is wrong with post-fire soil erosion modelling? A meta-analysis on current approaches,research gaps,and future directions

In the near future, a higher occurrence of wildfires is expected due to climate change, carrying social, environmental, and economic implications. Such impacts are often associated with an increase of post-fire hydrological and erosive responses, which are difficult to predict. Soil erosion models have been proven to be a valuable tool in the decision-making process, from emergency response to long-term planning, however, they were not designed for post-fire conditions, so need to be adapted to include fire-induced changes. In recent years, there have been an increasing number of studies testing different models and adaptations for the prediction of post-fire soil erosion. However, many of these adaptations are being applied without field validation or model performance assessment. Therefore, this study aims to describe the scientific advances in the last 20 years in post-fire soil erosion modelling research and evaluate model adaptations to burned areas that aim to include: (i) fire-induced changes in soil and ground cover; (ii) fire-induced changes in infiltration; (iii) burn severity; and (iv) mitigation measures in their predictions. This study also discusses the strengths and weaknesses of these approaches, suggests potential improvements, and identifies directions for future research. Results show that studies are not homogeneously distributed worldwide, according to the model type used or by region most affected by wildfire. During calibration, 73% of cases involved model adaptation to burned conditions, and only 21% attempted to accommodate new processes. Burn severity was addressed in 75% of cases, whilst mitigation measures were simulated in 27%. Additionally, only a minor percentage of model predictions were validated with independent field data (17%) or assessed for uncertainties (13%). Therefore, further efforts are required in the adaptation of erosion models to burned conditions, to be widely used for post-fire management decisions. © 2020 John Wiley & Sons, Ltd.     

Monitoring of the plume from the basaltic phreatomagmatic 2004 Grímsv?tn eruption—application of weather radar and comparison with plume models

The Grímsv?tn eruption in November 2004 belongs to a class of small- to medium-sized phreatomagmatic eruptions which are common in Iceland. The eruption lasted 6?days, but the main phase, producing most of the 0.02?km³ of magma erupted, was visible for 33?h on the C-band weather radar of the Icelandic Meteorological Office located in Keflavík, 260?km to the west of the volcano. The plume rose to 8–12?km high over sea level during 33?h. The long distance between radar and source severely reduces the accuracy of the plume height determinations, causing 3.5-km steps in recorded heights. Moreover, an apparent height overestimate of ~1.5?km in the uncorrected radar records occurs, possibly caused by wave ducting or super-refraction in the atmosphere. The stepping and the height overestimate can be partly overcome by averaging the plume heights and by applying a height adjustment based on direct aircraft measurements. Adjusted weather radar data on plume height are used to estimate the total mass erupted using empirical plume models mostly based on magmatic eruptions and to compare it with detailed in situ measurements of the mass of erupted tephra. The errors arising because of the large radar plume distance limit the applicability of the data for detailed comparisons. However, the results indicate that the models overestimate the mass erupted by a factor of three to four. This supports theoretical models indicating that high steam content of phreatomagmatic (wet) plumes enhances their height compared to dry plumes.