Large-scale Land Surface Models Research Articles

Simulation of groundwater-flow dynamics in the U.S. Northern High Plains driven by multi-model estimates of surficial aquifer recharge

There is growing interest in incorporating higher-resolution groundwater modeling within the framework of large-scale land surface models (LSMs), including processes such as three-dimensional flow, variable soil saturation, and surface water/groundwater interactions. Conversely, complex groundwater models (e.g., the U.S. Geological Survey Groundwater-Flow Model, MODFLOW) often use simpler representations of land surface dynamics (e.g., surface vegetation, evapotranspiration, recharge) and may benefit from higher process fidelity and temporal resolutions in these inputs. This study investigates the potential of improving groundwater representation in LSMs and land surface dynamics in MODFLOW through forcing MODFLOW with recharge from LSMs. Groundwater simulations build on an existing and well-calibrated MODFLOW model of the U.S. Northern High Plains aquifer, a hydrologically complex basin under the dual impacts of conversion of native vegetation to intense irrigated agricultural fields and climate change. Simulated groundwater recharge from four different land models are used to drive MODFLOW groundwater simulations. Results show relatively large discrepancies between recharge estimates among simulations. Forcing MODFLOW using recharge simulated by some of the LSMs in place of a simple water balance model marginally improves MODFLOW groundwater simulation. Further, our results support the efficacy of coupling LSMs to a sophisticated groundwater model such as MODFLOW. The coupling results in notable improvements in matching the historical groundwater levels through reduction of the skewness coefficient in percent bias histogram (from 1.50 and 1.41 in original LSMs to 0.44 and 0.27, respectively, when MODFLOW is forced by groundwater recharge from LSMs) and reduction of bias. This modeling effort seeks to identify the best compromise between comprehensive land surface processes from global LSMs and advanced representation of groundwater from regional models.

Optimizing a backscatter forward operator using Sentinel-1 data over irrigated land

Abstract. Worldwide, the amount of water used for agricultural purposes is rising, and the quantification of irrigation is becoming a crucial topic. Because of the limited availability of in situ observations, an increasing number of studies is focusing on the synergistic use of models and satellite data to detect and quantify irrigation. The parameterization of irrigation in large-scale land surface models (LSMs) is improving, but it is still hampered by the lack of information about dynamic crop rotations, or the extent of irrigated areas, and the mostly unknown timing and amount of irrigation. On the other hand, remote sensing observations offer an opportunity to fill this gap as they are directly affected by, and hence potentially able to detect, irrigation. Therefore, combining LSMs and satellite information through data assimilation can offer the optimal way to quantify the water used for irrigation. This work represents the first and necessary step towards building a reliable LSM data assimilation system which, in future analysis, will investigate the potential of high-resolution radar backscatter observations from Sentinel-1 to improve irrigation quantification. Specifically, the aim of this study is to couple the Noah-MP LSM running within the NASA Land Information System (LIS), with a backscatter observation operator for simulating unbiased backscatter predictions over irrigated lands. In this context, we first tested how well modelled surface soil moisture (SSM) and vegetation estimates, with or without irrigation simulation, are able to capture the signal of aggregated 1 km Sentinel-1 backscatter observations over the Po Valley, an important agricultural area in northern Italy. Next, Sentinel-1 backscatter observations, together with simulated SSM and leaf area index (LAI), were used to optimize a Water Cloud Model (WCM), which will represent the observation operator in future data assimilation experiments. The WCM was calibrated with and without an irrigation scheme in Noah-MP and considering two different cost functions. Results demonstrate that using an irrigation scheme provides a better calibration of the WCM, even if the simulated irrigation estimates are inaccurate. The Bayesian optimization is shown to result in the best unbiased calibrated system, with minimal chances of having error cross-correlations between the model and observations. Our time series analysis further confirms that Sentinel-1 is able to track the impact of human activities on the water cycle, highlighting its potential to improve irrigation, soil moisture, and vegetation estimates via future data assimilation.

Land surface modeling over the Dry Chaco: the impact of model structures, and soil, vegetation and land cover parameters

Abstract. In this study, we tested the impact of a revised set of soil, vegetation and land cover parameters on the performance of three different state-of-the-art land surface models (LSMs) within the NASA Land Information System (LIS). The impact of this revision was tested over the South American Dry Chaco, an ecoregion characterized by deforestation and forest degradation since the 1980s. Most large-scale LSMs may lack the ability to correctly represent the ongoing deforestation processes in this region, because most LSMs use climatological vegetation indices and static land cover information. The default LIS parameters were revised with (i) improved soil parameters, (ii) satellite-based interannually varying vegetation indices (leaf area index and green vegetation fraction) instead of climatological vegetation indices, and (iii) yearly land cover information instead of static land cover. A relative comparison in terms of water budget components and “efficiency space” for various baseline and revised experiments showed that large regional and long-term differences in the simulated water budget partitioning relate to different LSM structures, whereas smaller local differences resulted from updated soil, vegetation and land cover parameters. Furthermore, the different LSM structures redistributed water differently in response to these parameter updates. A time-series comparison of the simulations to independent satellite-based estimates of evapotranspiration and brightness temperature (Tb) showed that no LSM setup significantly outperformed another for the entire region and that not all LSM simulations improved with updated parameter values. However, the revised soil parameters generally reduced the bias between simulated surface soil moisture and pixel-scale in situ observations and the bias between simulated Tb and regional Soil Moisture Ocean Salinity (SMOS) observations. Our results suggest that the different hydrological responses of various LSMs to vegetation changes may need further attention to gain benefits from vegetation data assimilation.

Averaging over spatiotemporal heterogeneity substantially biases evapotranspiration rates in a mechanistic large-scale land evaporation model

Abstract. Evapotranspiration (ET) influences land–climate interactions, regulates the hydrological cycle, and contributes to the Earth's energy balance. Due to its feedback to large-scale hydrological processes and its impact on atmospheric dynamics, ET is one of the drivers of droughts and heatwaves. Existing land surface models differ substantially, both in their estimates of current ET fluxes and in their projections of how ET will evolve in the future. Any bias in estimated ET fluxes will affect the partitioning between sensible and latent heat and thus alter model predictions of temperature and precipitation. One potential source of bias is the so-called “aggregation bias” that arises whenever nonlinear processes, such as those that regulate ET fluxes, are modeled using averages of heterogeneous inputs. Here we demonstrate a general mathematical approach to quantifying and correcting for this aggregation bias, using the GLEAM land evaporation model as a relatively simple example. We demonstrate that this aggregation bias can lead to substantial overestimates in ET fluxes in a typical large-scale land surface model when sub-grid heterogeneities in land surface properties are averaged out. Using Switzerland as a test case, we examine the scale dependence of this aggregation bias and show that it can lead to an average overestimation of daily ET fluxes by as much as 10 % across the whole country (calculated as the median of the daily bias over the growing season). We show how our approach can be used to identify the dominant drivers of aggregation bias and to estimate sub-grid closure relationships that can correct for aggregation biases in ET estimates, without explicitly representing sub-grid heterogeneities in large-scale land surface models.

A spatially explicit database of wind disturbances in European forests over the period 2000–2018

Abstract. Strong winds may uproot and break trees and represent a major natural disturbance for European forests. Wind disturbances have intensified over the last decades globally and are expected to further rise in view of the effects of climate change. Despite the importance of such natural disturbances, there are currently no spatially explicit databases of wind-related impact at a pan-European scale. Here, we present a new database of wind disturbances in European forests (FORWIND). FORWIND is comprised of more than 80 000 spatially delineated areas in Europe that were disturbed by wind in the period 2000–2018 and describes them in a harmonized and consistent geographical vector format. The database includes all major windstorms that occurred over the observational period (e.g. Gudrun, Kyrill, Klaus, Xynthia and Vaia) and represents approximately 30 % of the reported damaging wind events in Europe. Correlation analyses between the areas in FORWIND and land cover changes retrieved from the Landsat-based Global Forest Change dataset and the MODIS Global Disturbance Index corroborate the robustness of FORWIND. Spearman rank coefficients range between 0.27 and 0.48 (p value < 0.05). When recorded forest areas are rescaled based on their damage degree, correlation increases to 0.54. Wind-damaged growing stock volumes reported in national inventories (FORESTORM dataset) are generally higher than analogous metrics provided by FORWIND in combination with satellite-based biomass and country-scale statistics of growing stock volume. The potential of FORWIND is explored for a range of challenging topics and scientific fields, including scaling relations of wind damage, forest vulnerability modelling, remote sensing monitoring of forest disturbance, representation of uprooting and breakage of trees in large-scale land surface models, and hydrogeological risks following wind damage. Overall, FORWIND represents an essential and open-access spatial source that can be used to improve the understanding, detection and prediction of wind disturbances and the consequent impacts on forest ecosystems and the land–atmosphere system. Data sharing is encouraged in order to continuously update and improve FORWIND. The dataset is available at https://doi.org/10.6084/m9.figshare.9555008 (Forzieri et al., 2019).

Detecting global irrigated areas by using satellite and reanalysis products

Despite the importance of irrigation in meeting the world's food demand and as an essential human modification to water and energy cycles, the reliable extent and distribution of the global irrigated areas remain undefined. In this study, an intuitive method is proposed, based on the aftereffects of irrigation, to detect global irrigated areas by combining satellite and reanalysis datasets. The proposed methodology assumes that irrigation is an unmodeled land surface process, while satellite observations can effectively detect irrigation signals in near real-time. The spatial extents of irrigation were derived by calculating the difference between the remotely sensed and reanalysis datasets. To detect the irrigated areas, three irrigation-dependent variables, soil moisture (SM), land surface temperature (LST), and surface albedo (AL), were used. In the absence of reliable ground truths, the proposed irrigation map was compared to the commonly used global irrigation maps, namely Global Map of Irrigated areas, Global Irrigated Area Map, and recently developed Global Irrigated Areas by Meier et al. (2018). Individual detection by SM, LST, and AL has discrepancies in detecting irrigation signals in highly irrigated, urbanized, and semi-arid regions. However, by combining the individual detection maps, the proposed method showed reasonable agreement with the reference irrigated maps overlapping with approximately 70% of the irrigated areas. We believe that the proposed method, as stand-alone or in combination with the existing irrigation maps, will benefit the studies regarding water and energy balance closure in near-real time for large-scale land surface models by minimizing the uncertainties in model parameterization.

A Framework for Diagnosing Factors Degrading the Streamflow Performance of a Soil Moisture Data Assimilation System

Abstract Data assimilation (DA) techniques have been widely applied to assimilate satellite-based soil moisture (SM) measurements into hydrologic models to improve streamflow simulations. However, past studies have reached mixed conclusions regarding the degree of runoff improvement achieved via SM state updating. In this study, a synthetic diagnostic framework is designed to 1) decompose the random error components in a hydrologic simulation, 2) quantify the error terms that originate from SM states, and 3) assess the effectiveness of SM DA to correct these random errors. The general framework is illustrated through a case study in which surface Soil Moisture Active Passive (SMAP) data are assimilated into a large-scale land surface model in the Arkansas–Red River basin. The case study includes systematic error in the simulated streamflow that imposes a first-order limit on DA performance. In addition, about 60% of the random runoff error originates directly from rainfall and cannot be corrected by SM DA. In particular, fast-response runoff dominates in much of the basin but is relatively unresponsive to state updating. Slow-response runoff is strongly controlled by the bottom-layer SM and therefore only modestly improved via the assimilation of surface measurements. Combined, the total runoff improvement in the synthetic analysis is small (<10% over the basin). Improvements in the real SMAP-assimilated case are further limited due to systematic error and other factors such as inaccurate error assumptions and SMAP rescaling. Findings from the diagnostic framework suggest that SM DA alone is insufficient to substantially improve streamflow estimates in large basins.

Spatial variation of the rain\u2013snow temperature threshold across the Northern Hemisphere

Despite the importance of precipitation phase to global hydroclimate simulations, many land surface models use spatially uniform air temperature thresholds to partition rain and snow. Here we show, through the analysis of a 29-year observational dataset (n = 17.8 million), that the air temperature at which rain and snow fall in equal frequency varies significantly across the Northern Hemisphere, averaging 1.0 °C and ranging from –0.4 to 2.4 °C for 95% of the stations. Continental climates generally exhibit the warmest rain–snow thresholds and maritime the coolest. Simulations show precipitation phase methods incorporating humidity perform better than air temperature-only methods, particularly at relative humidity values below saturation and air temperatures between 0.6 and 3.4 °C. We also present the first continuous Northern Hemisphere map of rain–snow thresholds, underlining the spatial variability of precipitation phase partitioning. These results suggest precipitation phase could be better predicted using humidity and air temperature in large-scale land surface model runs.

Simulating damage for wind storms in the land surface model ORCHIDEE-CAN (revision 4262)

Abstract. Earth system models (ESMs) are currently the most advanced tools with which to study the interactions among humans, ecosystem productivity, and the climate. The inclusion of storm damage in ESMs has long been hampered by their big-leaf approach, which ignores the canopy structure information that is required for process-based wind-throw modelling. Recently the big-leaf assumptions in the large-scale land surface model ORCHIDEE-CAN were replaced by a three-dimensional description of the canopy structure. This opened the way to the integration of the processes from the small-scale wind damage risk model ForestGALES into ORCHIDEE-CAN. The integration of ForestGALES into ORCHIDEE-CAN required, however, developing numerically efficient solutions to deal with (1) landscape heterogeneity, i.e. account for newly established forest edges for the parameterization of gusts; (2) downscaling spatially and temporally aggregated wind fields to obtain more realistic wind speeds that would represents gusts; and (3) downscaling storm damage within the 2500 km2 pixels of ORCHIDEE-CAN. This new version of ORCHIDEE-CAN was parameterized over Sweden. Subsequently, the performance of the model was tested against data for historical storms in southern Sweden between 1951 and 2010 and south-western France in 2009. In years without big storms, here defined as a storm damaging less than 15 × 106 m3 of wood in Sweden, the model error is 1.62 × 106 m3, which is about 100 % of the observed damage. For years with big storms, such as Gudrun in 2005, the model error increased to 5.05 × 106 m3, which is between 10 and 50 % of the observed damage. When the same model parameters were used over France, the model reproduced a decrease in leaf area index and an increase in albedo, in accordance with SPOT-VGT and MODIS records following the passing of Cyclone Klaus in 2009. The current version of ORCHIDEE-CAN (revision 4262) is therefore expected to have the capability to capture the dynamics of forest structure due to storm disturbance on both regional and global scales, although the empirical parameters calculating gustiness from the gridded wind fields and storm damage from critical wind speeds may benefit from regional fitting.

A New Global fAPAR and LAI Dataset Derived from Optimal Albedo Estimates: Comparison with MODIS Products

We present the first comparison between new fAPAR and LAI products derived from the GlobAlbedo dataset and the widely-used MODIS fAPAR and LAI products. The GlobAlbedo-derived products are produced using a 1D two-stream radiative transfer (RT) scheme designed explicitly for global parameter retrieval from albedo, with consistency between RT model assumptions and observations, as well as with typical large-scale land surface model RT schemes. The approach does not require biome-specific structural assumptions (e.g., cover, clumping, understory), unlike more detailed 3D RT model approaches. GlobAlbedo-derived values of fAPAR and LAI are compared with MODIS values over 2002–2011 at multiple flux tower sites within selected biomes, over 1200 × 1200 km regions and globally. GlobAlbedo-derived fAPAR and LAI values are temporally more stable than the MODIS values due to the smoothness of the underlying albedo, derived via optimal estimation (assimilation) using an a priori estimate of albedo derived from an albedo “climatology” (composited multi-year albedo observations). Parameters agree closely in timing but with GlobAlbedo values consistently lower than MODIS, particularly for LAI. Larger differences occur in winter (when values are lower) and in the Southern hemisphere. Globally, we find that: GlobAlbedo-derived fAPAR is ~0.9–1.01 × MODIS fAPAR with an intercept of ~0.03; GlobAlbedo-derived LAI is ~0.6 × MODIS LAI with an intercept of ~0.2. Differences arise due to the RT model assumptions underlying the products, meaning care is required in interpreting either set of values, particularly when comparing to fine-scale ground-based estimates. We present global transformations between GlobAlbedo-derived and MODIS products.

Combining the [ABA] and net photosynthesis-based model equations of stomatal conductance

Stomatal conductance gs is variously depicted as being dependent on environmental conditions (Jarvis, 976), transpiration (Monteith, 1995), net photosynthesis (Leuning, 1995) or chemical signalling arriving in the xylem (Tardieu and Davies, 1993). Accurate descriptions of gs are being increasingly demanded in the large-scale land surface model components of General Circulation Models (GCMs) to predict future land-atmospheric fluxes of water vapour, heat and carbon dioxide. The JULES model, for instance, uses the net photosynthesis description combined with a relatively simple semi-linear dependence on soil moisture content that modulates the photosynthesis dependence (Cox et al., 1998).Dewar (2002) combines the Leuning (1995) and Tardieu and Davies (1993) models. We revisit that combination, and discuss whether the Vapour Pressure Deficit (VPD) implicit in both components is different or in common. Further, we show a potential re-arrangement of the combined equations reveals that this model for gs can be considered as being dependent on only four variables: evaporative flux Jw, net photosynthesis an, soil moisture content θ and ambient CO2 concentration ca. Expressed this way, gs is influenced by two relatively slowly varying stores of the hydrological and carbon cycles (soil water content and atmospheric CO2) and two more rapidly fluctuating fluxes from both cycles (evaporation and net photosynthesis).We consider how the modelling structure and its response to both canopy-level and soil environmental controls may make it suitable for inclusion in GCMs, and what this entails in terms of parameterisation.

Hydrologic Drought Atlas for Texas

AbstractThis paper presents a hydrologic drought atlas for the state of Texas. The atlas depicts the spatial variation of drought severity for durations of 3, 6, 12, and 24 months corresponding to the return periods of 10, 25, 50, and 100 years. Using the variable infiltration capacity model which is a large-scale land surface model, drought characteristics were derived from monthly stream flow simulated at 1/8° resolution over the state of Texas. Appropriate marginal distributions of drought severity and duration were selected from amongst exponential, gamma, log-normal, and Weibull distributions, whereas the joint distribution of severity and duration was selected from the Archimedean, extreme value, Plackett, and elliptical families. Then, severity-duration-frequency (SDF) curves were obtained which were used for preparing the drought atlas that depicts the drought severity for any drought duration and return period at any location in Texas. Results indicate a decreasing pattern of severities from west...

Effect of sub-layer corrections on the roughness parameterization of a Douglas fir forest

This study investigated two aerodynamic methods to estimate the momentum roughness length and displacement height (z0 and d0 respectively) of a 32m-tall Douglas fir forest using simultaneous micrometeorological and flux measurements. When the flux–gradient method was used to objectively determine z0 and d0, corrections for roughness sub-layer effects proved to be important. A new iterative method is employed to solve the set of equations when the corrections were made. The estimated average values of z0 and d0 with corrections vary from 1.2m to 1.4m and from 26.2m to 28.4m respectively depending on the correction method applied. In the absence of experimentally determined roughness sub-layer height, the corrections of Harman and Finnigan (2007) yielded the best overall estimates of aerodynamic parameters. Comparison with results of over 25 other studies has shown that the results obtained here fit the general trend rather well. The relationship between displacement height d0 and mean tree height hc is well described by the quadratic equation: d0=0.0087hc2+0.566hc. Similarly the quadratic equation: ha=0.006hc2+0.865hc appears to be the best to illustrate the relation between aerodynamic canopy height ha and hc. These simple relationships can be easily incorporated to large scale land surface models, provided that spatially distributed canopy height information is available. Furthermore the alternative flux–variance technique cannot be objectively used to estimate z0 and d0 as no explicit method exists to select the exact value for coefficient C1, which mainly causes the variance method to compute systematically higher or lower sensible heat flux.

Adaptive Soil Moisture Profile Filtering for Horizontal Information Propagation in the Independent Column-Based CLM2.0

Abstract Data assimilation aims to provide an optimal estimate of the overall system state, not only for an observed state variable or location. However, large-scale land surface models are typically column-based and purely random ensemble perturbation of states will lead to block-diagonal a priori (or background) error covariance. This facilitates the filtering calculations but compromises the potential of data assimilation to influence (unobserved) vertical and horizontal neighboring state variables. Here, a combination of an ensemble Kalman filter and an adaptive covariance correction method is explored to optimize the variances and retrieve the off-block-diagonal correlations in the a priori error covariance matrix. In a first time period, all available soil moisture profile observations in a small agricultural field are assimilated into the Community Land Model, version 2.0 (CLM2.0) to find the adaptive second-order a priori error information. After that period, only observations from single individual soil profiles are assimilated with inclusion of this adaptive information. It is shown that assimilation of a single profile can partially rectify the incorrectly simulated soil moisture spatial mean and variability. The largest reduction in the root-mean-square error in the soil moisture field varies between 7% and 22%, depending on the soil depth, when assimilating a single complete profile every two days during three months with a single time-invariant covariance correction.

Impact of Watershed Geomorphic Characteristics on the Energy and Water Budgets

Abstract The GEOtop model makes it possible to analyze the short- and long-term effects of geomorphic variation on the partitioning of the lateral surface and subsurface water and surface energy fluxes. The topography of the Little Washita basin (Oklahoma) and of the Serraia basin (Trentino, Italy) have been used as base topographies from which virtual topographies with altered slopes and elevations have been created with corresponding modifications of the soil thickness and the extension of the channel network, according to applicable geomorphological theories, in order to quantify the contribution of these topographic features to the spatial and temporal variability of energy and water fluxes. Simulation results show that both a more extended channel network and more accentuated slopes cause an increase in the discharge balanced by a diminution of the evapotranspiration. The diminution of the latent heat flux is balanced by the increase in the sensible heat flux. Net radiation shows a minor sensitivity to topography. Evaporative fraction, on the contrary, is shown to be strongly dependent on geomorphic characteristics. The results confirm the importance of including an adequate treatment of topography in large-scale land surface models.