Abstract:Evapotranspiration (ET) is, after precipitation, the second largest flux at the land surface in the water cycle and occurs mainly during daytime. Less attention has been given to water fluxes from the land surface into the atmosphere during nighttime (i.e., between sunset and sunrise). The nighttime ET (ETN) may be estimated based on models that use meteorological data; however, due to missing experimental long‐term data, the verification of ETN estimates is limited. In this paper, the amount of ETN for two gr… Show more
“…Specifically, LE predictions by the hybrid model are most sensitive to h_canopy and SM. The strong dependence of ET and surface resistance on SM is well known (e.g., Akbar et al, ; Dirmeyer, ; Douville et al, ; Entekhabi et al, ; Gentine et al, ; Koster et al, ; Koster et al, ; Seneviratne et al, ; Seneviratne et al, ; Vogel et al, ), but the impact of vegetation height has been only recently recognized (Groh et al, ; Klein et al, ; Ringgaard et al, ; Xu et al, ), and not thoroughly quantified. Here we find that canopy height is even more important than SM.…”
Section: Resultsmentioning
confidence: 99%
“…We then test the sensitivity of the hybrid model to different environmental variables by adding a perturbation to each input variable (10-90% standard deviation increase, with 10% increment) to understand which variables are the most important regulators of r s and LE (Figures 4a and 4b), as well as to assess their nonlinearity. The most important variables (ranked by their mean R 2 with and without perturbation in Dirmeyer, 2011; Douville et al, 2016;Entekhabi et al, 1996;Gentine et al, 2012;Koster et al, 2004;Koster et al, 2006;Seneviratne et al, 2006;Seneviratne et al, 2010;Vogel et al, 2017), but the impact of vegetation height has been only recently recognized (Groh et al, 2019;Klein et al, 2015;Ringgaard et al, 2014;Xu et al, 2018), and not thoroughly quantified. Here we find that canopy height is even more important than SM.…”
Section: Interpretation Of Surface Resistance Predictions By the Hybrmentioning
Estimating ecosystem evapotranspiration (ET) is important to understanding the global water cycle and to study land-atmosphere interactions. We developed a physics constrained machine learning (ML) model (hybrid model) to estimate latent heat flux (LE), which conserves the surface energy budget. By comparing model predictions with observations at 82 eddy covariance tower sites, our hybrid model shows similar performance to the pure ML model in terms of mean metrics (e.g., mean absolute percent errors) but, importantly, the hybrid model conserves the surface energy balance, while the pure ML model does not. A second key result is that the hybrid model extrapolates much better than the pure ML model, emphasizing the benefits of combining physics with ML for increased generalizations. The hybrid model allows inferring the structural dependence of ET and surface resistance (r s ), and we find that vegetation height and soil moisture are the main regulators of ET and r s .
Plain Language SummaryA physics constrained machine learning model is developed using the FLUXNET2015 Tier 1 data set. This new approach is able to effectively retrieve latent heat flux while constraining energy conservation in the surface energy budget. This hybrid model has better performance in extrapolation than a pure machine learning model. Key Points: • A physics-constrained machine learning model of evapotranspiration (hybrid model) is developed and trained using the FLUXNET 2015 data set • The evapotranspiration retrieved by the hybrid model is as accurate as pure machine learning model and also conserves surface energy balance • The hybrid model better reproduces extremes and thus better extrapolates compared to the pure machine learning approach Supporting Information:• Supporting Information S1• Figure S1 • Table S1
“…Specifically, LE predictions by the hybrid model are most sensitive to h_canopy and SM. The strong dependence of ET and surface resistance on SM is well known (e.g., Akbar et al, ; Dirmeyer, ; Douville et al, ; Entekhabi et al, ; Gentine et al, ; Koster et al, ; Koster et al, ; Seneviratne et al, ; Seneviratne et al, ; Vogel et al, ), but the impact of vegetation height has been only recently recognized (Groh et al, ; Klein et al, ; Ringgaard et al, ; Xu et al, ), and not thoroughly quantified. Here we find that canopy height is even more important than SM.…”
Section: Resultsmentioning
confidence: 99%
“…We then test the sensitivity of the hybrid model to different environmental variables by adding a perturbation to each input variable (10-90% standard deviation increase, with 10% increment) to understand which variables are the most important regulators of r s and LE (Figures 4a and 4b), as well as to assess their nonlinearity. The most important variables (ranked by their mean R 2 with and without perturbation in Dirmeyer, 2011; Douville et al, 2016;Entekhabi et al, 1996;Gentine et al, 2012;Koster et al, 2004;Koster et al, 2006;Seneviratne et al, 2006;Seneviratne et al, 2010;Vogel et al, 2017), but the impact of vegetation height has been only recently recognized (Groh et al, 2019;Klein et al, 2015;Ringgaard et al, 2014;Xu et al, 2018), and not thoroughly quantified. Here we find that canopy height is even more important than SM.…”
Section: Interpretation Of Surface Resistance Predictions By the Hybrmentioning
Estimating ecosystem evapotranspiration (ET) is important to understanding the global water cycle and to study land-atmosphere interactions. We developed a physics constrained machine learning (ML) model (hybrid model) to estimate latent heat flux (LE), which conserves the surface energy budget. By comparing model predictions with observations at 82 eddy covariance tower sites, our hybrid model shows similar performance to the pure ML model in terms of mean metrics (e.g., mean absolute percent errors) but, importantly, the hybrid model conserves the surface energy balance, while the pure ML model does not. A second key result is that the hybrid model extrapolates much better than the pure ML model, emphasizing the benefits of combining physics with ML for increased generalizations. The hybrid model allows inferring the structural dependence of ET and surface resistance (r s ), and we find that vegetation height and soil moisture are the main regulators of ET and r s .
Plain Language SummaryA physics constrained machine learning model is developed using the FLUXNET2015 Tier 1 data set. This new approach is able to effectively retrieve latent heat flux while constraining energy conservation in the surface energy budget. This hybrid model has better performance in extrapolation than a pure machine learning model. Key Points: • A physics-constrained machine learning model of evapotranspiration (hybrid model) is developed and trained using the FLUXNET 2015 data set • The evapotranspiration retrieved by the hybrid model is as accurate as pure machine learning model and also conserves surface energy balance • The hybrid model better reproduces extremes and thus better extrapolates compared to the pure machine learning approach Supporting Information:• Supporting Information S1• Figure S1 • Table S1
“…For the other crops the spread of fungal pathogens under a more humid climate (Talley et al, 2002;Agam and Berliner, 2006) and frequent occurrence of dew formation (Xiao et al, 2009;Groh et al, 2018a;Brunke et al, 2019;Groh et al, 2019) could explain the generally lower yield of grain crops for soils under a wet climate in Selhausen. However, an appropriate crop management with one to three applications of fungicides during the growing season (see appendix impact on crop yield such that other reasons have to be considered.…”
Section: Crop Yield and Water Use Efficiencymentioning
Abstract. Future crop production will be affected by climatic changes. In several regions, the projected changes in total rainfall and seasonal rainfall patterns will lead to lower soil water storage (SWS) which in turn affects crop water uptake, crop yield, water use efficiency, grain quality and groundwater recharge. Effects of climate change on those variables depend on the soil properties and were often estimated based on model simulations. The objective of this study was to investigate the response of key variables in four different soils and for two different climates in Germany with different aridity index: 1.09 for the wetter (range: 0.82 to 1.29) and 1.57 for the drier climate (range: 1.19 to 1.77), by using high-precision weighable lysimeters. According to a “space-for-time” concept, intact soil monoliths that were moved to sites with contrasting climatic conditions have been monitored from April 2011 until December 2018. Evapotranspiration was lower for the same soil under the relatively drier climate whereas crop yield was significantly higher, without affecting grain quality. Especially non-productive water losses (evapotranspiration out of the main growing period) were lower which led to a more efficient crop water use in the drier climate. A characteristic decrease of the SWS for soils with a finer texture was observed after a longer drought period under a drier climate. The reduced SWS after the drought remained until the end of the observation period which demonstrates carry-over of drought from one growing season to another and the overall long term effects of single drought events. In the relatively drier climate, water flow at the soil profile bottom showed a small net upward flux over the entire monitoring period as compared to downward fluxes (ground water recharge) or drainage in the relatively wetter climate and larger recharge rates in the coarser- as compared to finer-textured soils. The large variability of recharge from year to year and the long lasting effects of drought periods on SWS imply that long term monitoring of soil water balance components is necessary to obtain representative estimates. Results confirmed a more efficient crop water use under less optimal soil moisture conditions. Long-term effects of changing climatic conditions on the SWS and ecosystem productivity should be considered when trying to develop adaptation strategies in the agricultural sector.
“…Seneviratne et al, 2012 for site details). The sensors are thoroughly described by Hirschi et al (2017). Given that the focus is on sensor comparison in this case, day and night are distinguished using a simple threshold of 10 W m −2 for measured incoming solar radiation; below this threshold, it is assumed that no photosynthesis occurs (Hirschi et al, 2017).…”
Section: Introductionmentioning
confidence: 99%
“…The EC data are processed using EddyPro (Fratini and Mauder, 2014;LI-COR, 2018) to obtain a latent heat flux time series with a temporal resolution of 30 min. Values are discarded for intervals when rain occurs, when the tower is in the upwind direction affecting the air flow (see Hirschi et al, 2017), and for cases with overly low turbulence (median threshold for friction velocity) based on Wutzler et al (2018). The resulting gaps are filled according to Reichstein et al (2005).…”
Abstract. Nocturnal water loss (NWL) from the surface into the atmosphere is often overlooked because of the absence of solar radiation to drive
evapotranspiration and the measuring difficulties involved. However, growing evidence suggests that NWL – and particularly nocturnal transpiration
– represents a considerable fraction of the daily values. Here we provide a global overview of the characteristics of NWL based on latent heat
flux estimates from the FLUXNET2015 dataset, as well as from simulations of global climate models. Eddy-covariance measurements at 99 sites indicate
that NWL represents 6.3 % of total evapotranspiration on average. There are six sites where NWL is higher than 15 %; these sites comprise mountain
forests with considerable NWL during winter that is related to snowy and windy conditions. Higher temperature, vapor pressure deficit, wind speed, soil
moisture, and downward longwave radiation are related to higher NWL, although this is not consistent across all of the sites. On the other hand, the global
multi-model mean of terrestrial NWL is 7.9 % of the total evapotranspiration. The spread of the model ensemble, however, is greater than 15.8 %
over half of the land grid cells. Finally, NWL is projected to increase everywhere with an average of 1.8 %, although with a substantial
inter-model spread. Changes in NWL contribute substantially to projected changes in total evapotranspiration. Overall, this study highlights the
relevance of water loss during the night and opens avenues to explore its influence on the water cycle and the climate system under present and
future conditions.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.