Improving Representation of Deforestation Effects on Evapotranspiration in the E3SM Land Model

Cai, Xitian; Riley, William J.; Zhu, Qing; Tang, Jinyun; Zeng, Zhenzhong; Bisht, Gautam; Randerson, James T.

doi:10.1029/2018ms001551

Cited by 31 publications

(26 citation statements)

References 60 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, it is likely that this variation is small relative to other types of PFT variation (e.g., phenology) which we are able to capture using this dataset. Using the equations of Campbell and Norman (1998), which relate leaf width to dleaf by a leaf shape-dependent factor, we produce PFTdependent uncertainty bounds for dleaf (Table S6). These minimum and maximum values are then applied consistently across all PFTs for the minimum and maximum dleaf perturbation simulations, respectively.…”

Section: Using Observations To Inform Pft-specific Parameter Rangesmentioning

confidence: 99%

“…It has also been shown previously that it is difficult to find global optimal parameter values which consistently improve skill in a climate model (Jackson et al, 2008;Williamson et al, 2015;Li et al, 2018), and we find a similar outcome when studying biophysical processes with a land model. Cai et al (2019) calibrated evapotranspiration (ET) across multiple paired FLUXNET sites in the Energy Exascale Earth System Model Land Model and found a reduction in the mean annual bias of ET when the global model was tested with the optimized parameters. This study demonstrates the potential for site-level calibration to improve a global model simulation.…”

Section: Global Meanmentioning

confidence: 99%

“…Hand tuning parameter values can be computationally inefficient, requiring many model simulations and large amounts of computer time, especially when the spatial domain is large or global. Single point simulations provide a lower computational cost and the ability to more easily run large ensembles; however, results from parameter estimation at a single site may not be transferable to other regions due to differences in climate, soil, and vegetation types (Post et al, 2017;Cai et al, 2019;Huang et al, 2016;Lu et al, 2018;Ray et al, 2015). Undirected model calibration can be limited in its ability to objectively assess whether the optimal parameter configuration has been found due to the ambiguous and subjective nature of the hand tuning process (Mendoza et al, 2015;Prihodko et al, 2008).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A machine learning approach to emulation and biophysical parameter estimation with the Community Land Model, version 5

Dagon

Sanderson

Fisher

et al. 2020

Adv. Stat. Clim. Meteorol. Oceanogr.

View full text Add to dashboard Cite

Abstract. Land models are essential tools for understanding and predicting terrestrial processes and climate–carbon feedbacks in the Earth system, but uncertainties in their future projections are poorly understood. Improvements in physical process realism and the representation of human influence arguably make models more comparable to reality but also increase the degrees of freedom in model configuration, leading to increased parametric uncertainty in projections. In this work we design and implement a machine learning approach to globally calibrate a subset of the parameters of the Community Land Model, version 5 (CLM5) to observations of carbon and water fluxes. We focus on parameters controlling biophysical features such as surface energy balance, hydrology, and carbon uptake. We first use parameter sensitivity simulations and a combination of objective metrics including ranked global mean sensitivity to multiple output variables and non-overlapping spatial pattern responses between parameters to narrow the parameter space and determine a subset of important CLM5 biophysical parameters for further analysis. Using a perturbed parameter ensemble, we then train a series of artificial feed-forward neural networks to emulate CLM5 output given parameter values as input. We use annual mean globally aggregated spatial variability in carbon and water fluxes as our emulation and calibration targets. Validation and out-of-sample tests are used to assess the predictive skill of the networks, and we utilize permutation feature importance and partial dependence methods to better interpret the results. The trained networks are then used to estimate global optimal parameter values with greater computational efficiency than achieved by hand tuning efforts and increased spatial scale relative to previous studies optimizing at a single site. By developing this methodology, our framework can help quantify the contribution of parameter uncertainty to overall uncertainty in land model projections.

show abstract

Section: Using Observations To Inform Pft-specific Parameter Rangesmentioning

confidence: 99%

Section: Global Meanmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A machine learning approach to emulation and biophysical parameter estimation with the Community Land Model, version 5

Dagon

Sanderson

Fisher

et al. 2020

Adv. Stat. Clim. Meteorol. Oceanogr.

View full text Add to dashboard Cite

show abstract

“…Fire has a clear impact on stand water cycle [120] and hydrology at watershed scale [121]. Forest wildfires, importantly, alter the hydrological processes, which may impact on the growth of the forest and climate at regional scale [122,123].…”

Section: Is It Possible To Accurately Estimate Burn Severity From Thementioning

confidence: 99%

Can Landsat-Derived Variables Related to Energy Balance Improve Understanding of Burn Severity From Current Operational Techniques?

2020

View full text Add to dashboard Cite

Forest managers rely on accurate burn severity estimates to evaluate post-fire damage and to establish revegetation policies. Burn severity estimates based on reflective data acquired from sensors onboard satellites are increasingly complementing field-based ones. However, fire not only induces changes in reflected and emitted radiation measured by the sensor, but also on energy balance. Evapotranspiration (ET), land surface temperature (LST) and land surface albedo (LSA) are greatly affected by wildfires. In this study, we examine the usefulness of these elements of energy balance as indicators of burn severity and compare the accuracy of burn severity estimates based on them to the accuracy of widely used approaches based on spectral indexes. We studied a mega-fire (more than 450 km2 burned) in Central Portugal, which occurred from 17 to 24 June 2017. The official burn severity map acted as a ground reference. Variations induced by fire during the first year following the fire event were evaluated through changes in ET, LST and LSA derived from Landsat data and related to burn severity. Fisher’s least significant difference test (ANOVA) revealed that ET and LST images could discriminate three burn severity levels with statistical significance (uni-temporal and multi-temporal approaches). Burn severity was estimated from ET, LST and LSA using thresholding. Accuracy of ET and LST based on burn severity estimates was adequate (κ = 0.63 and 0.57, respectively), similar to the accuracy of the estimate based on dNBR (κ = 0.66). We conclude that Landsat-derived surface energy balance variables, in particular ET and LST, in addition to acting as useful indicators of burn severity for mega-fires in Mediterranean ecosystems, may provide critical information about how energy balance changes due to fire.

show abstract

“…Specifically, a process‐based model of the terrestrial surface, which integrates biogeochemical and biogeophysical land surface processes with physical climate models, is critical to understanding the temporal and spatial patterns of ecosystem processes and their response to climate change (Atkin et al, 2015; Lin et al, 2015). Over the years, ecosystem models have grown in sophistication to include multilayer hydrology, multilayer canopies, and carbon and nitrogen cycling (Cai et al, 2019; Collalti et al, 2014; Pitman, 2003; Zaehle et al, 2014).…”

Section: Introductionmentioning

confidence: 99%

Incorporating Spatial Variations in Parameters for Improvements of an Evapotranspiration Model

Keenan

et al. 2020

JGR Biogeosciences

View full text Add to dashboard Cite

Ecosystem models are important tools for exploring the temporal and spatial patterns of ecosystem processes and their responses to climate change. However, the implications of uncertainty in model parameters are often overlooked, especially in regional ecosystem model simulations. Here, we use eddy-covariance observations to estimate parameters in an ecosystem model, which was developed from Shuttleworth-Wallace model, and examine the effect on estimates of evapotranspiration (ET). Using a simple ecosystem model as an example, we use Monte Carlo techniques to optimize key model parameters using eddy covariance (EC) data from 163 FLUXNET sites. The optimization process revealed a strong spatial correlation between key parameters and environmental variables, particularly leaf area index (LAI) and soil characteristics (e.g., clay fraction). The optimization of parameters related to canopy conductance and soil surface resistance greatly improved model performance, particularly when incorporating the identified spatial variation of parameters. The improved model agreed well with the measurements with an increase in the coefficient of determination (R 2) from 73% to 80% in the 8-day averaged ET estimation and a decrease in the root mean square error (RMSE) from 130.2 to 104.3 mm year −1 compared with the original model. The results suggest the potential of eddy-covariance flux observations to identify predictable spatial variations of key parameters, which can be used to better constrain ecosystem models. And in this case, a universal and efficient method for reducing the uncertainties in key parameters across different PFTs and ecosystem applications is suggested.

show abstract

Improving Representation of Deforestation Effects on Evapotranspiration in the E3SM Land Model

Cited by 31 publications

References 60 publications

A machine learning approach to emulation and biophysical parameter estimation with the Community Land Model, version 5

A machine learning approach to emulation and biophysical parameter estimation with the Community Land Model, version 5

Can Landsat-Derived Variables Related to Energy Balance Improve Understanding of Burn Severity From Current Operational Techniques?

Incorporating Spatial Variations in Parameters for Improvements of an Evapotranspiration Model

Contact Info

Product

Resources

About