Surface Energy Flux Estimation in Two Boreal Settings in Alaska Using a Thermal-Based Remote Sensing Model

Abstract

Recent Arctic warming has led to changes in the hydrological cycle. Circum-Arctic and circumboreal ecosystems are showing evidence of “greening” and “browning” due to temperature warming leading to shrub encroachment, tree mortality and deciduousness. Increases in latent heat flux from increased evapotranspiration rates associated with deciduous-dominated ecosystems may be significant, because deciduous vegetation has extremely high-water use and water storage capacity compared to coniferous and herbaceous plant species. Thus, the impact of vegetation change in boreal ecosystems on regional surface energy balance is a significant knowledge gap that must be addressed to better understand observed trends in water use/availability and tree mortality. To this end, output from a two-source energy balance model (TSEB) with modifications for high latitude boreal ecosystems was evaluated using flux tower measurements and Terra/Aqua MODIS remote sensing data collected over the two largest boreal forest types in Alaska (birch and black spruce). Data under clear and overcast days and from leaf-out to senescence from 2012 to 2016 were used for validation. Using flux tower observations and local model inputs, modifications to the model formulation for soil heat flux, net radiation partitioning, and canopy transpiration were required for the boreal forest. These improvements resulted in a mean absolute percent difference of around 23% for turbulent daytime fluxes when surface temperature from the flux towers was used, similar to errors reported in other studies conducted in warmer climates. Results when surface temperature from Terra/Aqua MODIS estimates were used as model input suggested that these model improvements are pertinent for regional scale applications. Vegetation indices and LAI time-series from the Terra/Aqua MODIS products were confirmed to be appropriate for energy flux estimation in the boreal forest to describe vegetation properties (LAI and green fraction) when field observations are not available. Model improvements for boreal settings identified in this study will be implemented operationally over North America to map surface energy fluxes at regional scales using long time series of remote sensing estimates as part of NOAA’s GOES Evapotranspiration and Drought Information System.

Evaluating European ECOSTRESS Hub Evapotranspiration Products Across a Range of Soil‐Atmospheric Aridity and Biomes Over Europe

“…Thus, a deterministic PT coefficient can lead to uncertainties at large scales. An adjustment of the initial PT coefficient based on land surface type may mitigate the uncertainty in ET estimates (Andreu et al., 2018; Cristóbal et al., 2017, 2020; Guzinski et al., 2013). Besides, the soil aerodynamic resistance parameterization and the associated empirical coefficients could be another source of uncertainty (Li et al., 2019).…”

A Global Implementation of Single‐ and Dual‐Source Surface Energy Balance Models for Estimating Actual Evapotranspiration at 30‐m Resolution Using Google Earth Engine

“…It should be noted that TSEB undergoes continuous refinements to its aerodynamic resistance, transpiration, and wind and radiation divergence formulations applied to different landscapes containing unique canopy distributions and structure and surface conditions (Colaizzi et al., 2014; Cristóbal et al., 2020; M. Anderson et al., 2008; Nieto, Kustas, Alfieri, et al., 2019; Nieto, Kustas, Torres‐Rúa, et al., 2019; W. P. Kustas et al., 2022; Y. Li et al., 2019).…”

“…The uncertainty of TSEB comes from the following aspects: (a) the assumed initial value of the P-T coefficient for all landscapes and climates, which is likely affected by vapor pressure, wind speed, LAI, and vegetation type and height, representing the combined effect of aerodynamic and vegetation attributes on vegetation transpiration (Yang et al, 2015); (b) the tendency for the P-T coefficient to be lower in arid environments with natural vegetation and higher in areas with advection (Agam et al, 2010;Komatsu, 2003), thus a universal value for the P-T coefficient is unlikely; and (c) TSEB is sensitive to resistances within and below canopy (R x , R A , and R S ) (Gan & Gao, 2015;Yao et al, 2017) that are controlled by wind speed, vapor pressure deficit, and canopy height. However, nominal adjustment of the P-T coefficient and estimating canopy green fraction from remote sensing for different biomes has been fairly successful in oak savannas (Andreu et al, 2018), the arctic tundra, and boreal forest (Cristóbal et al, 2017(Cristóbal et al, , 2020, and in coniferous and deciduous forests and semiarid grasslands (Guzinski et al, 2013).…”

Evaluating European ECOSTRESS Hub Evapotranspiration Products Retrieved from Three Structurally Contrasting SEB Models over Europe

“…Thus, a deterministic P-T coefficient can lead to uncertainties at large scales. An adjustment of the initial P-T coefficient based on land surface type may mitigate the uncertainty in ET estimates (Andreu et al 2018;Cristóbal et al 2020;Cristóbal et al 2017;Guzinski et al 2013). Also, the soil aerodynamic resistance parameterization and the associated empirical coefficients could be another reason (Li et al 2019).…”