Angela J. Rigden

Changes in autumn senescence in northern hemisphere deciduous trees: a meta-analysis of autumn phenology studies.

BACKGROUND AND AIMS Many individual studies have shown that the timing of leaf senescence in boreal and temperate deciduous forests in the northern hemisphere is influenced by rising temperatures, but there is limited consensus on the magnitude, direction and spatial extent of this relationship. METHODS A meta-analysis was conducted of published studies from the peer-reviewed literature that reported autumn senescence dates for deciduous trees in the northern hemisphere, encompassing 64 publications with observations ranging from 1931 to 2010. KEY RESULTS Among the meteorological measurements examined, October temperatures were the strongest predictors of date of senescence, followed by cooling degree-days, latitude, photoperiod and, lastly, total monthly precipitation, although the strength of the relationships differed between high- and low-latitude sites. Autumn leaf senescence has been significantly more delayed at low (25° to 49°N) than high (50° to 70°N) latitudes across the northern hemisphere, with senescence across high-latitude sites more sensitive to the effects of photoperiod and low-latitude sites more sensitive to the effects of temperature. Delays in leaf senescence over time were stronger in North America compared with Europe and Asia. CONCLUSIONS The results indicate that leaf senescence has been delayed over time and in response to temperature, although low-latitude sites show significantly stronger delays in senescence over time than high-latitude sites. While temperature alone may be a reasonable predictor of the date of leaf senescence when examining a broad suite of sites, it is important to consider that temperature-induced changes in senescence at high-latitude sites are likely to be constrained by the influence of photoperiod. Ecosystem-level differences in the mechanisms that control the timing of leaf senescence may affect both plant community interactions and ecosystem carbon storage as global temperatures increase over the next century.

Evapotranspiration based on equilibrated relative humidity (ETRHEQ): Evaluation over the continental U.S.

A novel method of estimating evapotranspiration (ET), referred to as the ETRHEQ method, is further developed, validated, and applied across the U.S. from 1961 to 2010. The ETRHEQ method estimates the surface conductance to water vapor transport, which is the key rate-limiting parameter of typical ET models, by choosing the surface conductance that minimizes the vertical variance of the calculated relative humidity profile averaged over the day. The ETRHEQ method, which was previously tested at five AmeriFlux sites, is modified for use at common weather stations and further validated at 20 AmeriFlux sites that span a wide range of climates and limiting factors. Averaged across all sites, the daily latent heat flux RMSE is ∼26 W·m−2 (or 15%). The method is applied across the U.S. at 305 weather stations and spatially interpolated using ANUSPLIN software. Gridded annual mean ETRHEQ ET estimates are compared with four data sets, including water balance-derived ET, machine-learning ET estimates based on FLUXNET data, North American Land Data Assimilation System project phase 2 ET, and a benchmark product that integrates 14 global ET data sets, with RMSEs ranging from 8.7 to 12.5 cm·yr−1. The ETRHEQ method relies only on data measured at weather stations, an estimate of vegetation height derived from land cover maps, and an estimate of soil thermal inertia. These data requirements allow it to have greater spatial coverage than direct measurements, greater historical coverage than satellite methods, significantly less parameter specification than most land surface models, and no requirement for calibration.

Attribution of surface temperature anomalies induced by land use and land cover changes

Land use/land cover changes (LULCC) directly impact the surface temperature by modifying the radiative, physiological, and aerodynamic properties controlling the surface energy and water balances. In this study, we propose a new method to attribute changes in the surface temperature induced by LULCC to changes in radiative and turbulent heat fluxes, with the partition of turbulent fluxes controlled by aerodynamic and surface resistances. We demonstrate that previous attribution studies have overestimated the contribution of aerodynamic resistance by assuming independence between the aerodynamic resistance and the Bowen ratio. Our results further demonstrate that acceptable agreement between modeled and observed temperature anomalies does not guarantee correct attribution by the model. When performing an attribution analysis, the covariance among attributing variables needs to be taken into consideration in order to accurately interpret the results.

The Pattern Across the Continental United States of Evapotranspiration Variability Associated with Water Availability

The spatial pattern across the continental United States of the interannual variance of warm-season water-dependent evapotranspiration, a pattern of relevance to land-atmosphere feedback, cannot be measured directly. Alternative and indirect approaches to estimating the pattern, however, do exist, and given the uncertainty of each, we use several such approaches here. We first quantify the water-dependent evapotranspiration variance pattern inherent in two derived evapotranspiration datasets available from the literature. We then search for the pattern in proxy geophysical variables (air temperature, streamflow, and NDVI) known to have strong ties to evapotranspiration. The variances inherent in all of the different (and mostly independent) data sources show some differences but are generally strongly consistent – they all show a large variance signal down the center of the U.S., with lower variances toward the east and (for the most part) toward the west. The robustness of the pattern across the datasets suggests that it indeed represents the pattern operating in nature. Using Budyko’s hydroclimatic framework, we show that the pattern can largely be explained by the relative strength of water and energy controls on evapotranspiration across the continent.

Evaporation estimates using weather station data and boundary layer theory

Global estimates of evapotranspiration remain a challenge. In this study, we show that the daily course of air temperature and specific humidity available at routine weather stations can be used to estimate evapotranspiration and the evaporative fraction, the ratio of latent heat flux to available energy at the surface. Indeed, the diurnal increase in air temperature reflects the magnitude of the sensible heat flux and the increase of specific humidity after sunrise reflects the amplitude of evapotranspiration. The method is physically constrained and based on the budget of heat and moisture in the boundary layer. Unlike land surface-based estimates, the proposed boundary layer estimate does not rely on ad hoc surface resistance parameterizations (e.g., Penman-Monteith). The proposed methodology can be applied to data collected at weather stations to estimate evapotranspiration and evaporative fraction under cloudy conditions and in the pre–remote sensing era.

Attribution of Local Temperature Response to Deforestation

Land use and land cover change such as deforestation can directly induce changes in land surface temperature (LST). Using observational data from four paired eddy covariance sites, we attribute changes in LST induced by deforestation to changes in radiation, aerodynamic resistance, the Bowen ratio or surface resistance, and heat storage using two different methods: the intrinsic biophysical mechanism (IBM) method and the two-resistance mechanism method. The two models are first optimized to reduce the root-mean-square error of the simulated daily LST change by using daily-averaged inputs and a weighted average approach for computing the sensitivities. Both methods indicate that the daytime warming effect of deforestation is mostly induced by changes in aerodynamic resistance as the surface becomes smoother after deforestation, and the nighttime cooling effect of deforestation is controlled by changes in aerodynamic resistance, surface resistance, radiation, and heat storage. Both methods also indicate that changes in atmospheric temperature have a large impact on LST and need to be included in the LST attribution. However, there are significant differences between the two methods. The IBM method tends to overestimate the contribution of aerodynamic resistance due to the assumption that aerodynamic resistance and the Bowen ratio are independent. Additionally, the IBM method underestimates the contributions of radiation and heat storage during the daytime but overestimates them at night. By highlighting the similarity and dissimilarity between the two methods, this study suggests that acceptable agreement between observed and modeled LST change is the prerequisite for attribution but does not guarantee correct attribution.

Partitioning Evapotranspiration Over the Continental United States Using Weather Station Data

Accurately characterizing evapotranspiration is critical when predicting the response of the hydrologic cycle to climate change. Although Earth system models estimate similar magnitudes of global evapotranspiration, the magnitude of each contributing source varies considerably between models due to the lack of evapotranspiration partitioning data. Here we develop an observation-based method to partition evapotranspiration into soil evaporation and transpiration using meteorological data and satellite soil moisture retrievals. We apply the methodology at 1,614 weather stations across the continental United States during the summers of 2015 and 2016. We evaluate the method using vegetation indices inferred from satellites, finding strong spatial correlations between modeled transpiration and solar-induced fluorescence (r = 0.87), and modeled vegetation fraction and leaf area index (r = 0.70). Since the sensitivity of evapotranspiration to environmental factors depends on the contribution of each source component, understanding the partitioning of evapotranspiration is increasingly important with climate change. Plain Language Summary Water moves from the land surface to the overlying atmosphere by evaporation. The two main sources of evaporation include (1) evaporation from soils and (2) evaporation from pores on plants, called transpiration. Although methods exist to measure total evaporation over an ecosystem, it is challenging to observe soil evaporation and transpiration separately over an ecosystem. Consequently, the amount of estimated soil evaporation and transpiration varies considerably across models. In this study, we develop an observation-based method to estimate the fraction of water moved from the land to the atmosphere by plants, or the fraction of total evaporation that comes from transpiration. The method primarily relies on weather station data and soil moisture estimates from a recently launched satellite. We apply the method across the continental United States during the summers of 2015 and 2016 and evaluate it using observations of plants inferred from other satellites. Looking toward the future, it is important to estimate transpiration and soil evaporation correctly because they respond differently to changes in climate.

Reconciling the Reynolds number dependence of scalar roughness length and laminar resistance

The scalar roughness length and laminar resistance are necessary for computing scalar fluxes in numerical simulations and experimental studies. Their dependence on flow properties such as the Reynolds number remains controversial. In particular, two important power laws (“1/4” and “1/2”), both having strong theoretical foundations, have been widely used in various parameterizations and models. Building on a previously proposed phenomenological model for interactions between the viscous sublayer and the turbulent flow, it is shown here that the two scaling laws can be reconciled. The 1/4 power law corresponds to the situation where the vertical diffusion is balanced by the temporal change or advection due to a constant velocity in the viscous sublayer, while the 1/2 scaling corresponds to the situation where the vertical diffusion is balanced by the advection due to a linear velocity profile in the viscous sublayer. In addition, the recently proposed “1” power law scaling is also recovered, which corresponds to the situation where molecular diffusion dominates the scalar budget in the viscous sublayer. The formulation proposed here provides a unified framework for understanding the onset of these different scaling laws and offers a new perspective on how to evaluate them experimentally.