Wind Risk Model Research Articles

Evaluating wind damage vulnerability in the Alps: A new wind risk model parametrisation

The risk of wind damage to European forests is expected to increase due to the changed climate. Research efforts in forestry have been focussing on the development of analytical and modelling tools to aid with the evaluation and the prediction of forests' vulnerability to wind damage, and ultimately to inform forest management decisions aimed at promoting wind-resistance in forest stands. Recent catastrophic wind damage to European forests has shown that wind damage risk applies also to montane forests. These are of particular importance for the various ecosystem services they provide, including protection from gravitational hazards and defence against soil erosion. At present, the available forest wind risk models have been tested and used mainly on production or planted forests in different countries, but never in the complexity of mountainous terrains. The aim of this study is to introduce a methodology for the validation of a new parametrisation of the ForestGALES wind risk model, specifically designed for the alpine scenario. To do so, we used parameters from pulling tests in the field, and laboratory mechanical tests. We also developed a workflow that, starting from data obtained by LiDAR, aims at producing wind damage vulnerability maps. After validating this new parameterisation, we investigated the variations in wind vulnerability for a case study in the Italian Alps by introducing the Difference of Vulnerability (DoV) raster. Among our results, the DoV proves to be an innovative tool whose future use can increase understanding of the factors predisposing a forest stand to wind instability.

Testing an individual tree wind damage risk model in a naturally regenerated balsam fir stand: potential impact of thinning on the level of risk

Abstract Widely distributed in Quebec, balsam fir (Abies balsamea (L.) Mill.) is highly vulnerable to wind damage. Recently, there has been a trend in forest management to increase the use of partial cuttings in naturally regenerating stands, leaving the remnant trees at increased risk of wind damage. In order to limit wind damage after partial cuttings, it is therefore important to find silvicultural practices that minimize the risk of wind damage in these fir stands. Our main objective was to find balsam fir-specific values of parameters to integrate into the wind risk model ForestGALES, in order to simulate the impact of different types of commercial thinning on wind damage risk, and to determine which practice potentially minimizes the risk in a naturally regenerated stand. An anemometer placed at canopy height and strain gauges attached to the trunks of balsam firs allowed us to measure the wind-induced bending moments experienced by a sample of balsam fir trees. This enabled the calculation of the turning moment coefficients specific to each of the trees in order to compare them with the ForestGALES model predictions and to adapt the model for balsam fir stands. The model was tested first with only tree diameter and height as input variables to calculate the turning moment coefficient, then with the addition of a competition index, and finally with the addition of crown dimensions. Wind climate parameters for prediction of the probability of damage were calculated using the Wind Atlas Analysis and Application Program airflow model. The model with the highest accuracy was then used to simulate two types of thinning and determine the impact on wind damage risk for each tree in the stand. According to the model’s predictions, thinning from below has a reduced risk of wind damage compared with thinning from above.

A root-to-foliage tree dynamic model for gusty winds during windstorm conditions

Improving forest wind risk models relies on a better understanding of the tree response to damaging gusts. However, our empirical knowledge of the tree dynamic response to wind is limited by the feasibility of field experiments at recording the dynamics of both, the tree aerial structure and the tree root-system, under windstorm conditions. Here, we present the first complete root-to-foliage tree dynamic model simulating the tree aerial structure and root-soil system responses to turbulent winds, including the possibility of tree breakage and tree uprooting. This model allowed us to investigate the impact of both (1) various anchorage conditions, related to the soil humidity or root rigidity, and (2) the history of gusts experienced by the tree, on the dynamics and windthrow vulnerability of a Pinus pinaster tree. The model suggests a negligible sensitivity of the upper tree swaying dynamics to the anchorage condition as opposed to the stem base movement. The low-frequency motions and natural vibration modes of the tree aerial structure in response to the wind are felt by the roots, with a propagation intensity depending on the root-soil system properties. Only the extremities of the transverse roots appear weakly impacted by the tree aerial motion. A study of the tree uprooting to the wind load demonstrates the importance of the gustiness history experienced by the tree in weakening the tree anchorage, and thus in progressively enhancing the tree vulnerability to uprooting. This result suggests that the critical overturning bending moment predicted by static tree-pulling experiments may overestimate the ultimate critical bending moment in situations where trees are damaged progressively by a succession of wind gusts. This novel model represents a promising tool for better understanding the tree response to gusty winds for various plant anchorage conditions.

Critical wind speeds suggest wind could be an important disturbance agent in Amazonian forests

Abstract Recent research in the central Amazon suggests that wind is a major agent of disturbance, however, a mechanistic understanding of how wind may lead to tree mortality in Amazonian forests remains unclear. Here we estimated wind speeds necessary to topple central Amazon trees by linking both static and dynamic versions of two wind speed estimation methods (four methods total) to field data on tree failure derived from a static winching study. Static versions of these methods assumed invariant wind characteristics as more trees failed, while dynamic versions updated tree spacing, leaf area index and wind profiles progressively after each tree failure. First, we used a profile method which estimates wind force on individual trees by segments. We calculated drag on each segment and converted drag into basal turning moment, and compared the summed turning moments to the critical turning moment measured in the winching study. Estimated critical wind speeds from the static profile method varied greatly, from 10.75 m s−1 to >120.0 m s−1 with a mean of 45.70 m s−1. Critical wind speeds estimated with static approaches decreased with tree size but were not significantly different between two focal genera. Primary drivers of variation in critical wind speed were tree height and crown size. Second, we used the turning moment coefficient method of Hale, S.E., Gardiner, B., Peace, A., Nicoll, B., Taylor, P. and Pizzirani, S. 2015 Comparison and validation of three versions of a forest wind risk model. Environ. Model. Softw.68, 27–41. doi:10.1016/j.envsoft.2015.01.016.; the static version of this method yielded less-variable estimates, ranging from 18.98 to 52.01 m s−1, with a mean of 30.88 m s−1. Notably, the two static methods for estimating critical wind speeds differed in the trees they identified as having the highest and lowest critical wind speeds. Dynamic variants of the above two methods produced greatly reduced ranges in CWS estimates for our study trees, because after the early tree failures, remaining trees were subject to greater wind penetration into the stand and thus greater loading for a given above-canopy wind speed. CWS estimated with dynamic approaches differed significantly between the focal taxa. Nevertheless, both estimates suggest that wind speeds commonly observed during Amazon storms are sufficient to produce widespread tree damage and mortality.

Use of machine learning techniques to model wind damage to forests

This paper tested the ability of machine learning techniques, namely artificial neural networks and random forests, to predict the individual trees within a forest most at risk of damage in storms. Models based on these techniques were developed individually for both a small forest area containing a set of 29 permanent sample plots that were damaged in Storm Martin in December 1999, and from a much larger set of 235 forest inventory plots damaged in Storm Klaus in January 2009. Both data sets are within the Landes de Gascogne Forest in Nouvelle-Aquitaine, France. The models were tested both against the data from which they were developed, and against the data set from the other storm. For comparison with an earlier study using the same data, logistic regression models were also developed. In addition, the ability of machine learning techniques to substitute for a mechanistic wind damage risk model by training them with previous mechanistic model predictions was tested.All models were accurate at identifying whether trees would be damaged or not damaged but the random forests models were more accurate, had higher discriminatory power, and were almost totally unaffected by the removal of any individual input variable. However, if all information relating to a stand was removed the random forests model lost accuracy and discriminatory power. The other models were similarly affected by the removal of all site information but none of the models were affected by removal of all tree information, suggesting that damage in the Landes de Gascogne Forest occurs at stand scale and is not controlled by individual tree characteristics. The models developed with the large comprehensive database were also accurate in identifying damaged trees when applied to the small forest data damaged in the earlier storm. However, none of the models developed with the smaller forest data set could successfully discriminate between damaged and undamaged trees when applied across the whole landscape. All models were very successful in replicating the predictions of the mechanistic wind risk model and using them as a substitute for the mechanistic model predictions of critical wind speed did not affect the damage model results.Overall the results suggest that random forests provide a significant advantage over other statistical modelling techniques and the random forest models were found to be more robust in their predictions if all input variables were not available. In addition, the ability to replace the mechanistic wind damage model suggests that random forests could provide a powerful tool for damage risk assessment over large regions and provide rapid assessment of the impact of different management strategies or be used in the development of optimised forest management with multiple objectives and constraints including the risk of wind damage.

Risk model of tree stand damage by winds and its evaluation based on damage caused by cyclone “Xaver”

Aim of study: To develop and evaluate the forest’s wind-risk model, dedicated for stand damage level.Area of study: Model was tested in the northeastern Poland.Material and methods: A risk model referring to the damage of forest stands by wind specifies, for every stand in a selected forest district, the risk factor within the range of 0 to 3. The higher value of the factor, the greater risk of damage, if wind occurs. The model was based on 11 features: average size of a tree stand, mean diameter breast high in the specified features’ ratio, species composition, degree of stand density, age of stand, forest site type, amount of damage caused by wind in the last 10-year period, location of forest district in the region of Poland, and three features for mountains: stand’s altitude above sea level, direction, and slope. The model used information from the State Forests’ Information System database (SILP), and since it was developed in JAVA computer language, the processing of data for one forest district lasted less than one minute.Main results: The model can create a forest district digital map, in which stands characterized by specific risk values are presented with high prediction accuracy.Research highlights: The risk model of tree stand damage by winds uses data provided by the SILP and what was proven in below study, can be used as an effective tool in a forestry practice.

How stand tree motion impacts wind dynamics during windstorms

Understanding how wind and trees interact during wind storms is crucial for better predicting forest wind damage. The complexity of this interaction is enhanced by the fragmented environment of forests. Here, we present an unprecedented field experiment (TWIST) where both the wind dynamics and the tree motion in the edge region of a maritime pine forest have been recorded simultaneously during four non-destructive wind storms. For three of them, the instrumented trees were under stand flow while for one of them they were under an edge flow. Our measurements demonstrate that the well-known characteristics of stand-flow dynamics remain valid under high wind conditions. Only the sub-canopy flow appeared more intermittent as canopy-top turbulent structures penetrate easier within the canopy due to the tree foliage reconfiguration. Under similar storm intensity, the tree motions were lower under edge flow than under stand flow due to the lower turbulence of the former flow while the mean wind speed was higher. This result demonstrates the importance of considering both the turbulence and the mean wind speed in wind risk models. No impact of tree motion other than tree reconfiguration were observed on the stand flow dynamics. On the other hand, for the edge flow, our measurements reveal a peak in frequency on the wind velocity fluctuations related to the fundamental tree vibration mode. This peak was especially visible at canopy top and in the upper trunk space under high wind conditions. Compared to the stand flow, we suspect that the velocity fluctuations induced by the tree motion emerge in the edge flow due to the lower background turbulence. Our edge storm was nonetheless not strong enough for tree motion to enhance flow turbulence and for trees to enter into resonance. These findings may suggest a higher susceptibility of near-edge trees to reach resonance than stand trees due to the motion of upwind trees in a lower background turbulence.

Variance-based sensitivity analysis of a wind risk model - Model behaviour and lessons for forest modelling

We submitted the semi-empirical, process-based wind-risk model ForestGALES to a variance-based sensitivity analysis using the method of Soboĺ for correlated variables proposed by Kucherenko et al. (2012). Our results show that ForestGALES is able to simulate very effectively the dynamics of wind damage to forest stands, as the model architecture reflects the significant influence of tree height, stocking density, dbh, and size of an upwind gap, on the calculations of the critical wind speeds of damage. These results highlight the importance of accurate knowledge of the values of these variables when calculating the risk of wind damage with ForestGALES. Conversely, rooting depth and soil type, i.e. the model input variables on which the empirical component of ForestGALES that describes the resistance to overturning is based, contribute only marginally to the variation in the outputs. We show that these two variables can confidently be fixed at a nominal value without significantly affecting the model's predictions. The variance-based method used in this study is equally sensitive to the accurate description of the probability distribution functions of the scrutinised variables, as it is to their correlation structure.

A simple wind–tree interaction model predicting the probability of wind damage at stand level

Forest wind damage occurs under the passage of strong intermittent wind gusts at tree level. In order to predict the probability of tree damage of a forest, a simple statistical wind speed model was developed and coupled to a tree swaying model. The wind model is based on a stochastic approach and on some universal characters of normalized wind statistics at canopy top. This model aims at generating high frequency (10Hz) time series of the three wind velocity components at canopy top knowing only the wind intensity from a nearby meteorological station and the height and cumulative plant area index of the forest. Compared to field measurements and large-eddy simulations over different canopy structures and densities, the wind model was able to reproduce accurately the main features of canopy-top wind dynamics, in particular the signature of the mixing-layer type coherent eddy structures developing at canopy top. Coupled with a tree swaying model, the model has allowed to predict the probability of wind damage of forest following the windstorm intensity and duration, and following the main tree characteristics resulting from silvicultural scenarios. By responding to some weaknesses of existing mechanistic wind risk models, this simple wind–tree interaction model may represent the first step toward a new generation of mechanistic wind risk models based on a probabilistic approach.

Modelling wind risk to Eucalyptus globulus (Labill.) stands

Wind damage to commercial plantations and natural forests is a serious concern for forest owners and managers all over the world, with notable losses having been reported in the last few decades in Europe, the Americas, and Oceania. Wind-risk models such as ForestGALES allow for a good understanding of the dynamics involved in wind damage, and for calculations of risk to be made, therefore providing vital information on the best practices to minimise such risk. In this paper we parameterise ForestGALES for Eucalyptus globulus (Labill.), arguably one the most widespread and commercially important species for pulp and biomass production, with tree-pulling data obtained in Asturias, Spain. Despite the scarce data on tree and stand characteristics available for wind damaged stands of Eucalyptus spp., we provide an evaluation of our model’s performance under different stocking densities by comparing our simulations with real wind damage data acquired from the literature. We show that ForestGALES is able to accurately model the critical wind speeds responsible for Eucalypts stand damage, hence extending the model’s applicability to this important commercial genus. In line with good modelling practice, we present the results of a sensitivity analysis of the model, performed with a Global variance-based method. Our sensitivity analysis confirmed the main role of Dbh, stocking density, and tree height in driving the model outputs, and highlighted the importance of accurately knowing the size of any upwind gaps adjacent to a stand to reduce uncertainty in model predictions.

Wind damage propagation in forests

A key issue for predicting and mitigating the risk of wind damage in forests is to understand the mechanics of damage propagation during windstorms. Until now the acting processes have been poorly understood due to the difficulty of performing measurements in such extreme conditions. Here we use an innovative wind–tree interaction model, which allows for large deflection and tree breakage, to unravel for the first time the mechanisms of damage propagation at forest scale. We find that damage propagation involves two stages. Firstly, initial damage is caused by the impact of strong downward wind gusts. Trees break preferentially at the end of such critical passing sweeps, as the tree motion decelerates. The second stage starts when the damaged areas reach about 5 canopy heights in length and 1 canopy height in width: as the flow accelerates within the damaged areas the mean wind load becomes sufficient to break newly-created edge trees. From this bifurcation point tree damage increases drastically, irrespective of the tree motion state and the type of passing gusts. In addition to demonstrating the possibility of simulating wind damage propagation, these results have considerable potential for improving wind risk models.

Predicting tree damage in fragmented landscapes using a wind risk model coupled with an airflow model

Forest mechanistic wind risk models are widely applied on heterogeneous landscapes, whereas their wind load parameterizations are often derived either from homogeneous stand conditions or from simple forest edge conditions. To evaluate the impact of improving the wind flow representation of the mechanistic wind risk model HWIND on tree damage predictions when applied on heterogeneous environments, we coupled HWIND with the airflow model Aquilon. Aquilon provides to HWIND the velocity profiles and the gust factor (deduced from an approach based on the probability distribution of the wind velocity and on the turbulent kinetic energy). HWIND–Aquilon is compared with HWIND alone on different stand configurations of Scots pine (Pinus sylvestris L.) and Norway spruce (Picea abies (L.) Karst.) comprising newly clearcuts or shelter stands. Although both models showed the same pattern of differences in edge-tree critical wind speeds with differences in clear-cut length and shelter stand height, the model comparison reveals significant differences in the magnitude of critical wind speeds between them. This discrepancy is explained by the wind velocity and gust factor parameterizations used in HWIND alone, as in other wind risk models that exhibit weaknesses in heterogeneous configurations. This result confirms the need for improving the wind flow representation in mechanistic wind risk models when applied to heterogeneous landscapes.

Comparison and validation of three versions of a forest wind risk model

Predicting the probability of wind damage in both natural and managed forests is important for understanding forest ecosystem functioning, the environmental impact of storms and for forest risk management. We undertook a thorough validation of three versions of the hybrid-mechanistic wind risk model, ForestGALES, and a statistical logistic regression model, against observed damage in a Scottish upland conifer forest following a major storm. Statistical analysis demonstrated that increasing tree height and local wind speed during the storm were the main factors associated with increased damage levels. All models provided acceptable discrimination between damaged and undamaged forest stands but there were trade-offs between the accuracy of the mechanistic models and model bias. The two versions of the mechanistic model with the lowest bias gave very comparable overall results at the forest scale and could form part of a decision support system for managing forest wind damage risk.

Building and Validating Geographically Refined Hurricane Wind Risk Models for Residential Structures

AbstractAccurate estimation of risk to residential structures from hurricane winds is critical for emergency planning and post-event recovery. Models based on fragility curves are widely used for assessing wind damage risk at the county and census tract levels. Large-scale evaluation of the predictive accuracy of these models has been hampered by the lack of detailed damage data. This paper has three aims: (1) to evaluate the predictive accuracy of fragility-curve-based models at the census tract level using a comprehensive damage data set for Harris County residences collected after Hurricane Ike in 2008, (2) to demonstrate the need for geographically refined models of wind damage risk at spatial scales of 1-km2 blocks and to analyze the performance of fragility-curve models at that level, and (3) to explore the sources of errors made by fragility-curve-based models at the census tract and 1-km2-block level using a statistical machine learning model constructed from 21 potential explanatory variables. Th...

Extreme Weather Loading Risk Model of Overhead Transmission Line

Power grid suffers tremendous economic loss in extreme ice disaster weather, suggesting that it lacks immediate precautionary system. On the basis of the standards of Q/GDW179-2008 and IEC60826-2003, the curve of transmission line design loads is built up. In view of ransom character for load-strength of transmission line, according to load-strength interference theory, a short-term wind and ice loading risk model is established, which is a time-dependent wind and ice loading model, and can be calculated unreliability probability and fault rate, showing risks about cluster fault and common fault. Furthermore, wind and ice loads are divided into five states, which show risk margin of loads. It also can provide precautionary information for operator, and can present risk measurement on time scale.

Forestry Applications for Satellite Lidar Remote Sensing

This paper presents a method to estimate forest parameters and surface topography from NASA's Geosciences Laser Altimeter System (GLAS). Their potential use as observational inputs to models is demonstrated using a wind-risk model for the UK, ForestGALES. Relative heights above ground were used as biophysical parameter estimators. Top Height was estimated with R 2 = 0.73, RMSE = 4.5 m. Diameter at breast height estimates differed for conifer-dominated stands (R 2 = 0.72, RMSE = 0.07 m) and for stands containing mostly broadleaves (R 2 = 0.41, RMSE = 0.11 m). Ground elevation estimation produced R 2 = 0.997, RMSE = 2.2 m. These three parameters were applied to F orestGALES for stand-level assessment of wind-throw risk. Stability is sensitive to small differences in tree dimensions, and therefore vegetation parameters require greater accuracy than those currently retrievable from GLAS to more reliably determine risk of wind-throw. Future satellite lidar missions such as NASA's DESDynI sensor aim to produce improved vegetation parameter estimation plus greater spatial coverage which would offer more appropriate inputs for forestry models.

Improvements in anchorage provided by the acclimation of forest trees to wind stress

Summary Trees are known to modify their growth in windy climates, but these responses have not previously been linked to alterations in anchorage strength. The effects of stem mass, soil group, rooting depth and wind exposure (Detailed Aspect Method of Scoring (DAMS)) on anchorage were modelled using data from experiments where Sitka spruce ( Picea sitchensis (Bong.) Carr.) trees had been mechanically overturned. Regressions of critical turning moment against stem mass were performed on combinations of soil group (mineral, gleyed mineral, peaty mineral and deep peat) and rooting depth ( 80 cm), with DAMS as an additional factor. Correlations of DAMS with critical turning moment were positive for all combinations, and DAMS had a positive effect on the slope of all critical turning moment against stem mass regressions. When DAMS was added to the model, it increased the variance accounted for, from 66.5 to 71.2 per cent. When revised anchorage coeffi cients were applied to the ForestGALES wind risk model, critical wind speeds and associated return times for a 55-year-old stand were reduced on a relatively sheltered DAMS 14 site and increased on a wind-exposed DAMS 18 site. The implications of trees developing stronger anchorage in locations with greater wind exposure are discussed in relation to management of wind risk.

Estimating long-term critical wind speed for wind damage by using mechanistic wind risk model, GALES

Estimating long-term critical wind speed for wind damage by using mechanistic wind risk model, GALES

Anchorage of coniferous trees in relation to species, soil type, and rooting depth

A database was constructed of tree-anchorage measurements from almost 2000 trees from 12 conifer species that were mechanically overturned on 34 sites in the United Kingdom between 1960 and 2000. Anchorage was compared among species, soil groups (freely-draining mineral, gleyed mineral, peaty mineral, and deep peat) and root depth classes (shallow, <40 cm; medium, 40–80 cm; and deep, >80 cm) using regressions of critical turning moment against stem mass. Sitka spruce (Picea sitchensis (Bong.) Carr.) was used as a benchmark because it formed the largest part of the database and was the only species with all soil-group and depth-class combinations. Anchorage of Sitka spruce was strongest on peat and poorest on gleyed mineral soils. Deep rooting increased critical turning moments by 10%–15% compared with trees of equivalent mass with shallower roots. Significantly better anchorage than Sitka spruce was found for grand fir (Abies grandis (Dougl. ex D. Don) Lindl.), with various rooting depths on freely draining and gleyed mineral soils and for Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) on medium-depth mineral soil. Lodgepole pine (Pinus contorta Dougl. ex Loud.) had poorer anchorage than Sitka spruce over a range of soil groups and root depth classes. Norway spruce (Picea abies (L.) Karst.) on shallow gleyed mineral soil, and Corsican pine (Pinus nigra subsp. laricio (Poir.) Maire) on medium depth mineral soil, also had poorer anchorage. Other combinations had similar anchorage to the equivalent Sitka spruce. These results are discussed with respect to the development of forest wind-risk models.

Are irregular stands more windfirm?

Summary The proposed transformation of substantial areas of even-aged plantation forests in Britain to irregular structures (‘continuous cover forestry’) has raised concerns about the likely wind stability of irregular stands. A review of the literature suggests that the major difference between irregular and regular stands is the lower (i.e. more stable) height : diameter ratio associated with the dominant trees in the former. This appears to be a consequence of the greater wind loading that these dominant trees have to withstand. Wind tunnel studies show no difference in wind profile within or above the two types of canopy. There have been few comparative root investigations in the two types of stand and no differences in rooting depth have been reported, although changes in root architecture could be anticipated as a result of greater wind loading. The implications of these findings upon windthrow risk in regular and irregular Sitka spruce stands has been explored using the Forest GALES wind risk model on sites of different wind exposure. The results suggest no difference in wind risk on sheltered sites. On sites of moderate exposure, an irregular stand at close to ‘steady state’ conditions could be more wind stable than a conventionally thinned regular stand. However, this advantage disappears with increasing exposure. The conclusion is that the promotion of irregular stands may provide structures with more stable characteristics, but these cannot be considered in isolation from the prevailing wind climate and the local site type.