Work

Developing models for winter processes in grasslands – Bayesian calibration and sensitivity analysis

Year: 2010
Type: dissertation
Abstract: For grass-based agriculture at high latitudes, poor overwintering of perennial forage grasses often has economic consequences due to yield loss and re-establishment of grass fields. In order to assess the performance of grass cultivars currently used in Norway under a future changing climate, a wholeyear grassland model has been developed. The basis of this whole-year model was a grassland model developed for the growing season. In order to incorporate the winter season, this grassland model needed additional sub-models for simulating snow cover, soil frost, ice encasement and the development of frost tolerance in the plants. The main objective of this thesis has been to develop these additional sub-models, calibrate them using Bayesian methods and identifying key parameters using sensitivity analysis. The sub-models were also used to construct agroclimatic indices in order to assess the impact of climate change on the winter survival of two forage grasses.\nThere are several challenges emerging when applying Bayesian calibration to a dynamic model. The Bayesian approach regards parameters as random and allows integration of prior knowledge. Using the snow cover sub-model as case study, it is here demonstrated how prior information and new data affect the calibration process, parameters and model outputs, with focus on uncertainty. Point estimates and uncertainties are calculated and visualized for both parameters and model outputs. Generally, uncertainty decreased when new data were incorporated. Uniformly distributed priors gave the best fit for this model according to root mean square error, while the more informative beta distributed priors gave more physically meaningful parameter estimates. Markov chains of samples from the posterior distribution of the parameters were obtained by the random walk Metropolis-Hastings algorithm. Crucial points when using these methods are reaching and determining convergence of these chains. In order to reach convergence faster, informative priors, Sivia’s likelihood, reflection and updating the proposal distribution with parts of the data gave successful results. To determine convergence objectively and correctly, the use of multiple chains and the Gelman Rubin method was found useful. Several decisions must be made when implementing Bayesian calibration, and we highlight and visualize the choices that were found to be most effective.\nWe developed a simple model SnowFrostIce which simulates depth of snow cover, the lower frost boundary of the soil and the freezing of surface puddles. We parameterised the model by means of Bayesian calibration, and identified important model parameters using the sensitivity analysis method of Morris. Verification of the model suggests that the results are reasonable. Due to the simple model structure, some overestimation occurs in snow and frost depth. Both the calibration and the sensitivity analysis suggested that the snow cover module could be simplified with respect to snow melt and liquid water content. The soil frost module should be kept unchanged, while the surface ice module should be changed when more detailed topographical data become available, such as better estimates of the fraction of the land area where puddles may form.\nTimothy (Phleum pratense L.) is the most important forage grass in Scandinavia and it is therefore highly interesting to study how it will perform in a changing climate. In order to model winter survival, the dynamics of hardening and dehardening must be simulated with satisfactory precision. We investigated an early timothy frost tolerance model (LT50 model), and an LT50 model for winter wheat. Based on the assumption that timothy has no vernalization requirement, unlike winter wheat, but does have the ability to adapt to cold temperatures in a process linked to stage of development, two alternative versions of the winter wheat model were also constructed. These four models were calibrated by a Bayesian approach using observations on LT50 for the timothy cultivar Engmo. The models were validated using independent observations at different locations reflecting differences in climate. A sensitivity analysis using the Morris method to identify important model parameters suggested that there is a connection between frost tolerance and stage of plant development, even if there is no vernalization requirement. The simplified winter wheat model was selected as the best candidate based on model selection criteria and its ability to capture the hardening and dehardening processes. The results from the Bayesian calibration suggests that there are no major regional differences in Norway calling for regional calibration. However, cultivar-specific calibration is probably required, since there are hardy and less hardy cultivars within the same species. A functional LT50 model would allow risk assessments to be made of future winter survival using specifically tailored and downscaled climate scenarios.\nWe assessed the impact of climate change on the winter survival of timothy (Phleum pratense L.), and perennial ryegrass (Lolium perenne L.) under Norwegian conditions using agroclimatic indices and a simulation model of frost tolerance. Available to this study was locally adjusted climate scenarios (two for the period 2071-2100; one for the period 2020-2049) for six important agricultural regions, represented by one location each. We proposed and validated a rough way to estimate the daily minimum air temperatures from scenario data. Compared to the control period 1961-1990, the hardening period would be shortened by up to 21 days. As a consequence the modelled maximum frost tolerance is expected to be reduced by up to 3.9℃ and 1.9℃ for timothy and perennial ryegrass, respectively, under the warmest scenario. In spite of this reduction in hardiness, the plants are expected to be hardy enough to withstand the predicted autumn frosts, and also we expect a general reduction in risk of winter frost injuries. The plant data available to this study suggests that the agroclimatic indices developed for Canadian conditions can be useful for assessing the hardening status in timothy and perennial ryegrass. They are, however, less suitable for assessing the risk of plant injury related to frost and ice encasement in Norway since the dynamics of cold adaptation is not accounted for by these indices. Although less snow is expected, this is in most cases not accompanied by an increase in the risk of ice encasement injuries. There is little risk of winter injuries related to frost and ice encasement in the hardier grass species timothy. The better overwintering conditions in general indicate that it will be possible to grow perennial ryegrass in areas where it is not grown today, given that the risk of fungal diseases is not increased.
Language: English
Open Access status: green
OpenAlex ID: W2586392903
Cites: 81
Cited by:
Related to: 10