Incorporating Stochastic Weather Generators into Studies on Climate Impacts: Methods and Uncertainties

By adopting various stochastic weather generators, different research groups in their recent studies have realized the importance of the effects of climatic variability on crop growth and development. The conventional assessments derived climate change scenarios from General Circulation Models (GCMs) experiments, however, are incapable of helping to understand this importance. The particular interest here is to review the general methodological scheme to incorporate stochastic weather generator into climate impact studies and the specific approaches in our studies, and put forward uncertainties that still exist. A variety of approaches have been taken to develop the parameterization program and stochastic experiment, and adjust the parameters of atypical stochastic weather generator called WGEN. Usually, the changes in monthly means and variances of weather variables between controlled and changed climate are used to perturb the parameters to generate the intended daily climate scenarios. We establish a parameterization program and methods for stochastic experiment of WGEN in the light of outputs of short-term climate prediction models, and evaluate its simulations on both temporal and spatial scales. Also, we manipulated parameters in relation to the changes in precipitation to produce the intended types and qualitative magnitudes of climatic variability. These adjustments yield various changes in climatic variability for sensitivity analyses. The impacts of changes in climatic variability on maize growth, final yield, and agro-climatic resources in Northeast China are assessed and presented as the case studies through the above methods. However, this corporation is still equivocal due to deficiencies of the generator and unsophisticated manipulation of parameters. To detect and simulate the changes in climatic variability is one of the indispensable ways to reduce the uncertainties in this aspect.

Temporal Statistical Downscaling of Precipitation and Temperature Forecasts Using a Stochastic Weather Generator

Statistical downscaling is based on the fact that the large-scale climatic state and regional/local physiographic features control the regional climate. In the present paper, a stochastic weather generator is applied to seasonal precipitation and temperature forecasts produced by the International Research Institute for Climate and Society （IRI）. In conjunction with the GLM （generalized linear modeling） weather generator, a resampling scheme is used to translate the uncertainty in the seasonal forecasts （the IRI format only specifies probabilities for three categories： below normal, near normal, and above normal） into the corresponding uncertainty for the daily weather statistics. The method is able to generate potentially useful shifts in the probability distributions of seasonally aggregated precipitation and minimum and maximum temperature, as well as more meaningful daily weather statistics for crop yields, such as the number of dry days and the amount of precipitation on wet days. The approach is extended to the case of climate change scenarios, treating a hypothetical return to a previously observed drier regime in the Pampas.

Visual Programming of Stochastic Weather Generator and Future Applications on Agroecological Study

Based on former studies on weather simulator modules in IPMist laboratory, study on visual programming of stochastic weather generator(VS-WGEN)was continued. In this study, Markov Chain, Monte Carlo, Fourier Series, and weak stationary process were used to generate the daily weather data in software Matlab 6. 0, with the data input from 40 years' weather data recorded by Beijing Weather Station. The generated data includes daily maximum temperature, minimum temperature, precipitation and solar radiation. It has been verified that the weather data generated by the VS-WGEN were more accurate than that by the old WGEN, when twenty new model parameters were included. VS-WGEN has wide software applications, such as pest risk analysis, pest forecast and management. It can be implemented in hardware development as well, such as weather control in weather chamber and greenhouse for researches on ecological adaptation of crop varieties to a given location over time and space. Overall, VS-WGEN is a very useful tool for studies on theoretical and applied ecology.

Changes in climatic variability and maize yield inNortheast China

The method linking general circulation models' (GCMs') outputs with crop growthsimulation models' inputs has been the first choice in the studies of impacts of climate change.Changes in climatic variability, however were not considered in most studies due to limitedknowledge concerned Changes in climatic means derived from a general circulation model DKRZOPYC were input into a stochastic weather generator WGEN run for synthetic daily climate scenarios.Monte Carlo stochastic sampling method was adopted to generate climate change scenarios withvarious possible climatic veriabilities. A dynamic simulation model for maize growth anddevelopment of MZMOD was used to assess the potenhal implication of the changes in both climaticmeans and variability nd the boacts of crop management in changing climate on maize productionin Northeast China. The results indicated that maize yield would be reduced to various degrees inmost of the sensitivity experiments of climatic variability associating with the shortening of theduration of phenological phase of different sowing dates. The Anpacts of the diverse distributions ofclimatic factors detetmined by multiple changes in climatic variability on maire production and itsvariation, however, are not identical and have distinct regional disparities. Yield reduction caused bychanges in climatic means may be alleviated or aggravated by didributions of certain climaticvariables in line with the corresponding climatic variability according to the sensitivity analyses.Consequently, the hypothesis keeping climatic variability constant in the traditional research imposesrestriction on the overall inveshgation of the impacts of climate change on maize production.

A study on the possible impacts of climate warming on rice production in China

Based on climate change scenarios projected from GCMs (GFDL, UKMO and MPI), this study evaluates possible impacts of climate warming on rice production in China using numerical simulation experiments. A stochastic weather generator is used to make the projected climatic change scenarios suitable to the input of crop model, ORYZA1. The results show that the duration of rice growing season will be lengthened by 6-11 days and the accumulated temperature will increase by 200℃.d-330℃.d when CO2 concentration in the atmosphere doubles. The probability of cool injury in reproductive and grain filling period will decrease while that of heat stress will increase. Rice yield will decrease if cultivars and fanning practices are unchanged. If the dates of rice development stages can be maintained unchanged through cultivar adjustment although rice yield in most parts of the areas will decrease, the decrements will be much less than that when cultivars and farming practices are unchanged.

Towards global coverage of gridded parameterization for CLImate GENerator(CLIGEN)

Stochastic weather generators create time series that reproduce key weather dynamics present in long-term observations.The dataset detailed herein is a large-scale gridded parameterization for CLImate GENerator(CLIGEN)that fills spatial gaps in the coverage of existing regional CLIGEN parameterizations,thereby obtaining near-global availability of combined coverages.This dataset primarily covers countries north of 40°latitude with 0.25°spatial resolution.Various CLIGEN parameters were estimated based on 20-year records from four popular global climate products.Precipitation parameters were statistically downscaled to estimate point-scale values,while point-scale temperature and solar radiation parameters were approximated by direct calculation from high-resolution datasets.Surrogate parameter values were used in some cases,such as with wind parameters.Cross-validation was done to assess the downscaling approach for six precipitation parameters using known point-scale values from ground-based CLIGEN parameterizations.These parameter values were derived from daily accumulation records at 7,281 stations and high temporal resolution records at 609 stations.Two sensitive parameters,monthly average storm accumulation and maximum 30-minute intensity,were shown have RMSE values of 1.48 mm and 4.67 mm hr^(−1),respectively.Cumulative precipitation and the annual number of days with precipitation occurrence were both within 5%of ground-based parameterizations,effectively improving climate data availability.

Rainfall erosivity and erosivity density through rainfall synthetic series for Sao Paulo State,Brazil:Assessment,regionalization and modeling

Rainfall is the main cause of erosion of Brazilian soils,which makes assessing the rainfall erosivity factor(RE)and the erosivity density(ED)fundamental for soil and water conservation.Therefore,the objectives of this study were:i)to estimate the RE and ED for Sao Paulo State,Brazil,using synthetic series of pluviographic data;ii)to define homogeneous regions regarding rainfall erosivity;and iii)to generate regression models for rainfall erosivity estimates in each of the homogeneous regions.Synthetic series of pluviographic data were initially obtained on a sub-daily scale from the daily rainfall records of 696 rainfall gauges.The RE values were then estimated from the synthetic rainfall data,and ED was calculated from the relationship between erosivity and rainfall amounts.Monthly and annual maps for RE and ED were obtained.Hierarchical clustering analysis was used to define homogeneous regions in terms of rainfall erosivity,and regionalized regression models for estimating RE were generated.The results demonstrate high spatial variability of RE in Sao Paulo,where the highest annual values were observed in the coastal region.December to March concentrate approximately 60%of the intra-annual erosivity.The highest values of annual ED were observed in regions with intense agricultural activity.The definition of five homogeneous regions concerning the rainfall erosive potential evidenced distinct seasonal patterns of the spatial distribution of erosivity.Finally,the high predictive accuracy of the regionalized models obtained characterizes them as essential tools for reliable estimates of rainfall erosivity,and contribute to better soil conservation planning.