资源环境科技发展态势分析平台

Available climate data for south east Australia is reliant upon elevational lapse rates, which do not account for mesoscale processes that can affect temperatures, such as cold air drainage. Additional predictor variables are available for generating new climate datasets such as topographic indices and Moderate Resolution Imaging Spectroradiometer land surface temperature (MODIS LST); however, these have not been thoroughly tested to date. In this study, the relative benefits of including a localized topographic index and standardized MODIS LST values for temperature interpolation were assessed using partial bivariate splines, full and partial trivariate splines, and regression kriging. Trivariate splines provided the best interpolation performance in most cases; however, the partial bivariate spline with a fixed dependence upon elevation performed marginally better than the full trivariate spline for minimum temperature. The local topographic index improved the RMSE of minimum temperature climate normals by 17% in comparison to the best performing elevation only model. A further improvement for minimum temperature performance was achieved by including standardized night time MODIS LST values as covariates (34-39% reduction in RMSE). Standardized day time MODIS LST values improved maximum temperature interpolation performance; however, the improvement was onlymarginal in comparison to the full trivariate spline (6% reduction in RMSE). Cross validation of daily maximum and minimum temperature anomalies reflected performance trends shown in the climate normal analysis. Results suggest that the use of alternative approaches to interpolating temperature data may have significant implications for the calculation of bioclimatic variables and provide new opportunities to study extremes at high spatial and temporal resolutions using existing weather station networks. Furthermore, improving minimum temperature surfaces by accounting for temperature inversions driven by cold air drainage regimes may improve our ability to incorporate mesoscale temperature variability into a variety of applications, such as deriving temperature dependent climatic variables, species distribution modelling and assessments of fire risk.