资源环境科技发展态势分析平台

Spatial variation of near surface temperature lapse rate is usually analysed by subjectively dividing the study area in to different sub-regions based on political or physiographic boundaries. This can often lead to unreliable results as near surface temperature lapse rate largely depends on regional conditions. In this study, homogeneous regions of near surface air temperature lapse rate across India have been identified for monthly mean minimum, mean, and mean maximum temperature using a probabilistic Gaussian mixture model clustering approach. For the delineated regions, lapse rates are estimated using weighted linear regression model. The weights corresponding to each station were obtained from the clustering approach. Results indicate that regions with uniform lapse rate vary across different months and a minimum of three regions is obtained for maximum temperature in April month. Further, temperature lapse rate estimates exhibit spatial and temporal variability, and are less steep than the Environmental lapse rate (-6.5 degrees C/km). The maximum spatial variability in lapse rate is observed in May for both maximum and minimum temperature. The delineation of entire India into homogeneous regions improves temperature interpolation when compared to results obtained by considering entire India as one region. A maximum improvement of 34% in root mean square error is observed in prediction of monthly mean temperature in March. The identified regions and the associated lapse rates are expected to improve the prediction of near surface temperature at ungauged locations across India.