Analyses On The Spatial Evaluation Of Disastrous Weather Events Using MODE

Numerical forecast products are usually evaluated against the site observations by classical statistics,and the object-based spatial evaluation techniques are recently developed and comes into play.However,performance of the disastrous high winds and heavy rainstorms' forecast,which are the most concerned in industrial meteorological service,have seldom been evaluated.Therefore,based on the high winds and heavy rainstorms of 4 different synoptic conditions,evaluations of numerical forecast and alert system for power grid are chosen as an example to discuss the applicability of an object-based spatial evaluation technique,MODE(Method for Object-Based Diagnostic Evaluation),in high winds and heavy rainstorms.Through the selection,determination of MODE pertinent parameters and objects matching and analysis in high winds and heavy rainstorms,it is found that(1)parameters,such as the convolution radius,pairwise attribute's weight and matching critical values,in MODE are crucial to the objectivity of detection of objects and expression of errors.In the process of determination of aforementioned parameters,comprehensive considerations should be paid to the spatial resolutions of forecast/observation grid data,user-specified temporal and spatial error tolerances in forecast,the resilience of users to the weather disaster and the characteristics of a certain synoptic system.The weights of the pertinent parameters should come out by targeted determinations based on the users' sensitivity to the forecast of severe weather events.(2)MODE evaluations could so quantitatively depicts the general performance of model forecast,such as spatial deviation of range and location of high wind/heavy rainstorm area and temporal difference of high wind/heavy rainstorm,as to be able to give out the false alarm and missing report area for each valid hour and the model forecast performance of the movement and life-cycle of a severe synoptic system.(3)the evaluation of a high wind event(critical value is 10 m/s),which could cause foreign body intrusion to the power grid,reveals that the short-term forecast is able to well predict the onset of cold air gales in Henan,but with a short life-cycle and low wind speeds.A significant missing report is spotted on the plain east to the mountainous area.The short-term forecast also performs well on an summer squall line in Beijing,with 2 hours in advance.Many false alarms are found in plain areas,meanwhile,objects in mountainous areas are will predicted.(4)the evaluation of a heavy rainstorm,which could cause waterlogging in transformer substation,reveals that the short-term forecast could accurately predict the location of a cold vortex precipitation over Beijing,but with a lower coverage and missing reports.The model forecast underestimates the precipitation intensity when the rainstorm system approaches to Beijing and overestimates when the system moves out of Beijing.The movement and location of the heavy rainstorm of summer squall line in Beijing could be well predicted by short-term forecast,but the fact remains that missing report areas exist.(5)during the severest period of an disastrous weather event,poor forecast performance can be derived in terms of classical statistics.However,MODE evaluations demonstrate that numerical forecast could be advantageously feasible to the users under an acceptable relaxation of temporal and spatial error tolerances.