Font Size: a A A

Research On Prediction Methods For Persistent Extreme Precipitation Incentered Eastern China Based On TIGGE Data

Posted on:2016-03-23Degree:MasterType:Thesis
Country:ChinaCandidate:B Q ZhouFull Text:PDF
GTID:2180330461952991Subject:Science of meteorology
Abstract/Summary:PDF Full Text Request
TIGGE(THORPEX Interactive Grand Global Ensemble) centers perform poorly in the prediction of persistent extreme precipitation(PEP), carrying out downscaling of large-scale circulation based on TIGGE data is an effective method to improve the forecast performance of PEP. This paper focuses on establishing a downscaling model for the forecast of PEP based on spatial-temporal structure of the predictors, which are the atmospheric variables in the key area of the key influential systems in TIGGE data.A necessary condition for the establishment of downscaling prediction model is that the predictors should already show good performance in the state-of-the-art numerical weather predictions. Therefore, we conduct verification of wind at high and low levels which have been seldom studied among the predictors. The specific objects of wind at high and low levels are East Asian Subtropical Westerly Jet(EASWJ) and low-level winds of East and South Asian summer monsoon(EASM and SASM), the conclusions are as follows:(1) The verification of the predictability of the position, spatial coverage and intensity of EASWJ in the summers of 2010-2012 show that each ensenmble prediction system(EPS) predicts all EASWJ properties well, while the levels of skill of all EPSs decline as lead time extends. Overall, there exist improvements from the control to ensemble mean forecasts for predicting EASWJ. For deterministic forecasts of all EPSs, the prediction of average axis is better than the prediction of spatial coverage and intensity of EASWJ. The European Center for Medium-Range Weather Forecasts(ECMWF) performs best with a lead of about 0.5-1 day in predictability over the second-best EPSs for all EASWJ properties throughout the forecast range. ECMWF leads the Japan Meteorological Agency(JMA) by about 0.5-1 day for predicting EASWJ axis and about 1-2 days for predicting the spatial coverage and intensity.The largest lead of ECMWF over the relatively worse EPSs such as National Centers for Environmental Prediction(NCEP) and China Meteorological Administration(CMA) is about 3-4 days for all EASWJ properties. To sum up, ECMWF shows the highest level of skill for predicting EASWJ, followed by JMA.(2) The forecast performances of EASM and SASM by six TIGGE centers in the summers of 2008–2013 were evaluated. The results show that the EASM is overestimated by all TIGGE centers except the Canadian Meteorological Center(CMC). The SASM is also overestimated by ECMWF, CMA, and CMC, but is under-predicted by JMA. The EASM surge is overestimated by the NCEP and CMA and mainly underestimated by the others. The bias predictabilities for the SASM surge are similar to those of the SASM. The peaks of the SASM and EASM, including their surges, are mainly underestimated, whereas the valleys are mostly overestimated. Overall, the ECMWF and United Kingdom Meteorological Office(UKMO) have the highest forecast skill in predicting the SASM and EASM and both have respective advantages. The TIGGE centers generally show higher skill in predicting the SASM than the EASM, and their skill in forecasting the SASM and EASM is superior to that for their respective surges.Based on the previous evaluation and analyses, we established an analog prediction model through cosine similarity and cuckoo search using the spatial-temporal strcture of predictors, which are the atmospheric variables in the key area of the key influential systems in TIGGE data. The model is named as Key Influential Systems based Analog Model, “KISAM” for short. Performance of KISAM is inferior to the direct model output(DMO) of ECMWF in prediction of the daily precipitation of the PEP event for the lead time of 1 day. When the lead time extends to 3 days and longer, performance of KISAM show advantage over DMO, and the longer the lead time, the more obious the advantage. Moreover, skill of KISAM is more stable than DMO for the distribution of precipitation amounts at the lead time of 1、3 and 6 days, and they have their own advantages for different PEP days. For the PEPs in 17-19 Jun, KISAM can forecast the three-day PEP event more than a week in advance of the event, which is much earlier than DMO. Besides, KISAM performs better in predicting the three-day event than DMO at the same lead time. KISAM has relatively strong capacity to discriminate whether PEP will occur, especially at the time the PEP happens and the time that is about 4 days before or after the PEP. However, KISAM shows poor performance at the time of 4-5 days before PEP and 2-3 days after PEP for KISAM often mistakes these days as PEP days.
Keywords/Search Tags:TIGGE data, statistical verification, persistent extreme precipitatin, key influential system, statistical downscaling
PDF Full Text Request
Related items