ISSN 1000-0526
CN 11-2282/P
CN 11-2282/P
Archive > Volume 42 Issue 12 > 2016,42(12):1483-1497. DOI:10.7519/j.issn.1000-0526.2016.12.006 Prev Next
Analysis of Forecast Modifying Trend of the Numerical Model
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
Modifying trend (MD) analysis for different forecasts at same valid time is a method to estimate the forecast error and calibrate the forecast. To verify the validation of this method, intensity and position index of the westerly wind trough (WWT) and area index, intensity index, west ridge point and north edge index of the subtropical high (sTH) of northwest Pacific Ocean are defined. Using the 500 hPa geopotential height forecast dataset from April 2011 to September 2015, the correlation and consistency of intensity and position MDs of sTH and WWT at different positions around China produced by ECMWF （EC） and T639 models are analyzed and compared. The results show that there are systematic biases for EC and T639 sTH forecast and the WWTs are nearly bias free. The systematic biases of T639 are larger than that of EC. As the lead times are prolonged, the sTH eastward systematic biases of west ridge points and decreasing intensity and area systematic biases are enlarged. The west ridge point, intensity and area of sTH show no significant correlations among the MDs of different lead times and the MDs are not consistent for EC and T639. The forecast error and MD characteristics of sTH are different as the west ridge points are located at different longitudes. As the west ridge point moves from the east Indian Ocean to the northwest Pacific Ocean east of Philippines Islands, the sTH forecast errores and systematic biases decrease. In different longitude ranges there is no significant correlation among MDs of different lead times and their covariances are negative, which means the MDs are inconsistent. For the WWT, EC and T639 models both show that the errors of WWT in high latitudes are larger than that in low latitudes and the errors of WWT intensity are small. The errors of T639 are significantly larger than that of EC. The covariances among MDs of different lead times are negative too, which means that the WWTs position MDs are not consistent for EC and T639 as well. The correlatons among MDs of different periods are nearly 0, which means that the upgrades of EC and T639 models do not have any effects on the inconsistency of MDs.
Keywords:
Project Supported:
Clc Number:
Mobile website
