Methodology
Ph.D. Thesis of radar nowcasting: from methodology to verification
In this paper, the authors detailed a way to blend radar nowcasting with NWP data via a space-reduced Kalman filter approach.
They transform the high-dimention space data to low space with the PCA, and reconstruct back after the correction stage in the Ensemble Kalman Filter.
The result extends the nowcasting range further than without blending.
STEPS
The description of STEPS system is as follows:
-
spatial scaling of precipitation fields (scale the precipitation fields based on FFT)
-
dynamic scaling of precipitation fields (scale the field dynamically by fitting a power law function of spatial scale)
-
spatial correlation of forecast errors (power law filter to white noise field in FFT field and inverse back)
-
temporal correlation of the forecast errors (Auto-regressive regression)
They verified the STEPS-BE algorithm with statistics within 2 hours leading time.
Warn-on-Forecast
In this paper, the authors introduced the methodologies for developing the WoF system for 0-6h weather nowcasting.