Precipitation, Radiance Products

Michael Uddstrom Group: NIWA, New Zealand Instrument: several Measurement: Assimilation of Satellite Data for NWP

Level 1 Level 2 Operational Research Global HRPT Weather Climate

Satellite Research at NIWA (New Zealand)

Mesoscale Data Assimilation and Weather Prediction (MDAWP)

At NIWA (the National Institute of Water and Atmospheric Research (http://www.niwa.co.nz), New Zealand) our remote sensing research is centred around the use of satellite data to improve the accuracy of a very-large, high resolution, cycling mesoscale weather prediction system.

The basic hypothesis underlying MDAWP research is that: Prediction of the evolution of mesoscale weather systems over New Zealand, their interaction with the orography and the synoptic systems in which they are embedded, will be significantly improved through assimilation of the information in high spatial and temporal resolution meteorological observations. An adjunct to this is that New Zealand is a near perfect location to test such a hypothesis; it is surrounded by ocean, thereby providing the best possible environment for utilising the information in high spatial resolution satellite data.

Figure 1 NZLAM (mesoscale) domain. Image GMS 11 um.

Key Science Problems

Current research focuses on the following science problems:-

1. Development of unbiased mesoscale-resolving datasets with well defined error characteristics, for use in a cycling mesoscale numerical weather prediction model;
2. Use of a data assimilation approach that will make optimal use of such datasets;
3. Use of a competitive numerical weather prediction model that can be integrated at mesoscale resolution;
4. Application to a domain that is large enough to allow synoptic development and takes advantage of New Zealand's remote location (thereby maximising the use of the information in satellite data); (see GMS image to right)
5. The development of validation systems, so that Observation System Impact Studies (OSISs) can be conducted and model background and observation error characteristics analysed.

It is assumed that data from "conventional" meteorological observing systems are also used "optimally".

NWP System

To support these research goals the Unified Model (UM) and 3Dvar data assimilation system from the Met Office have been implemented at NIWA on a T3E supercomputer (http://www.niwa.co.nz/rc/hpcf/ ). This model is run at two resolutions:-

• Global scale: 432 ´ 325 ´ 30 (60 km grid spacing) to generate lateral boundary conditions for the mesoscale domain, and
• Mesoscale: 324 ´ 324 ´ 38 (12 km grid spacing), with 3 hourly cycling and two 48 hour forecasts each day.

The mesoscale model is called the New Zealand Local Area Model (NZLAM) and the complete forecast system, NZLAM-VAR.

Mesoscale Resolving Datasets

NIWA receives and processes all GMS and NOAA transmission visible at its Greta Point Wellington reception facility. These data are are being used to develop a number of mesoscale data resolving products, as summarized below.

Sea Surface Temperature

Sea Surface Temperature (SST) is an important input to the Unified Model. An automatic multi-spectral and spatial Bayesian cloud detection algorithm has been developed (Uddstrom et al. 1999), and an SST analysis system developed that retrieves (and validates) SSTs at 1 km resolution over the NZLAM domain (Uddstrom and Oien, 1999). These data are being used in NZLAM-VAR integrations.

Figure 2: 14 day temporal composite of SST on NZLAM domain.

ATOVS Data

The ATOVS and AVHRR Processing Package (AAPP) runs operationally on all NOAA15 and NOAA16 passes, routinely producing level 1D ATOVS radiances for HIRS, AMSU-A and AMSU-B.

Cloud Classification

Bayesian cloud classification and infrared rain-rate algorithms (called SRTex) have been developed that discriminate meteorological cloud classes and predict rain-rate from AVHRR multi-spectral and radiative textural information (Uddstrom and Gray, 1996). These data have been used to develop cloud climatologies (Uddstrom et al. 2000), and will be used to validate NZLAM-VAR cloud fields, and are an important component in the interpretation of AMSU-A and -B data (Korpela and Uddstrom, 2001).

NACA

The NIWA ATOVS Collocation Archive (NACA) implements a data fusion approach to the problem of identifying radiative "contaminants" in ATOVS data (both HIRS and AMSU), and provides an opportunity to use surface sensing AMSU channels to estimate instantaneous rain-rate, cloud-liquid water and column precipitable water (pcw). NACA data are being used to identify and cloud-clear HIRS data, identify rain contamination in AMSU-A, and ice contamination in AMSU-B data, as well as retrieve pcw over the whole domain. (Korpela and Uddstrom, 2001).

GPS

Zenith-delay and total column precipitable water vapour retrieval data products from GPS stations used for seismic monitoring have been developed. These data have been validated against radiosonde soundings (e.g. Falvey and Beavan, 2000), and will be combined with NACA to provide a validation dataset for AMSU-A pcw retrievals, and hence a pcw observations over the entire NZLAM domain.

GMS High Resolution Winds

Using hourly and half-hourly (around the synoptic times) GMS data, a high resolution cloud-motion vector winds product (using the 11 m m channel) has been developed and validated against upper winds data. The NIWA winds tend to have better error characteristics than those produced by the satellite operator.

Other Local Datasets

Other datasets available to NZLAM-VAR include upper temps (i.e. rawindsondes), upper winds (i.e. wind soundings), AMDAR, ships, buoys, GMS winds (i.e. JMA), and SYNOPs. A decoding system has been developed which transforms MetService WMO encoded database records into meteorological elements for assimilation in the mesoscale model (this is discussed further below).

Radiative Transfer Modelling

In anticipation of the advent of AIRS and IASI data, a fast model (together with associated tangent linear and K models) - Gastropod - is being developed. Sherlock (2001).

NZLAM-VAR and ATOVS

Present experiments with the NZLAM-VAR system are concentrating on understanding the error covariance structure of the forecast fields. This is a preliminary step to looking at the impact of data from different observing systems on the accuracy of forecasts. Multiple experiments are being conducted for a month long period in calendar year 2000.

Data being assimilated include:-

• Conventional observations (Rawindsondes, ships, buoys, SYNOP, AMDAR, etc)
• SSM/I winds
• GMS winds
• SST
• ATOVS (HIRS and AMSU-A only (at present)

The following diagrams show some diagnostic output. The left hand panel shows a 36 hour forecast of total water (liquid, vapour and ice) at level 15 (approximately 750 hPa). The right hand panel shows NOAA15 AMSU Channel 17 (i.e. AMSU-B 150 GHz) for the same time

Click for larger version

The frontal feature in the forecast lies very close to its observed location in the AMSU data (i.e. the strong scattering signature)

Click for larger version

The panels above show an NZLAM-VAR cloud prediction (12 hour forceast) and verifying GMS image.

Research Group

Michael Uddstrom (leader), Phil Andrews, Aarno Korpela, Hilary Oliver, Vanessa Sherlock, Xiagou Zheng.

References

Korpela, A.V. and M.J. Uddstrom, 2001: The use of ATOVS, AVHRR and radar data in the development and validation of rain-rate algorithms. In the Technical Proceedings of the Eleventh International ATOVS Study Conference, Budapest, Hungary, 20 - 26 September, 2001. 191 - 202 pp

Korpela, A.V., M.J. Uddstrom and W.R. Gray, 1999: ATOVS Research at NIWA: Rain-rate algorithms. In Technical Proceedings of the Tenth International TOVS Study Conference, Boulder, Colorado, 27 Jan. - 2 Feb. 1999, 310-320.

Sherlock, V.J., 2001: Gastropod, a fast radiative transfer model for IASI and AIRS - User guide, Met Office Forecasting Research, Technical Report No 362. 37 pp.

Uddstrom, M.J., J.A. McGregor, W.R. Gray and J.W. Kidson: 2001: A high resolution analysis of cloud amount and type over complex topography. Journal of Applied Meteorology 40, 16-33.

Uddstrom, M.J. and N.A. Oien 1999: On the use of high resolution satellite data to describe the spatial and temporal variability of sea surface temperatures in the New Zealand Region. Journal of Geophysical Research (Oceans) 104, C9, 20729 - 20751.

Uddstrom, M.J., W.R. Gray, R. Murphy, N.A. Oien and T. Murray, 1999: A Bayesian cloud mask for sea surface temperature retrieval. Journal of Atmospheric and Oceanic Technology, 16, 117 - 132.

Uddstrom M.J. and W.R. Gray, 1996: Satellite cloud classification and rain rate estimation using multispectral radiances and measures of spatial texture. Journal of Applied Meteorology, 35, 839 - 858.

Acknowledgements

This research is part funded by the New Zealand Foundation for Research Science and Technology. The Met Office has kindly provided the UM, 3DVAR (and associated software and data sets) and support for implementation.