TROPICAL CYCLONE BOGUS
Tropical cyclone bogus observations are manually entered by the Darwin RSMC when tropical cyclones are in the TXLAPS_PT375 domain.
These tropical cyclone data are used by the bogus program to implant the tropical cyclone data into the three TXLAPS_PT375 analyses.
The bogus program:
1) locates any circulations in the initial GASP analysis (T-12) and in the GASP forecasts at T-6 and T-12 close to the bogused tropical cyclones;
2) locally removes these circulations by careful filtering;
3) inserts an axisymmetric vortex as close as possible to the bogused locations;
4) builds large scale wind field asymmetries (β-gyres) consistent with the past 12-hour motion of the tropical cyclone;
5) modifies the existing box data files used in the objective analysis so that they contain synthetic observations in and about the tropical cyclone which are consistent with the modified GASP fields;
The modified GASP analysis, with the synthetic TC vortex implanted, is used as the first guess for the TXLAPS_PT375 analysis in the first objective analysis.
The modified box data files, with synthetic TC observations are then used in all three TXLAPS_PT375 analyses.
The overall tropical cyclone bogus scheme is described in APOB57 and APOB48 and is detailed in Davidson & Weber 2000.
DYNAMICAL NUDGING
Cloud top temperatures (CTT) are extracted from hourly GOES satellite imagery (MTSAT imagery will be used from mid 2005) and averaged in each 0.5 degree box over the region 80oE to 180o, 50oS to 48oN. This data is used to input predetermined synthetic moisture and empirical heating profiles into the model in observed cloudy regions at hourly intervals while the model in being nudged towards the target T-6 and T-0 analyses. This procedure forces more physically realistic up-motion and divergence in and about meso-scale cloud regions while retaining the rotational wind field of the target analyses.
Data from MTSAT rather than GOES will be used when MTSAT becomes operational in May 2005.
SOIL MOISTURE ADJUSTMENT
The surface soil moisture and temperature is initialised over the Australian continent and Tasmania at the start of each model forecast, using results from the previous 00UTC run of the Pescod soil moisture scheme. For land areas other than Australia a soil climatology is used.
The subsurface moisture and temperature is initialised during the soil-moisture adjustment step. The 6 hour forecast from either the assimilation or in the case of the first assimilation cycle from the previous TXLAPS_PT375 run is used as the first guess. The ECMWF (Viterbo, 1996) land surface scheme is then used in the prediction step to adjust the surface and subsurface moisture and temperature.
The ECMWF land surface scheme provides a detailed vegetation and soil type.
DIGITAL FILTER INITIALISATION AND PREDICTION MODEL
Initialisation is incorporated in the prediction model to control the generation of spurious gravity waves; and is based on a digital filtering technique. The forecast component is a hydrostatic primitive equation model formulated on sigma levels for a non‑staggered ("Arakawa A") latitude-longitude grid. Higher order numerics are a feature of the system. Detailed physical parameterisations, in line with those in GASP include: a mass-flux convective scheme (for deep, mid-level and shallow convection), large-scale rain, radiative transfer with a diurnal cycle, diagnostic clouds, stability dependent surface fluxes, and interactive soil moisture. The horizontal grid and vertical level structure used in the prediction model is identical with that of the analysis component.
The nudging phase of the prediction model runs from T-12 to T-0. The prediction model proper then integrates out to 72 hours.
BOUNDARY CONDITIONS
The GASP analysis and forecasts, from T-12 to T+72, are used to define the necessary lateral boundary conditions for TXLAPS_PT375. Absolute values of the mean sea level pressure, wind components, temperatures and mixing ratios are used at 6 hourly intervals throughout the nesting procedure. The nesting files are currently derived from the 1.50 latitude-longitude post-processed files from GASP (at present the T239/29L version).
Model Changes
The previous operational version of TLAPS_PT375 was using the "unified" bam physics code (version 3.3). TXLAPS_PT375 was tested using bam 4.0 physics but reverted to bam 3.3 after testing by BMRC found problems with the 4.0bam code. During the TLAPS_PT375 versus TXLAPS_PT375 trial period (described below) both models were running using bam 3.3 physics.
Objective Assessment
A parallel trial of TXLAPS_PT375 versus TLAPS_PT375 was conducted. Traditional verification parameters such as S1 Skill score, bias and root mean square (RMS) error, along with surface weather parameters such as the screen temperature and 10m winds were also verified over the period 15th November 2004 – 14th December 2004. A synoptic trial was also conducted by forecasters in the Darwin Regional Office for the same period. A comparison of rainfall statistics was done for the period 15th November 2004 – 14th February 2005.
The traditional skill scores indicate little difference between the two models however the comparison of rainfall statistics and surface observations show some improvements. The forecaster trial also demonstrated a number of improvements of the new system over the old system.
1. S1 Skill Score, Bias and Root Mean Square error for MSLP and WIND
Table 3 presents the mean S1 skill score, bias and RMS error for TLAPS_PT375 versus TXLAPS_PT375 over the Australian Region and the Australian Tropical verification region. The S1 skill scores for MSLP were consistently better for TXLAPS_PT375 than for TLAPS_PT375. There were mixed results for the BIAS and RMS results over both regions but differences were small.
| Table 3 |
Mean S1 Skill Score, Bias and RMS Error for MSLP and WIND comparing TLAPS_PT375 TXLAPS_PT375. |
|
| Regions |
1. Australian Region (45S - 15S, 100E - 170E) |
|
| |
2. Australian Tropics (25N - 25S, 100E - 160E)
|
|
| Time Period |
15 Nov 2004 - 14 Dec 2004 using 00 and 12 UTC runs (62 cases) |
|
| Verifying Analyses |
Self |
|
Change |
The difference between TLAPS_PT375 and TXLAPS_PT375 score; where the sign is set so that a positive (blue) value indicates that TXLAPS_PT375 has better score than TLAPS_PT375. |
|
| |
|
Australian Region |
Australian Tropics |
|
| Forecast Period |
Verification statistics |
Field |
TLAPS |
TXLAPS |
Change |
TLAPS |
TXLAPS |
Change |
| +24HRS |
S1 Skill Score |
MSLP |
31.3 |
31.0 |
0.3 |
50.1 |
45.7 |
4.4 |
| BIAS |
MSLP |
0.2 |
0.3 |
-0.1 |
-0.5 |
-0.2 |
0.3 |
| 850hpa U,V |
0.0, -0.2 |
0.0, -0.2 |
0.0, 0.0 |
0.2,-0.2 |
-0.1,-0.2 |
0.1,0.0 |
| 500hpa U,V |
0.0, -0.1 |
-0.1, -0.1 |
-0.1, 0.0 |
0.2, 0.1 |
0.2, 0.1 |
0.0,0.0 |
| 200hpa U,V |
0.0, -0.1 |
0.1, 0.0 |
-0.1, 0.1 |
0.2, 0.3 |
0.3, 0.4 |
-0.1, -0.1 |
| RMS |
MSLP |
1.3 |
1.2 |
0.1 |
1.1 |
1.0 |
0.1 |
| 850hpa U,V |
2.5, 2.5 |
2.5, 2.6 |
0.0, -0.1 |
2.7, 2.5 |
2.7, 2.6 |
0.0, -0.1 |
| 500hpa U,V |
3.1, 3.2 |
3.3, 3.5 |
-0.2, -0.3 |
2.8, 2.7 |
3.0, 2.9 |
-0.2, -0.2 |
| 200hpa U,V |
4.6, 4.9 |
4.8, 5.0 |
-0.2, -0.1 |
4.3, 4.4 |
4.6, 4.6 |
-0.3, -0.2 |
| +48HRS |
S1 Skill Score |
MSLP |
40.3 |
39.8 |
0.5 |
55.6 |
51.2 |
4.4 |
| BIAS |
MSLP |
0.6 |
0.6 |
0.0 |
-0.1 |
0.0 |
0.1 |
| 850hpa U,V |
0.0, -0.3 |
-0.1, -0.3 |
-0.1, 0.0 |
0.2, -0.1 |
-0.1, -0.1 |
0.1,0.0 |
| 500hpa U,V |
-0.1, -0.3 |
-0.2, -0.2 |
-0.1, 0.1 |
0.3, -0.1 |
0.3, 0.0 |
0.0,0.1 |
| 200hpa U,V |
-0.4, -0.4 |
-0.2, -0.3 |
0.2, 0.1 |
-0.2, 0.3 |
0.0, 0.4 |
0.2,-0.1 |
| RMS |
MSLP |
2.1 |
2.0 |
0.1 |
1.2 |
1.2 |
0.0 |
| 850hpa U,V |
3.2, 3.3 |
3.2, 3.4 |
0.0, -0.1 |
3.2, 3.1 |
3.3, 3.1 |
-0.1, 0.0 |
| 500hpa U,V |
4.2, 4.5 |
4.4, 4.7 |
-0.2, -0.2 |
3.5, 3.3 |
3.7, 3.5 |
-0.2, -0.2 |
| 200hpa U,V |
6.3, 6.9 |
6.4, 6.8 |
-0.1, 0.1 |
5.0, 5.0 |
5.1, 5.1 |
-0.1, -0.1 |
| +72HRS |
S1 Skill Score |
MSLP |
47.0 |
46.3 |
0.7 |
59.4 |
54.9 |
4.5 |
| BIAS |
MSLP |
0.7 |
0.7 |
0.0 |
0.1 |
0.2 |
-0.1 |
| 850hpa U,V |
0.0, -0.1 |
-0.2, -0.1 |
-0.2, 0.0 |
0.3, 0.0 |
0.1, 0.1 |
0.2,-0.1 |
| 500hpa U,V |
-0.2, -0.1 |
-0.3, -0.1 |
-0.1, 0.0 |
0.2, 0.1 |
0.2, 0.1 |
0.0, 0.0 |
| 200hpa U,V |
-0.5, -0.6 |
-0.2, -0.4 |
0.3, 0.2 |
-0.8, 0.0 |
-0.5, 0.2 |
0.3,-0.2 |
| RMS |
MSLP |
2.6 |
2.6 |
0.0 |
1.4 |
1.4 |
0.0 |
| 850hpa U,V |
3.7, 3.9 |
3.7, 3.9 |
0.0, 0.0 |
3.6, 3.4 |
3.7, 3.5 |
-0.1, -0.1 |
| 500hpa U,V |
4.9, 5.3 |
5.1, 5.4 |
-0.2, -0.1 |
3.9, 3.6 |
4.1, 3.7 |
-0.2, -0.1 |
| 200hpa U,V |
7.6, 8.3 |
7.6, 8.2 |
0.0, 0.1 |
5.4, 5.5 |
5.5, 5.6 |
-0.1, -0.1 |
|
|
|
|
|
|
|
|
|
|
S1 skill scores are non-dimensional. Units for bias and RMS:- MSLP (hPa) for U& V wind (ms-1).
| 2. Forecaster Trial
Darwin Regional Office forecasters compared the analyses and model output at different time steps for TLAPS_PT375 versus TXLAPS_PT375. The forecaster’s brief was to look at significant synoptic differences:-
- between the models;
- between the model and the analysis for the same time;
- and between both models and Darwin Regional Specialized Meteorological Analyses for the same time.
The following parameters were considered:- MSLP 700hPa, 200hPa winds, 850hPa vorticity and the 24hour rainfall. A number of consistent trends appeared. The rainfall trends were corroborated by some rainfall statistics. A summary of results are given in table 4.
|
Table 4: Results of Forecaster Trial |
| Parameter |
comments |
| 700hPa Wd |
Both models underestimated the SE trade flow across the tropics. But TLAPS_PT375 generally did worse than TXLAPS_PT375. Occasionally TXLAPS_PT375 did worse than TLAPS_PT375. Winds about systems, including tropical cyclones were too weak, particularly at longer prognosis times:- this was worse for TLAPS_PT375 than for TXLAPS_PT375. |
| 200hPa Wd |
Both models regularly underdeveloped upper troughs. |
| 850hPa/MSLP |
Strength of systems, including tropical cyclone underdone, particularly at longer prognosis times:- this was worse for TLAPS_PT375 than for TXLAPS_PT375. |
| 24hr Rainfall |
Rain area and rain amounts often overestimated by TLAPS_PT375. Occasionally TXLAPS_PT375 did the same. For middle level cloud bands across central Australia - TXLAPS_PT375 sometimes underestimated or missed areas or rain, particularly if rain was light; while TLAPS_PT375 did better. |
| Location of TCs |
Position of tropical cyclones a problem with both models particularly with longer term prognosis. |
|
3. Surface Weather Parameter Verification against the Aviation METAR Observations
NMOC’s Surface Weather Parameter Verification System, was developed by Kevin Tory of the BMRC Model Development Group, and verifies weather elements such as surface (2m, or screen level) temperature and dewpoint and surface (10m) wind direction and speed against METAR observational data. RMS errors and biases are calculated at regular intervals out to a specified forecast period using all available Australian METARS (approximately 297).
 
|
| Figures 3 and 4 show the time series plots of the mean bias and RMS errors averaged over all
METAR sites and over all hourly model forecast periods out to +72 hrs. Results are shown
for the period 15 November to 14 December 2004. These results show a similar skill between
the two models.
|
|
5a)
5b)
5c)
Figure 5. Screen (2m) dry bulb temperature comparison of forecasts against METAR observations from
(a) Alice Springs, (b) Townsville and (c) Darwin for the previous version (TLAPS) and the new version (TXLAPS).
6a)
6b)
6c)
|
Figure 6. 10m wind speed comparison of forecasts against METAR observations from
(a) Alice Springs, (b) Townsville and (c) Darwin for the previous version (TLAPS) and the new version (TXLAPS).
.
Figures 5 and 6 show 2m temperatures and wind speed forecasts respectively, for 48-hour period commencing at 0000UTC on 15 November 2004 for (a) Alice Springs, (b) Townsville and (c) Darwin. There were some improvements in temperature at Townsvillle and Darwin; while results for wind speed were mixed.
|
|
4. Rainfall Verification
Table 6 and figure 7 show a comparison of a number of rainfall forecast skill parameters which were generated using the Bureau's RAINVAL system (McBride & Ebert 1999) for the period from 15th November 2004
to the 14th February 2005. These parameters are defined in table 5. The statistics in table 6 and figure 7 are calculated over the Australian mainland.
| Table 5. Definitions of rainfall statistical parameters. |
| Name of statistical parameter |
Denoted by |
Given by |
| Rain Area |
RA |
Forecast and observed rain area |
| Probability of detection |
POD |
H/(M+H) |
| false alarm ratio |
FAR |
F/(F+H) |
| Hanssen & Kuipers Score |
HKS |
accuracy for event(POD) + accuracy for nonevents -1 or in other words (H/(M+H) + Z/(Z+F) - 1)
|
| Equitable Threat Score |
ETS |
(ZH – FM) / ((F + M)(total no. forecasts) + (ZH - FM)) |
| Mean Absolute Error |
MAE |
| Forecast – Observation | |
Where
|
|
Predicted |
Observed |
No Rain |
Rain |
No Rain |
Z |
F |
Rain |
M |
H |
TXLAPS_PT375 generally performed better than TLAPS_PT375. Trends in the 00hours-24hour period were generally mirrored at 24hour-48hours and 48hours-72hours.
While both model RA time series followed the general trends of the observed RA time series there were some significant differences between the observed and forecast values, and between the forecasts from the two models. The RA time series for TLAPS_PT375 overestimate RA to a greater extent than TXLAPS_PT375; particularly during periods of maximum observed RA. The overestimation was more pronounced for later forecast periods. Not only were the TLAPS_PT375 RA maxima too high, but the TLAPS_PT375 time series tended to pre-empt the observed increases in RA. TXLAPS_PT375 forecast RA trends were more in phase with the observed RA trends than TLAPS_PT375. For the later forecast periods both models kept RA values high for too long.
MAE increased (ie became worse) for both models from 00hours-24hour to 24hours-48hours. Average values of MAE were better for TXLAPS_PT375 than for TLAPS_PT375 over all forecast periods. Both models had peaks in MAE time series which corresponded to maxima in the observed RA. Generally values of the MAE time series were fairly similar for both models.
The average value of POD for TLAPS_PT375 was higher (ie better) than for TXLAPS_PT375. The time series values sometimes showed better values for TXLAPS_PT375 than for TLAPS_PT375. TLAPS_PT375 RA values were generally overestimated.
The average value of FAR and the time series values were consistently higher (ie worse) for TLAPS_PT375 than for TXLAPS_PT375 over all forecast time periods. HKS measures the accuracy of events and non events. The average value and the time series values of the HKS were consistently higher (ie better) for TXLAPS_PT375 than for TLAPS_PT375.
The ETS is a measure of the correct event forecasts (ie hits) divided by the number of cases forecast and/or observed events; adjusted to take into account the number of forecast events that would be expected purely by chance. Like the HKS the average value and the time series values of the ETS were consistently higher (ie better) for TXLAPS_PT375 than for TLAPS_PT375.
The Darwin Regional Office forecaster trial picked up on some of the issues above such as the over estimation of rain and rain areas by TLAPS_PT375 as compared to TXLAPS_PT375. In other words the forecasters picked that the FAR for TLAPS_PT375 was higher than for TXLAPS_PT375. Also that TXLAPS_PT375 missed rain events to a greater extent thanTLAPS_PT375, but that this was mainly when the rain was light.
|
Table 6: Average Australian statistics 00hrs-24hrs 15th November 2004 to 14th February 2005
|
|
TLAPS_PT375 |
TXLAPS_PT375 |
|
Rain Area (km2) |
1263 |
1274 |
|
Mean Absolute Error (mm/day) |
1.92 |
1.81 |
|
Probability of Detection |
0.74 |
0.72 |
|
False Alarm Ration |
0.44 |
0.36 |
|
Hanssen & Kuipers Score |
0.58 |
0.62 |
|
Equitable Threat Score |
0.35 |
0.41 |
Figure 7: Time Series statistics 00hrs-24hrs (All Australia averages) 15th November 2004 to 14th February 2005
|
Figure 8: Darwin time series 24hr Rainfall (mm/day) – 15th Nov 2004 to 14th February 2005
Figure 8 shows a time series
of observed, TLAPS_PT375 and TXLAPS_PT375 Darwin 24hour rainfall data. During
the first month of the time series, when the rainfall totals were lower than
later in the period, there was a far greater coincidence between the observed
and the 00hour-24hours and 24hours-48hours TXLAPS_PT375 forecasts than for the
TLAPS_PT375 forecasts. Both models
missed the start of the rain event on 11th -12th Jan and continued to forecast rain after
the event had finished. The 24-48hour
TLAPS_PT375 forecast over estimated the rainfall from the 4th-6th February event to a greater extent than TXLAPS_PT375.
Product Availability
The move from TLAPS_PT375 to TXLAPS_PT375 should be
transparent to most, apart from the change in labelling from TLAPS to
TXLAPS. A number of full domain charts are on the Bureau's external web.
There is an extra suite of charts covering the
central South Pacific. These new charts are available on the Bureau's RSMC external web page.
Email Joan Fernon (J.Fernon@bom.gov.au) for more information.
DIFACS
There are no changes to the
difacs suite of charts. Those charts that were TLAPS_PT375 are now TXLAPS_PT375 charts.
Any
requests concerning DIFACS should be sent by email to: difacs@bom.gov.au
Real-Time Database (rtdb)
Sigma and pressure level
data from TXLAPS_PT375 is written to the real time database (rtdb) during each
model run. Forecast fields out to +72 hours are available for the full domain,
in 6 hourly intervals. The database currently holds TXLAPS_PT375 model fields
for the last 8 days and analysis fields for the past 32days.
Currently TXLAPS_PT375 runs
on a 420x250 horizontal grid and on 29 sigma levels in the vertical. However
due resource limitations (both in terms of CPU power and disc storage), the
multi-level fields are put into NMOC's real time data base (rtdb3)
at a coarser horizontal resolution of 210x125 and only a small number of single level-fields are available in rtdb3 at the full horizontal resolution of 420x250.
Tables 7a and 7b show the TXLAPS_PT375 fields which are available through
rtdb3.
| Table 7a: Coarse resolution TXLAPS_PT375 fields in the NMOC rtdb |
| Horizontal Resolution |
210x125 lat-lon grid |
| Vertical Resolution |
29 sigma levels (sgma_lvl), as shown in table 2.
12 pressure levels (isbr_lvl): 1000, 950, 900, 850, 700, 500, 400, 300, 250, 200, 150, 100 hPa
(Note: Dew point temperatures, Mixing Ratio and Relative Humidity are only ingested to 300hpa.) |
| Temporal Resolution |
6-hourly from 00 to +72 (at 00 and 12UTC) |
FIELD
(Common Name) |
FIELD
(rtdb Name) |
surface |
isbr_lvl |
sgma_lvl |
UNITS |
| air temperature |
air_temp |
Yes |
Yes |
Yes |
K |
| wind u-component |
wnd_ucmp |
Yes |
Yes |
Yes |
m s-1 |
| wind v-component |
wnd_vcmp |
Yes |
Yes |
Yes |
m s-1 |
| wind speed |
wnd_spd |
No |
Yes |
No |
m s-1 |
| pressure |
pres |
Yes
(and MSL) |
No |
No |
pa |
| precipitation |
prcp |
Yes |
No |
No |
mm |
| geopotential height |
geop_ht |
No |
Yes |
Yes |
m |
| mixing ratio |
mix_rat |
No |
Yes |
Yes |
kg kg-1 |
| vertical velocity |
omega |
No |
Yes |
Yes |
pa s-1 |
| dew point temperature |
dwpt |
No |
Yes |
No |
K |
| vorticity |
vor |
No |
Yes |
No |
s-1 |
| relative humidity |
rel_hum |
No |
Yes |
No |
% |
| total-totals index |
tot_tot |
Yes |
No |
No |
- |
| topography |
topg |
Yes |
No |
No |
m |
.
| Table 7b: Full resolution TXLAPS_PT375 fields in the NMOC rtdb. |
| Horizontal Resolution |
420x250 lat-lon grid |
| Vertical Resolution | |
|
