[Previous Article] [Next Article]

LAD Multiple Linear Regression Forecasts of Atlantic

Tropical Storm Activity for 1997

contributed by William Gray¹, Christopher Landsea², Paul Mielke³,

Kenneth Berry³ and John Knaff¹

¹Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

²Hurricane Research Division, AOML, NOAA, Miami, Florida

³Department of Statistics, Colorado State University, Fort Collins, Colorado

A specific version of multiple linear regression is used by Dr. Gray and his associates to make forecasts separately for each of several parameters of tropical storm activity in the Atlantic Basin (Gray et al. 1992). Least absolute deviation (LAD) multiple linear regression is used to develop the prediction model for these forecasts. LAD regression was discussed briefly in the December 1992 issue, and is described more fully in Gray et al. (1993a) and references therein. In essence, the calculation of the weights for the predictor variables is based on least absolute deviation (LAD) of forecasts with respect to observations, rather than on least squared deviation as in ordinary regression. Forecasts for each year's storm season are made at four times: in late November of the preceding calendar year, in early April, early June (Gray et al. 1994), and finally in early August at the beginning of the storm season. Forecast parameters include (1) named storms, (2) named storm days, (3) hurricanes, (4) hurricane days, (5) intense hurricanes, (6) intense hurricane days, (7) hurricane destruction potential, (8) net tropical cyclone activity, and (9) maximum potential destruction. The predictors used, and the skill expected, at each of the three forecast times were tabulated in the September 1993 issue of this Bulletin. However, the prediction system has been revised several times as opportunities for skill increases have emerged. In 1994, for example, prediction of the state of ENSO to be expected during the fall storm season was incorporated into the regression equation rather than being an a posteriori human adjustment (Gray et al. 1994). The following list identifies six predictor clusters, their expected influence on Atlantic tropical storm activity, and their status regarding the 1997 storm season. More detail on the influences of most of these predictor groups is found in Gray et al. (1993a, 1993b) and in the forecast bulletins on http://tropical.atmos.colostate.edu/fore-casts/index.html.

(1) The Quasi-Biennial Oscillation (QBO) at 50 and 30 mb in the northern tropics expected at the onset of the hurricane season: Westerly QBO winds enhance storm activity, while easterly winds suppress it (Gray et al. 1992). A large shear between the two levels is also a storm-inhibiting factor. In late summer 1997 the QBO is expected to be westerly at both 50 and 30 mb, which will enhance hurricane activity. The shear is not expected to be significant.

(2) El Niño/Southern Oscillation (ENSO): Warm east-central equatorial Pacific sea surface temperature (SST), or El Niño, reduces storm activity, while anom-alously cool SST enhances it. Going into summer we currently have at least a moderate El Niño in place. The outlook for this event, however, is for it not to become a strong El Niño, because the QBO is becoming westerly (which tends to inhibit El Niño) and the Singapore 100 mb temperature anomalies have not been strongly positive. Therefore, while the ENSO situation will clearly have an inhibiting effect on hurricane activity, its influence is not expected to dominate with respect to conflicting influences for 1997.

(3) African rainfall: Intense hurricane activity is enhanced when the Western Sahel and Gulf of Guinea regions in West Africa have above average precipitation the previous late summer and fall (implying favored chances for the same anomaly sign for the upcoming late summer and fall, which is the actual "predictor"), and is suppressed when that precipitation is below average (Landsea et al. 1993). Rainfall in the Western Sahel in Jun-Sep of 1996 was near average, and Gulf of Guinea rainfall last August through November was slightly below average. From these observations and from the more favorable current rainfall outlook, a near to somewhat above normal 1997 rainy season is anticipated and thus there will be a neutral to somewhat positive influence on intense hurricane activity.

(4) West Africa west-to-east surface pressure and temperature gradients: Above average west-to-east surface pressure and (often associated) east-to-west surface temperature gradients from February to May are associated with enhanced hurricane activity later that year. For February through May of this year the gradients for these variables were somewhat in the direction favoring 1997 tropical storm activity.

(5) Caribbean basin sea level pressure anomaly (SLPA) and upper tropospheric (12 km) zonal wind anomaly (ZWA): Negative anomalies of either one of these weakly imply enhanced storm activity, while positives weakly associate with reduced activity. For April-May 1997 SLPA averaged +1.1 mb, exerting some inhibiting influence on storm activity, and ZWA was near to somewhat above its mean. A slightly inhibiting influence on 1997 storm activity is therefore indicated. However, it is noted that during May and very early June the SLPA has been decreasing through the average category while the SST anomaly in both the subtropical and extratropical Atlantic has been continuing to warm (see cluster [6]). These recent factors tilt the odds against having a quiet storm season in 1997.

(6) Atlantic sea surface temperatures (SST): Negative subtropical south Atlantic SST anomalies coupled with positive subtropical north Atlantic SST anomalies enhance west African boreal summer rainfall, which are associated with increased Atlantic hurricane activity. Furthermore, broad scale positive north Atlantic SST anomalies are a rough indicator of the strength of the Atlantic Ocean thermohaline circulation, which is also positively correlated with hurricane activity. At this time, both factors point in the direction of an above average storm season for 1997.

In a newly developed prediction procedure for early June forecasts, a set of 15 potential predictors is introduced for each predictand, and the set 4 to 7 of these that maximize the partially cross-validated estimate of sample explained variance is used for each of the 9 predictands on an individual predictand basis. The use of SST in the Niño 3.4 region serves as a substitute for the previously used Niño 3 SST and the SOI (see Goldenberg and Shapiro 1995). Table 1 shows which predictors were selected for each predictand.

Table 1. The predictors chosen out of the potential pool of 15 for each of the 9 tropical storm predictands.

	N	NSD	H	HD	IH	IHD	HDP	NTC	MPD
Number of Predictors Used==>	4	5	6	6	7	5	4	5	6
1. QBO: Zonal wind, 50mb	x		x	x	x	x	x	x	x
2. QBO: Zonal wind, 30mb
3. QBO: |30mb-50mb shear|	x	x	x	x					x
4. Aug-Nov Gulf of Guinea rain		x	x	x	x	x	x	x	x
5. Jun-Sep West Sahel rain			x	x	x
6. Feb-May E-W T, W Sahel						x		x
7. Apr-May Caribbean SLP					x
8. Apr-May 200mb Caribbean Zonal Wind
9. Oct-Nov Azores SLP ridge	x		x	x					x
10. Mar Azores SLP ridge
11. Apr-May Niño 3.4 SST		x				x	x	x	x
12. (Apr-May) - (Feb-Mar) Niño 3.4 SST				x	x
13. Mar-Apr South Atlantic SST					x
14. Mar-Apr Tropical Atlantic SST		x
15. Mar-Apr Mid-Atlantic SST	x	x	x		x	x	x	x	x

The LAD multiple regression predictions, made first in November 1996 and again in early June 1997, for each tropical storm parameter are shown in Table 2. An early April 1997 forecast (not shown) is identical to the November and June forecasts this year. The mean values based on the 1950-90 period are shown to the right. The first column shows expected skill for the forecasts in terms of percent variance explained (r-squared), first for partly sample-dependent skill evalua-tions and then for forecasts on independent cases such as real-time forecasts. The latter skill estimates are based on a nonparametric resampling procedure.

Table 2. Predicted Atlantic tropical storm historical hindcast skills and forecasts for the 1997 season, as of late November 1996 and early June 1997. The most recent statistical tools are used. Skill expected for independent (real-time) forecasts made in June (shown after the slash in bold) has been estimated using resampling simulations. These are equal to approximately 84% (usually between 75% and 90%) of the hindcasting skill levels (shown before the slash) obtained using a partial cross-validation design.
ATLANTIC TROPICAL CYCLONE PARAMETER	Skill (% variance)	Nov 1996 Forecast	Jun 1997 Objctv Fcst	Jun 1997 Final Fcst	1950-90 Mean
Named storms	.54/.45	11	10.7	11	9.3
Named storm days	.58/.49	55	43.6	55	46.9
Hurricanes	.53/.45	7	6.5	7	5.8
Hurricane days	.60/.51	25	34.0	25	23.7
Intense hurricanes	.63/.53	3	6.0	3	2.2
Intense hurricane days	.63/.53	5	8.8	5	4.7
Hurricane destruction potential	.58/.49	75	116	75	70.6
Net tropical cyclone activity (% of avg)	.65/.56	110%	129%	110%	100%
Maximum potential destruction	.68/.57	70	68.7	70	61.7

Somewhat above-average tropical storm activity is predicted for 1997. The season is not expected to be as active as 1995 and 1996, however.

It should be noted that if the 1997 season is as active as predicted, the 1995-1997 period will be the most active 3-year period for Atlantic hurricanes in this century.

Goldenberg, S.B. and L.J. Shapiro, 1995: A new look at the relationships between El Niño, West African rainfall, and north Atlantic tropical cyclone activity. Preprints, 21st Conference on Hurricanes and Tropical Meteorology. American Meteorological Society, April 24-28, Miami, Florida, 585-587.

Gray, W.M., C.W. Landsea, P.W. Mielke, and K.J. Berry, 1992: Predicting Atlantic seasonal hurricane activity 6-11 months in advance. Wea. Forecasting, 7, 440-455.

Gray, W.M., C.W. Landsea, P. Mielke and K. Berry, 1993a: Predicting Atlantic basin seasonal tropical cyclone activity by 1 August. Wea. Forecasting, 8, 73-86.

Gray, W.M., J.D. Sheaffer, P.W. Mielke, K.J. Berry and J.A. Knaff, 1994: Predicting ENSO 9-14 months in advance. Proceedings of the 18th Annual Climate Diagnostics Workshop, Boulder, Colorado, November 1-5, 1993, 390-393.

Gray, W.M., C.W. Landsea, P.W. Mielke and K. Berry, 1994: Predicting Atlantic basin seasonal tropical cyclone activity by 1 June. Wea. Forecasting, 9, 103-115.

Landsea, C.W., W.M. Gray, P.W. Mielke and K.J. Berry, 1993: Predictability of seasonal Sahelian rainfall by 1 December of the previous year and 1 June of the current year. Preprints, 20th Conference on Hurricane and Tropical Meteorology, AMS, San Antonio, Texas, 473-476.

[Previous Article] [Next Article]