CASE #2

Date: 1-28-05

Requested Flight Track: 56

Verification region / event : (40N, 123W) Northern California precipitation event 1/29 also (33N, 88W) Southeast wave 2/1

Maximum Signal Variance: The maximum signal variance is mainly in the Pacific and obtains it's maximum value of 1.2 for the 5 day forecast. The signal variance maximum is constant for the 1 and 2 day forecasts then increases at a fairly steady rate. This suggest that the signal would both grow and propagate into the easternmost verification region.

Maximum Ensemble Spread: The maximum ensemble spread is also in the Pacific with a fairly large value of 140. The ensemble spread is higher for the 1 and 2 day European forecasts but the signal variance is small since observations in the Pacific, where the ensemble spread is greatest, would probably not propagate more than 150 degrees in 2 days. The regions of maximum ensemble spread are roughly co-located with the regions of maximum signal variance.

The areas of high ensemble spread are located over the center and east of a large surface low in the central Pacific and at the center of two lows in the Atlantic at 35W and 57W. There is also significant ensemble spread in areas of local low pressure off the California coast in the U.S. north of Mexico, and across western Europe. There are also many small regions along 70N that are not associated with any verification regions or targets.

Target Regions and Characteristics: The main target region is in the Pacific from 35-50N, 180-160W in an upper-level trough extending eastward over the downstream ridge. This trough is just downstream of a surface low where the maximum ensemble spread is located. A smaller target is located at 50N,135W just east of another upper trough and an area of low surface pressure. There is a weak target about 25 degrees to the southeast at the center of an upper level shortwave trough. For later forecast times there is a target region in the Atlantic at 40N, 35W just east of a closed upper low and above a surface low where there is high ensemble spread. The target regions using later ensembles for the European verification region are very similar with the exception of a low that develops off the east coast of the U.S. becoming an important target as well. The Pacific target region does not change much since the two lows merge and stay in the same general area, however the other target regions propagate with the upper level features. There were no ensembles available for 1-29 but the target regions would be expected to be consistent with the earlier and later times.

Observations: Since the Pacific lows are relatively stationary the target information would not be likely to propagate into any of the eastern verification regions (especially not the European region) at the verification time. The target information should be useful however for the intended California verification region since these lows are the source of the predicted heavy precipitation. It is also possible that these lows can contribute to the downstream development of cyclones. The Pacific target is associated with an upper level feature that is likely to enhance development of the surface low

Signal Variance Evolution: The initial signal variance is greatest just west of the observations. Some of this signal propagates upstream, some remains stationary while increasing and the rest propagates downstream. The downstream propagation can be followed until t = 108 when it begins to merge with the large Atlantic signal. That signal begins a s small blob of signal variance  at about 35W that is associated with a developing trough. The signal grows and spreads out along the downstream edge of this trough and over the ridge to the east. Strong local growth in both the Atlantic and Pacific overpower the sonde signal.

KE Signal:  The KE signal is the difference between 2 runs of the GFS model, one with and one without the observations from the dropsondes. The initial KE signal is located around the sondes. The signal propagates eastward and is concentrated in the base of a developing trough and spreads out downstream. Some of the signal appears to jump to a shortwave trough downstream. The signal grows much stronger as this trough deepens over the Atlantic. Again some of the signal appears to jump to a downstream trough, this time over northern Europe.

CASE #3

Date: 2-2-05

Requested Flight Track: 18

Verification region / event : (42N,71W) Heavy rain in New England 2-05 also (39N, 80W) snow in central Appalachians 2-4

Maximum Signal Variance: The signal variance increases dramatically for the 2 day forecast and remains approximately constant for the later forecasts. The maximum value of 1.6 is found in the Pacific for all verification times and regions and also in the Atlantic for the 3-5 day forecasts.

Maximum Ensemble Spread: There is fairly high ensemble spread in both the Atlantic and Pacific with a maximum value of 140. The Atlantic region of maximum ensemble spread does not coincide with the region of maximum signal variance in the Atlantic although there is fairly high ensemble spread in that region.

Target Regions & Characteristics: The main target region is centered at about 150W between 40 and 60N along the downstream side of an upper level trough and extending over the ridge into western Canada. This target is just southwest of a strong surface low. For the more eastern verification regions there is a minor target situated on the downstream side of an upper trough that is upstream from a weak low in the Gulf of Mexico and a strong target in the mid Atlantic where there is a closed upper level low directly over a surface low. This target is strongest on the east side where it is eastward of the surface feature. The maximum ensemble spread is in similar locations as the target regions however the ensemble spread is more localized over the surface features and there is an additional area of large ensemble variance off the east coast of Greenland that is not reflected in the target regions except in the case of the 24 hour European forecast. The European target regions are similar to the U.S. ones except that the target in the mid Atlantic becomes much stronger as the upper level low strengthens and a surface cyclone develops. In this particular case the targets are are often situated inside the verification region. For the European forecasts the targets propagate with the upper level features. The large Pacific target initially propagates then fizzles out. This target is again strongly indicated for the 2 day European forecast although information from this remote area is not likely to propagate into the verification region in such a short time. For this forecast the ensemble spread in the Pacific is much greater than the spread in the more dynamically relevant Atlantic regions.

Observations: The Pacific target is fairly strong for the 5 day European forecast and it is possible that these observations are able to impact the forecast for that area. It appears that there is offshore cyclogenesis or a nor-easter forecast which may effect the European verification region as well as the original U.S. east coast target. The targets are generally associated with upper level features that would be likely to intensify the surface features.

Signal Variance Evolution: In this case the initial signal is in the observation region or slightly to the east with a small signal in the Atlantic in a trough region at 40W. There is clear propagation of the signal until it merges with the signal in the Atlantic and blows up. After t = 108 it is difficult to tell whether there is further propagation or simply local growth.

CASE # 4

Date: 2-5-05

Requested Flight Track: 20/56

Verification region / event : (31N, 92W) Lower Mississippi Valley/ East Southern Plains storm system 2-6 also South Central U.S. heavy precipitation.

Maximum Signal Variance: There is strong signal variance in the Pacific for the 1 day forecast with a maximum value of 1.6 which decreases for the 2 day forecast then increases again to a maximum of 1.7 for the 4 day forecast when there is equally high signal variance in the Atlantic.

Maximum Ensemble Spread: There is a high maximum ensemble spread of 180 in the Pacific. The spread in the Atlantic is much lower at all ensemble times. The regions of maximum ensemble spread correlate to the regions of maximum signal variance but it appears as if the ensemble spread is greatest over the Hawaiian islands while the signal variance seems to be a bit larger in the central Pacific.

Target Regions & Characteristics : There are two Pacific target regions, one large regions from 170E- 160W between 40 and 55N on the downstream side of an upper low extending over an upper level ridge and along the ridge axis and another smaller region over the Hawaiian islands These regions are co-located with the maximum ensemble variance. There is also a strong target in the Atlantic centered on an upper level closed low at 40N, 60W and extending east and northward over the downstream ridge. The are surface cyclones located directly beneath the Atlantic and central Pacific targets and a closed low over the Hawaiian islands. There is also a minor target at 33N, 107W just east of an upper level low over the Baja. The target regions are similar for the European verification region with the Baja and Atlantic targets gaining importance as the forecast time decreases. The European target regions propagate with the upper level features for the later ensemble/observation dates.

Observations: The targets are generally associated with upper level features that would impact surface features by vorticity advection. The upper level features associated with the target regions as well as the targets are forecast to propagate into the verification regions. The Pacific targets remain relatively strong for 2 day European forecasts even though this information is not likely to propagate into the verification region by the verification time. There is no data for 2-8.

Signal Variance Evolution: In addition to an initial signal in the southern observation region there are two strong signals upstream in upper level lows to the west and southwest of the sonde region and another in an Atlantic low at 60W. The upstream Pacific signals remain stationary. The signal to the west grows while the one to the south decays. The sonde signal grows and propagates into the Atlantic until it merges with the Atlantic signal by about 120 hours. The Atlantic signal grows but does not appears to propagate until it starts to spread out upstream at 108 hours.

KE Signal:  The actual signal is at the  southern edge of the sondes  and is strongest in an area to the southwest of the sondes. The sonde signal bifurcates with one signal propagating rapidly in the jetstream while the other propagates to the south in  a split flow regime. The northernmost signal reaches the  Atlantic in 48 hours. The signal southwest of the sondes grows rapidly as a ridge in this region intensifies while propagating downstream with the ridge and in the downstream trough.

CASE #5

Date: 2-6-05

Requested Flight Track: 26

Verification region / event : (43N,90 W) East Central U.S. storm and (15.5N,155W)Hawaiian precipitation associated with a closed low both on 2-9

Maximum Signal Variance: There is initially relatively large signal variance(.8) along the Mexican west coast and into the U.S. southwest which decreases for the 2 and 3 day forecasts then increases again for the 4 and 5 day forecasts with a maximum of 1.2. The signal variance in the other target regions is nearly as high.

Maximum Ensemble Spread: There is a high ensemble spread maximum in this case with a value of 200. The maximum ensemble spread is found in a more localized area over Mexico and also over the Hawaiian islands. There is significant ensemble spread over two surface lows in the Atlantic and a surface cyclone and a high pressure area in the Pacific.

Target Regions & Characteristics: The main target region in this case is not over the Pacific but in an upper level trough over northwest Mexico. There are two minor Pacific targets, one over the Hawaiian islands where there are upper and surface lows and likely deep convection, and another to the northeast where there is a surface high and possibly an upper short wave. There is a weak target to the west on the east side of an upper level low which is situated over a strong surface low. For more eastern forecast regions targets are situated on the east side of a low in the west Atlantic at 40N 60W and at the midpoint between and upper level trough and downstream ridge near the southern tip of Greenland. For the European forecasts all the targets propagate with the upper level features. The target over Hawaii as well as the associated low remain stationary.

Observations: The target over Hawaii is associated with a local stationary feature and information from this area would not be likely to propagate into the other verification regions. The Pacific target is still strong for all the European forecast but would not be expected to be dynamically linked to this verification region for short term forecasts. The Atlantic targets would seem to be more viable and do seem to be related to the strengthening of a surface low over northern Europe. Again the missing ensembles and control forecast for 2-8 limit the information available for the European verification region.

Signal Variance Evolution: The strongest initial signal is at the base of a trough to the southeast of the sonde region. There are signal upstream to the southwest and west at low pressure centers and in the Atlantic as well. The upstream signals remain stationary and the signal grows in one of these regions (northwest) and decays in the other. The main signal propagates but decays somewhat and is overwhelmed by local growth once it reaches the Atlantic.

KE Signal: The actual signal is fairly strong  in the sonde region with a small signal to the southeast and a strong signal in a large area to the southwest. The sonde signal  dissapears or merges with the signal to the southeast and clearly propagates eastward into Europe. The strong Pacific signal to the southwest is relatively stationary and exhibits growth. At 36 hours an Atlantic signal appears and grows rapidly.

CASE #6

Date: 2-7-05

Requested Flight Track: 29/56

Likely Flight Track: 21

Verification region / event : (42N,90W) Winter storm across northern Plains/Great Lakes region and (60N, 150W) Alaska heavy precipitation, both on 2-9

Maximum Signal Variance: The maximum signal variance is constant for the 1 to 3 day forecast then increases for the 4 and 5 day forecasts to a maximum of 1.6. The areas of maximum signal variance are found in the Pacific for the 1-3 day forecasts and also in the Atlantic and over the continental U.S. for the 4 and 5 day forecasts.

Maximum Ensemble Spread: A large areas of maximum ensemble spread exists in the northwest Pacific and also to the south and over Hawaii. There is a smaller area across the south central U.S. and small localized areas in the Atlantic. The maximum ensemble spread is relatively low at 80.

The areas of high ensemble spread are associated with low pressure centers in the Atlantic at 35N, 60W and 65N, 35W and 55N, 80W, a large low in the Pacific at 55N, 180W, and areas of local low pressure over Hawaii, Mexico and in the Pacific at 40N, 150W.

Target Regions and Characteristics: The main target region is in the Pacific on the downstream side of an upper level trough extending eastward to a maximum at 45N, 165W just to the east of a large surface low. There is another strong target along the downstream ridge axis. There are also targets located over the Hawaiian islands, at 25N,140W where there is an upper level short wave , and also over the central U.S. on the downstream side of an upper level through that extends southward into Mexico. For the eastern verification regions there is another target area associated with a closed upper low at 30N, 60W. This target is to the north and east of the upper level feature. For the European verification region there is also a target centered in a upper trough just east of Greenland. For the European verification region the targets propagate eastward along with the upper level features for the different observation times.

Ensemble Spread: The maximum ensemble spread is located at 55N, 180W centered on a strong surface low which is almost directly beneath the upper low. The spread is also fairly large in the other target regions. There are two other areas of large ensemble spread that are not associated with targets. One is in northern Oregon where there is a fairly strong surface low and the other is north of the Great Lakes, on the downstream side of a very strong, large surface high.

Observations: The targets are generally associated with upper level features that would be expected to influence the development of the surface features. The targets are all located either downstream or in the verification regions except for the 48 hour forecast for the east coast of the U.S. where there is a target region over the low in the Atlantic. Even this target seems reasonable since variations in the strength and position of this low would be expected to influence onshore flow and moisture in the verification region. The Pacific targets are strong for the 5 day European forecast suggesting observations here might be expected to impact this forecast. The Pacific target is weaker for shorter European forecasts where there would not be sufficient time for the information to propagate into this verification region. However the Pacific target becomes strong again for the 24 hour European forecast even though it probably would not influence the short term forecast. The ensemble variance in the Pacific is not significantly greater for this time so it is not clear why this is a strong target.

Signal Variance Evolution: There is a large area of signal initially around and west of the observation region and smaller areas over the central U.S. and in the Atlantic. Some of the signal remains in the Pacific but some can also be clearly followed through t=144 where it has grown and spread out into an Atlantic trough. There is strong local growth of the Atlantic signal but it is far enough to the north to distinguish it from the other signal.

KE Signal: The KE signal is around the sondes and to the west, similar to the predicted signal. The signal spreads out and propagates in three different branches. One part immediately propagates eastward, another grows and propagates with a developing trough in the Pacific and the third takes a northern route staying in the jet stream. There is strong signal growth in the Pacific trough and weaker growth in the Atlantic. The signal appears to jump to a developing downstream trough in the Atlantic at about 120 hours.

CASE #7

Date: 2-8-05

Requested Flight Track: 21/37

Likely Flight Track: 19

Verification region / event : (43N,74W) New England snow storm and (61N,150W)Alaska heavy precipitation event both on 2-11

Maximum Signal Variance: The maximum signal variance is east of a closed low in the Atlantic for the 5 day forecast. The signal variance is pretty much constant for th 1-3 day forecast increasing for the 4 and 5 day forecasts to a maximum of 1.5.

Maximum Ensemble Spread: The maximum ensemble spread is found in the Pacific centered on a surface low and to the northwest towards another low. Ensemble spread is nearly as large in the Atlantic near the center of 2 surface lows. The ensemble spread is relatively small with a maximum value of 90.

Target Regions and Characteristics: The target regions are very similar to those from the previous case. There are targets on the downstream side of the upper level trough at 57N, 160W and over the surface low at 40N,140W. These targets have propagated downstream and northward with the upper level features and have spread outward. The eastern portion of the target region has become the most significant and it appears that the surface low there has become more well organized . There is still a target over the Hawaiian islands and the other Pacific target from the previous case has moved eastward to 20N,120W where the short wave has deepened the central U.S. trough. The central U.S. target associated with this upper trough is weaker and more localized, and has propagated eastward about 20 degrees. There is also the same Atlantic target associated with the low there that is in approximately the same position as the previous day. For the European verification region there are additional targets over the surface low east of Greenland and at the center of an upper level trough which is eroding the strong surface high at 30N,20W

Ensemble Spread: The maximum ensemble spread is located at the pacific surface low

at 40N,140W. There is also large ensemble spread at the two southern Pacific target regions, along the downstream edge of the central U.S. trough and on the northeast side of the Atlantic low. There is also a large ensemble spread associated with a low to the east of Greenland.

Observations: All the targets are either upstream from or inside the verification regions or else slightly downstream on the far side of a low where the flow would be directed towards the verification region. The only exception is the Greenland target which is very strong only for the 2 day European forecast. As in the previous cases the Pacific targets are reasonable for the 5 day European forecast but too strong for the 1 and 2 day forecasts. The Hawaiian target appears to be of local significance only as it does not propagate but weakens and disappears. Again the missing ensembles and control forecast from February 8 limit the evaluation of the European targets.

Signal Variance Evolution: The initial signal is all around the observation region and also over an Atlantic low. The Pacific signal splits into two parts, one that propagates to the north over a ridge and another that stays to the south in the center of a trough. Some of the signal remain behind of Alaska before propagating eastward starting at about 84 hours.The Atlantic signal grows locally and spreads out into the northern Atlantic where it grows more rapidly. The Pacific signal and the Atlantic one are merged by t = 120.

KE Signal:  The initial KE signal is in a large area aroung the sondes similar to the predicted signal variance. The signal splits into a northern and southen branch and at 36 hours a blob of signal appears on the U.S. east coast then dissapears. At 72 hours part of the northern signal appears to jump to the downstream side of a trough just off the east coast and grow while the more western signals propagate into this region. At 120 hours this signal spreads out over the downstream ridge into northern Europe.

CASE #8

Date: 2-9-05

Requested Flight Track: 46/21

Verification region / event : 35N,93W on 2-13

Maximum Signal Variance: The maximum signal variance for the 1 day forecast is nearly as large as for the 5 day forecast while the other forecast times have slightly lower signal variance. The maximum is in the Pacific on the east side of a closed low just west of Central America with a value of .9 for the 5 day forecast. For the shorter forecasts the maximum signal variance is in the Atlantic at the center of a closed low at 25N, 20W.

Maximum Ensemble Spread: The maximum ensemble spread is relatively low for the 2-4 day forecast at 100 and high for the 1 day forecast at 180.

Target Regions and Characteristics: Since the ensembles and control forecast is missing for 2-8 only the European verification regions are able to be evaluated. For an observation time of 2-9 the main target region is just west of Mexico on the downstream side of an upper level low. Additional targets are located on the downstream edge of upper level troughs at 40N, 150W and 40N, 80W. Another strong target exists just west of an upper level ridge at 50N, 30W. For the 1-3 day forecasts the target regions propagate eastward with the upper level features and there is an additional target at 30N, 20W where an upper level cut off low is eroding a strong surface high. These targets are similar to those for the previous day.

Ensemble Spread: The maximum ensemble spread is on the northwest side of an upper level low on the southern coast of Alaska. Large ensemble spreads are also found east of Greenland, at 30N, 130W , 55N, 120W and to a lesser extent across the U.S. along the general storm track.

Observations: The Pacific targets do not appear to propagate rapidly enough to reach the verification region. They are more likely a result of the extremely high ensemble spread in this area. The Atlantic targets look more promising. The 30N, 20W target is strong even though the ensemble spread is relatively weak for most of the observation times. The north/central Atlantic target is fairly strong until the associated feature has moved into the verification region and would be expected to be downstream of the region by the verification time.

Signal Variance Evolution: The signal variance is initially in the observation region and to the southeast. There is a small signal in an Atlantic trough region as well. Both signal grow and propagate eastward but the growth is much greater for the Atlantic signal. The Pacific signal appears to propagate into western Europe by t =144.

CASE #9

Date: 2-10-05

Requested Flight Track: 20

Verification region / event : (55N,135W) Southeast Alaska medium precipitation event on 2-11

Maximum Signal Variance: The maximum signal variance is high at .8 for the 1 day forecast then decreases slightly then increases for the 4 and 5 day forecasts to a maximum of 1.8 in the Pacific.

Maximum Ensemble Spread: The maximum ensemble spread is fairly small with a value of 100 mainly in the Pacific but also in a small area in the north Atlantic off Norway.

Target Regions and Characteristics: The main target region is centered at 20N, 120W on the downstream side of an upper level low and just upstream of a developing surface low. The next strongest target is at 42N, 150W on the downstream edge of a weak upper level trough and fairly low surface pressure. For the longer forecast times this target spreads out towards the downstream ridge. There are also minor targets over the Hawaiian islands, at the center of a shortwave upper trough at 50N,115W, on the downstream side of upper level troughs just south of the Great Lakes and in the central Atlantic, For the European forecasts there is the same additional target as the previous 2 cases at 30N, 20W. The central Atlantic and east U.S. targets are stronger for the European verification region and propagate with the upper level features. The central Atlantic target is past the verification region by 2-13. The westernmost Pacific target is strongest for the 5 day European forecast again suggesting that observations in this region are significant for these longer forecasts.

Ensemble Spread: The maximum ensemble spread is located in the Pacific at 60N,160W and 30N, 130W in the Aleutian Low and and a upper level cut off low just off the southern coast of California. Other areas of high ensemble spread are at 75N, 25W in the vicinity of the Icelandic Low, on the west U.S. coast at 55N,120W just downstream of an upper level ridge,and along the downstream side of an upper trough at 30N, 50W. Weaker but significant ensemble spread exists along the storm track at 40N extending from 170W to 50W where it then spreads out to the north along the downstream side of an upper level trough.

Observations: The Pacific targets weaken as the forecast time is shortened but they are still fairly strong even though observations here after 2-10 would not be expected to propagate into the European verification region. Targets inside the European verification region do not appear to be associated with large ensemble spread indicating that these targets are specific to this verification region. Although both the Aleutian and Icelandic lows are areas of high ensemble spread these regions are not selected as targets for any of the verification regions.

Signal Variance Evolution: In this case there is a signal in the observation region and another in the north Atlantic. Some of the Pacific signal propagates but most of it becomes locked to a trough that remains offshore in the Pacific. The growth from the Atlantic signal overwhelms any Pacific signal within 36 to 48 hours.

CASE #10

Date: 2-12-05

Requested Flight Track: 21/37

Verification region / event : (46N,88W) Southeast precipitation event and Great Lakes snow on 2-14

Maximum Signal Variance: The maximum signal variance increases pretty much linearly to a relatively low maximum of .8. The maximum is found in the Pacific and just off the southwest U.S. coast.

Maximum Ensemble Spread: The ensemble spread is about average with a maximum of 110 in the Pacific and over northern Europe. The ensemble spread increases for the late ensembles due to an area in the south Atlantic which propagates out of the search region by the 15^th.

Target Regions and Characteristics: The main target region is located at 30N, 117W just downstream from a strong upper level trough with a low pressure center. There is also a strong low pressure center at the surface here. Other strong target are at 45N,175W halfway between an upper-level trough and the downstream ridge, 40N, 160W along an upper level trough axis and extending downstream towards a ridge and just east of a surface low, at 40-55N, 70W and 20N, 5W downstream of upper level troughs and at 50-65N,25W where there is a weakness in an otherwise strong upper level ridge. The east Atlantic trough is very pronounced and almost directly over a surface cyclone. The maximum signal variance increases with the length of the forecast

Ensemble Spread: The maximum ensemble spread is located at 37N, 117W just west of the surface low, 33N, 155W, 33N in an area of weakness in a surface high to the west, 50N, 130W on the southwest side of a surface low, at 60N, 20W and 30N, 20W both in weaknesses in surface highs, and across the Icelandic low. Weaker but significant ensemble spread occurs in the area of the Aleutian low and at 55N, 70W in the northwest quadrant of a surface cyclone.

Observations: The east coast and Atlantic verification regions have upstream targets that are as strong as the ones inside the region although not as strong as the Pacific targets. These upstream targets are close enough to the verification region for the expected cyclonic flow to have some impact in the region on the regions from northeasterly flow. The Pacific targets remain strong up until the 5 day European forecast when it weakens suggesting that observations here would not be likely to impact the forecast in this region. For 2-4 day European forecasts the Pacific region is again very strong relative to the other target regions due to large ensemble variance associated with a strong upper level low and possible developing surface low. The maximum signal variance however is very small for these forecasts. Because of the low signal and distance from the verification region this target region would not be expected to be viable for European forecasts for fewer than 4 days.

Signal Variance Evolution: In this case there is very strong signal inside the dropsonde region and smaller signal in an east Atlantic low. At 12 hours a signal appears off the east U.S. coast which rapidly grows and spreads eastward so that the signal propagation from the Pacific cannot be tracked past 48 hours.

CASE #11

Date: 2-13-05

Requested Flight Track: 37

Likely Flight Track: 46

Verification region / event : (38N,123W) California precipitation event on 2-15

Maximum Signal Variance: The signal variance doubles on day 3 then increases again on day 5 to 1.4. The maximum is in the Pacific.

Maximum Ensemble Spread: The maximum ensemble spread of 180 is found in the east Atlantic but there is also fairly high spread(140) in the Pacific as well. The ensemble spread is smaller for the later ensembles.

Target Regions and Characteristics: In this case the target regions are more localized than in the previous cases. There are five target regions. The main targets are located at 45N, 165W along the downstream edge of a strong broad upper level trough and midway between a surface low and high, 33N, 140W at the center of an upper low slightly downstream of a strong surface low, and at 65N, 5W on the west side of the Icelandic low at the edge of a strong upper level ridge above the surface low pressure center. Lesser targets are found at 35N, 100W where there appears to be a short wave near the center of an upper level ridge axis and a weak surface low,and at 25N, 5W on the east side of a closed upper level low and on the northwest quadrant of a surface low. The signal increases greatly for the 5day forecast and the European verification region. At other times the targets are relatively weak. The target regions all coincide with areas of high ensemble spread. Ensemble data for 2-15 is not available

Ensemble Spread: Like the target regions the ensemble spread is also very localized. The maximum ensemble spread is located at 25N, 25W in the northwest quadrant of a weak surface low and at 33N, 145W also in the northwest quadrant of a low that is much stronger. Significant ensemble spread also occurs at 60N, 120W and 30N, 100W at the center of surface lows , at 45-70N, 165W on the east side of the Aleutian low and a small area on the southwest side of the Icelandic low.

Observations: The targets for the European forecasts propagate with the upper level features however the low off the west coast of the U.S. appears to be stationary and therefore not relevant to forecasts for this region or any other regions further than about 40 degrees downstream of this feature. This target is however the most logical for the intended precipitation forecast for California. The Icelandic low target is only strong for the European forecasts and in these instances the target is strong relative to the ensemble spread in this area.

Signal Variance Evolution: The signal variance is large at the center of the sondes but it also nearly as large at the center of a cutoff low in the southern Atlantic. There are also signals around the Aleutian islands and in the southwestern U.S.. By 12 hours a strong signal begins to appear in in Atlantic in an offshore trough. The Pacific signal spreads downstream leaving a strong signal in the sonde region for 48 hours. The Atlantic signal grows and propagates but remains far enough to the east to distinguish it from the sonde signal. The growth and propagation from the more northwest signal might add to the sonde signal making it appear stronger than in reality.

CASE # 12

Date: 2-19-05

Requested Flight Track: 25

Verification region / event : (61N,150W) Southeast Alaska precipitation on 2-21

Maximum Signal Variance: The signal variance increases in small increments to a maximum of only .8. This value is quite low compared to the other cases. The maximum values are found in a large area of the Pacific and also in the Atlantic.

Maximum Ensemble Spread: Although the maximum signal variance is low the ensemble spread is fairly high at 140, mainly in the Atlantic The spread is higher (180) for the 2/20 ensembles.

Target Regions and Characteristics: The target regions in this case consist of 3 main targets in the Pacific and one in the central Atlantic. The Pacific targets are a large area centered at 35N, 120W on the eastern side of an upper level low, 45N, 180W along the axis of a broad weak upper trough,and at 40N, 130W on the downstream side of a shortwave trough. There are surface lows almost directly beneath all of these targets. The central Atlantic target extends southward from 60N to 25N and spans the longitudes from 60W to 40W. This target is parallel and to the east of a large upper level trough axis. There is a surface cyclone at the northwest area of this target. This target becomes significant beginning with the forecast for the 21^st when it is slightly east of the verification region. This target remains significant as it propagates westward for the European forecasts. There are also minor targets along the southern Alaska coast and the western Canadian coastline. There is no target region associated with the region of high ensemble spread that is to the east of all the verification regions.

Ensemble Spread: The maximum ensemble spread is located in a large region encompassing the surface cyclone at 55N, 60W. There is also large ensemble spread in the Pacific at 45 and 55N, 175W over a double surface low. Significant ensemble spread also exists at 60N from about 160-120W not associated with any significant surface feature and at 30N, 35E in a low pressure center to the east of all the verification regions.

Observations: The Pacific targets are strong for all of the European forecasts although they would not be expected to impact short term forecasts and the associated low is still relatively stationary, moving only slightly inland. These targets would be significant for the west coast and the original forecast region in Alaska. There is a surface cyclone just east of the upper level trough associated with the central Atlantic target which would likely be affected by the strength and location of the trough so these observations would be significant for forecasts in this region. There is a strong high and two weak lows downstream that would also be affected by this feature so observations here would impact a large forecast region over a period of a few days.

Signal Variance Evolution: ab

CASE #13

Date: 2-20-05

Requested Flight Track: 26/46

Verification region / event : (39N,118W) Heavy California rainfall and also Gulf coast precipitation on 2-22

Maximum Signal Variance: The maximum signal variance is in the Pacific for the 5 day forecast with a value of 1.8. The signal variance is initially quite high at 1.1 then decreases on day 2 then jumps back up on the 3^rd day. For the European forecast the increase in signal variance with time is more steady with a doubling at day 4 then a more linear increase on day 5.

Maximum Ensemble Spread: The maximum ensemble spread is about average at 130 and is located in the Atlantic. The spread is significantly greater for ensembles on the 20^th.

Target Regions and Characteristics: The main target regions are located over the Pacific ocean along an upper level trough axis running from 35-50N, and from140-110W and along the downstream side of the same trough starting a bit further south, ending with a maximum just west of the ridge at 45N, 100W. A slightly weaker target is found along the downstream side of another upper trough off the northeast U.S. coast. For the European verification region a strong target appears for a five day forecast with observations on the 20^th at 30N, 30E at the center of the downstream side of an upper level trough. This target is initially at the southeast corner of the verification region and the upper feature moves out of the search region for late observation times. There is one more target for the European forecasts located at 20N, 20E in an area where the same upper trough protrudes westward. This target as well as the Atlantic and west-central U.S. targets move with the upper level features. The Pacific target stays in the same area and becomes concentrated in strong low that develops from the initial trough. This target would only seem to be relevant for the precipitation event that this case was targeted for. The maximum signal variance is significantly stronger for the 3-5 day forecasts for observations on the 20^th.

Ensemble Spread: There is a large area of strong ensemble spread at 70N, 60W at the center of a strong low, that appears to be a westward displacement of the Icelandic low. There are two other areas of fairly strong ensemble spread surrounding a surface low at 40N, 115W with the greatest spread to the west/southwest just below the upper trough axis and at the center of an Atlantic surface cyclone at 43N, 50W which is downstream of the upper trough.

Observations: The Pacific target remains basically stationary with the upper level feature while the target just downstream appears to be associated with a short wave that propagates eastward through the N. Atlantic trough. The Pacific target is very strong for the 4 and 5 day European forecasts although there is little propagation. The target weakens considerably for later forecasts. Although there is such large ensemble spread in the Labrador Sea area the Europeans targets in this region are about the same as those to the southeast and become minimal for later observation times, even though the ensemble spread is still large. The targets to the east are much stronger for the shorter European forecasts. This seems to suggest that other than the Pacific targets the ETKF is doing a good job of locating targets that are dynamically connected to the verification regions and not just mimicking the ensemble spread.

Signal Variance Evolution: ab

CASE #14

Date: 2-21-05

Requested Flight Track: 26/46

Likely Flight Track: 4

Verification region / event : (39N,118W) Heavy California rainfall on 2-22

Maximum Signal Variance: The maximum signal variance doubles for the 3 day forecast and then doubles again for the 4 day forecast with a maximum on day 5 of 1.8. The maximum is found in the Pacific but almost equally large values are found in the Atlantic and across the continental U.S..

Maximum Ensemble Spread: There is a large maximum ensemble spread of 180 that is found in the Atlantic. For later ensembles the spread is not so great.

Target Regions and Characteristics: The main targets are located at the center of the upper level low at 40N, 125W with a region of smaller signal variance 20 degrees southeast extending diagonally to the Great Lakes, on the eastern side of a large upper level low over the Pacific at 45N 160W, and on the downstream side of the trough in the north-central Atlantic. The two minor targets for the European region from the previous case are also indicated for observations times up until the 23^rd when the upper level features have moved east of the verification region. The more western Pacific target is stronger at all times and the maximum signal variance increases greatly for the longer range forecasts.

Ensemble Spread: The maximum ensemble spread is located at the centers of surface lows at 45N, 160W, 35N, 125W, 45N, 90W and 55N, 50W with the greatest spread near the easternmost low. The ensemble spread is large for all forecast times while the signal variance is greatest for the longer range forecasts.

Observations: The Pacific target s are indicated for all the European forecasts although the westernmost target is pretty much stationary and appears to be due to slight variations in the structure of the Aleutian low. This is probably because the ensemble spread becomes large in this region for the later observation times and not because of a dynamic connection to the verification region. These targets are however valid for the case verification region on the California coast since the upper level features would influence the surface low just off the coast. The northwest Atlantic target is important to the predicted strengthening off a surface low to the southeast. Another low further west is not influenced by this feature but might be affected by the other smaller targets. The upper level features associated with the two easternmost targets and the one that extends eastward from the south Pacific are predicted to be fast moving and therefore very likely to impact verification regions downstream.

Signal Variance Evolution: ab

CASE #15

Date: 2-22-05

Requested Flight Track: 11

Verification region / event : (53N,138W) Heavy precipitation on Southeast Alaska coast 2-23

Maximum Signal Variance: The signal variance maximum increases at a greater rate with each day up to the 4 day forecast then increases slightly on day 5 to 1.3. The maximum is found in the Pacific.

Maximum Ensemble Spread: The maximum ensemble spread is relatively small at 100 and is found in the Atlantic. The ensemble spread increases significantly for later ensembles with regions of high spread in the Pacific as well.

Target Regions and Characteristics: The main target is centered on the upper level low in the northwest Pacific at 50N, 160W and extends eastward and towards the downstream ridge. A smaller but almost equally strong target is located to the west at 40N, 170W just west of a downstream ridge. Smaller weaker targets are found at 35N, 120W in the low just off the California coast, 20 degrees to the southeast, at 40N, 90W where the height lines are closely packed, and along the downstream edge of the upper level trough over the east Atlantic from 35-60N, 50-30W. For the European verification region small targets exist around the southeast corner of the region which is an exit region for an upper trough. The main target remains strong for all verification regions but weakens greatly for later observation times. The target just upstream appears to be associated with a short wave propagating through the region which remains a target for all the European forecasts. The signal variance doubles for the 4 and 5 day forecasts and if alway much greater in the main target region than for any of the other targets except for a small area of the target off California for the 1 day forecast. All the targets except the main one propagate with the upper level features. The main target is mostly stationary while the most western target and the central U.S. targets appear to be associated with fast moving short waves.

Ensemble Spread: For the initial observation time there is a large area of high ensemble spread at about 70-75N extending from 70W to at least 20E. The ensemble spread here is not noticeable for later observation times. Ensemble spread is also large in the Pacific at 50N, 155W where a surface cyclone is located just beneath the upper level low. Slightly lower values of ensemble spread also are found in large areas encompassing the remaining target regions. There is also noticeable ensemble spread around a double centered surface low at 50N, 0 and 20E. For later observation times the ensemble spread is mainly concentrated around the lows

Observations: With the exception of the north Atlantic the target regions are all closely related to the ensemble spread and all but one of the surface features are initially at the same location as the upper level features. The exception is the surface low in the northeast Atlantic which is just downstream of the upper level trough. The main Pacific target region does not appear to be dynamically connected to the European verification region since the upper level feature is relatively stationary so that this target simply reflects the large ensemble spread. The targets associated with transient waves that move through this region would not be expected to reach the verification region either, considering the forecast rate of propagation. The northeast Atlantic trough is expected to aid in the development of a surface cyclone to the east so it would make sense to take supplemental observations in this target region. The target off the California coast is almost directly over the surface feature at the observation time and would likely only affect the short term forecast for the California verification region. The central U.S. target also appears to be associated with a short wave that is forecast to move through the Atlantic trough so that observation of this feature might impact forecasts for the eastern U.S. and eastward towards Europe.

Signal Variance Evolution: ab

CASE #16

Date: 2-23-05

Requested Flight Track: 12

Verification region / event : (60N, 140W) Southeast Alaska precipitation on 2-24

Maximum Signal Variance: The maximum signal variance increases at a greater rate each day through the 4 day forecast then decreases slightly for 5 days. The signal variance maximum of 2.0 is found in both the Atlantic and Pacific

Maximum Ensemble Spread: The maximum ensemble spread is fairly high at 140 and in found both in the Atlantic and Pacific.

Target Regions and Characteristics: The targets are in similar regions as in the previous case but the strength and structure of the targets is somewhat different. The main targets are at 60N, 150W east of the upper level low with a maximum over the downstream ridge, 25N, 100W, southeast of the low off the California coast, and on both sides of the north Atlantic trough from 40-55N,60-30W. Lesser targets are found at 35N, 120W in the low off California and to the west at 40N, 170W on the downstream side of an upper trough which is likely to be associated with a predicted short wave moving through that area. The target at 150W is strong for all verification regions and all observations times. For the European verification region this target is not so dominant. It is at best equal in magnitude to the other targets. For later observation times the targets to the east propagate eastward and are incorporated into the Atlantic target. The westernmost target appears to be associated with at least two short waves that propagate eastward through the downstream trough.

Ensemble Spread: The maximum ensemble spread is found to the east of the main target region at 40-60N, 140W to the east of both the upper and lower level low pressure centers. There is significant ensemble spread over a large area to the west. From the surface pressure values this area is predicted to develop up to three strong surface cyclones. There is a second maximum in the north Atlantic at 60N, 70W where there is a surface cyclone directly beneath the upper level trough. Fairly large ensemble spread extends downstream in the vicinity of two developing lows. There is an area of noticeable ensemble spread along the southern U.S. , gulf coast.

Observations: The only target regions that would appear to be relevant to European verification regions are the Atlantic targets and the one near Central America which propagates eastward towards the verification region. The strong Pacific target does not propagate and the westernmost target only propagates into the the stronger target region. The main target region and the target off the coast of California are directly over the surface features and are relatively stationary, their impact on local forecast would seem to be from changes due to short waves moving through the target features.

Signal Variance Evolution: ab

CASE #17

Date: 2-25-05

Requested Flight Track: 21/55

Verification region / event : (40N,75W) East coast system on 3-1, (40N,123W) North California system on 2-28 and (60N,140W) Southeast Alaska system on 2-27

Maximum Signal Variance: The signal variance is greatest in the Pacific where it initially increases slightly with forecast length then doubles for the 4 day forecast and increases greatly again for a maximum at 5 days of 2.4. This is very high in relation to the other cases. The signal variance increases similarly for the European forecasts.

Maximum Ensemble Spread: The ensemble spread is about average at a value of 110 with high ensemble spread in both the Atlantic and Pacific. The ensemble spread is much greater for ensembles on the25th in the Pacific and 28^th in the Atlantic with a maximum of 200. The signal variance is large for the 25^th but not the 28^th although both would be expected to be large.

Target Regions and Characteristics: There are two main target regions, one in the Pacific at 40N, 160W on the downstream side of a short wave trough to the south of the Aleutian low and another in the Atlantic centered on an upper level low at 45N, 45W. Fairly strong signal variance also exists along the entire Gulf of Mexico and the east coast of the U.S. both on the downstream side of upper troughs and also in the Atlantic storm track, downstream from the low. For later observation times the Pacific target stays in the basically the same location but it's structure evolves so that it is centered on the low pressure center that develops as the short wave passes. The waves associated with the eastern and gulf coast targets is forecast to merge with the Atlantic trough while the associated low pressure center propagates eastward. The offshore trough remains a target but a stronger target follows the movement of the low. As in the previous cases the Pacific target remains strong until the one day forecast without eastward propagation.

Ensemble Spread: The maximum ensemble spread is found in the same vicinity as the two main target regions, around a developing surface cyclone at 45N, 145W downstream of the upper trough and at 45N, 40W on the eastern side of a surface cyclone. Fairly strong ensemble spread is also found to the east of the Pacific region of strong spread and at 40N, 75W where there is a surface low. For the 2-4 day European forecasts there is also large ensemble spread in the western Pacific at 45N.

Observations: The Pacific target is strong at all times due to the large ensemble spread in this region. The relevance of this area to forecasts would seem to be from the modification of the Aleutian low due to transient waves. The forecast effect is the development of a surface low off the California coast. Observations in this region would only be expected to affect the west coast of the U.S.. The Atlantic targets propagate with the upper level features and appear to be dynamically connected to the verification regions. The Atlantic targets seem to be slightly downstream from the areas of high ensemble spread.

Signal Variance Evolution: ab

CASE #18

Date: 3-3-05

Requested Flight Track: 37

Verification region / event : (41N, 86W) Midwest/Northeast snow on 3-5

Maximum Signal Variance: The maximum signal variance is 1.4 for the 5 day forecast and is found in the Pacific for all verification regions and times. The signal variance almost doubles on day 3 from .5 to .9 then increases to 1.3 on day 4.

Maximum Ensemble Spread: The ensemble spread is fairly high at 130 and is also found in the Pacific except for the 3/5 ensembles where there the maximum increases to 180 and is found in both the Pacific and the Atlantic. I would expect there to be large signal variance at this time as well since the area in the Atlantic of high ensemble spread is a valid target region but this is not the case.

Target Regions and Characteristics: The strongest targets are in the Pacific at 40N, 170W and 40N, 140W. The more eastern target is along the downstream side of an upper level trough while the more western target is on the downstream side of a shortwave trough moving through the larger one. Other strong targets are located east of the Baja peninsula along what appears to be a short wave trough axis, off the eastern U.S. coast at 55N, 75 and 40N, 60W on the downstream side of an upper level trough. The more northern target extends northeast towards the downstream ridge while the southern target is more localized in an area where the trough seems to have a subtropical branch. There are also small targets at the center of an upper trough at 30N, 20W and at the middle of the downstream edge of the same trough.

Ensemble Spread: The areas of maximum ensemble spread are co-located with the target regions in the Pacific and just east of California. Large ensemble spread is found in the Atlantic on the north side of surface cyclones at 50N, 60W and 30N, 20W. These locations also coincide with two of the target regions. There is minor ensemble spread in the remaining Atlantic target regions

Observations: The Pacific targets are strong up until the 1 day European forecast because the ensemble spread is large over the entire eastern Pacific but don't appear to be dynamically connected to this verification region. The Atlantic targets are much stronger in comparison to the ensemble spread for the European verification region. These targets propagate eastward for later observation times. The targets over the Baja and east of California seem to be connected to a deepening of the upper level trough and the development of a surface low in this region. The is no propagation related to this target and observations here would only be expected to influence the forecast for this location. The western Atlantic targets are indicated for all verification regions east of 90W. For the westernmost region this target is downstream but it is close enough so that the some advection of the atmospheric properties would be expected in the verification region. These targets do appear to be dynamically related to all of the indicated verification regions. The Pacific targets appear to be stationary and only relevant to the formation of surface cyclones over the ocean which should not affect any of the verification regions for the times chosen.

Signal Variance Evolution: ab

CASE #19

Date: 3-9-05

Requested Flight Track: 36

Verification region / event : (57N,130W) Southeast Alaska precipitation on 3-10

Maximum Signal Variance: The signal variance increases to a maximum of 1.6 for the 5 day forecast. The maximum signal variance is small (.4-.7) and increases almost linearly for the European verification region.

Maximum Ensemble Spread: The maximum ensemble spread is very large 220 and is found in the Atlantic or just off the east U.S. coast. For later verification times the spread in this area is much less so the maximum spread drops to as low as 100 for 3/10 ensembles. However for the 3/11 ensembles the ensemble spread is large again. Since the ensemble spread is greatest in the Atlantic, the signal variance would be expected to be greater for verification regions that would be affected by observations in the Atlantic more than those affected by Pacific observations.

Target Regions and Characteristics: There is a huge Pacific target region in this case ranging from 30-60N, 160-120W along the downstream side of an upper level trough to the crest of the downstream ridge. Two other strong targets are located at 40N, 70W on the east side of a large upper level trough and at 40N from 40 to 20W east of a closed upper low. A smaller target is found at 20N, 120 along an upper trough axis. This target is indicated mainly for the U.S. verification regions

Ensemble Spread: The ensemble spread is largest in the Pacific target area on the southeast side of the Aleutian low and in the Atlantic at 40N,70W, and 35N, 40W at the center of two surface cyclones. There is also noticeable ensemble spread in the vicinity of Central America and central Canada.

Observations: abc

Signal Variance Evolution: ab

CASE #20

Date: 3-12-05

Requested Flight Track: 39

Verification region / event : (21N,157W) Heavy rain event across the Hawaiian islands on 3-13

Maximum Signal Variance: The maximum signal variance is in the Atlantic with a value of 1.6 for the 5 day forecast. The signal variance is low for the 1 and 2 day forecasts (.4) then begins to increase for the longer forecast periods. For later ensembles the signal variance decreases greatly and is now located in the Pacific.

Maximum Ensemble Spread: The maximum ensemble spread is very high at 220 and is also found in the Atlantic.

Target Regions and Characteristics: For the western U.S. verification region the strongest targets are centered on an upper level low off the California coast at 33N, 125W and also a large area of area over Mexico where there is a sharp pressure gradient in the eastward direction. Weaker targets are found in the Pacific at 37N, 142W at the center of an upper low and from 45-60N, 155-165W along the downstream side of an upper level trough. For more western verification regions these targets become less significant and a strong low pressure center in the Atlantic at about 45N, 45W is a much stronger target There is a minor target for the European verification region at the center of low and along the associated trough at 20N and another minor target in the Pacific on the downstream side of a shortwave trough at 37N, 145W.

Ensemble Spread: The maximum ensemble spread is in the region of the strong Atlantic target where there is a surface cyclone co-located with the upper low. The ensemble spread is also significant in the area extending northwest of the cyclones. The other regions of high ensembles spread are in the Pacific on the western edge of surface lows that are downstream of the upper level lows. There are small areas of high ensemble spread in the western Pacific which is not a target region and in the target region at 20N where there is a weak surface low.

Observations: The maximum ensemble spread is large in comparison with the previous cases so observations would be expected to have a significant impact on the analysis. The Pacific targets fizzle out for the eastern verification regions suggesting that there is little propagation from this region to the remote verification regions.

Signal Variance Evolution: ab

CASE #21

Date: 3-13-05

Requested Flight Track: 54

Verification region / event : (32N,81W) Heavy rainfall event in the Southeast U.S. on 3-17

Maximum Signal Variance: The signal variance increases steadily for days 2-5 to a maximum of 1.3 for observations taken on 3/13. For the European verification region there is a doubling of signal variance for the 4 day forecast with observations taken on 3/14.

Maximum Ensemble Spread: The maximum ensemble spread is about average for the 3/13 observations at 120 in both the Atlantic and Pacific. For the later observation time there is greater ensemble spread with a maximum of 180 for the 3-16 ensembles. This higher spread is confined to the Pacific.

Target Regions and Characteristics: There are numerous more targets for this case that are all about equally strong. In The Pacific there are targets at 30N, 160W along a deep upper trough axis, just west of the downstream ridge at 60N and midway between the same trough and ridge where there appear to be a short wave propagating through. Pacific target are also found at the center of a weak low off California at 33N, 130W and about 10 degrees southeast where there is no obvious upper level feature. There is a large strong target is the eastern Pacific at 40N, 160E along the downstream edge of a strong upper trough that would seem to be too remote to affect any of the verification regions. For later verification times this target merges with the one to the east where a strong low is forecast to develop at 30N and propagate eastward. This target is strong for all the European forecasts despite it's remote location.There are no targets over land but two fairly large targets in the Pacific at 45N, 70W and 55N, 40W at the center and slightly asymmetric to the east of two strong upper level lows. Small, slightly weaker targets are found to the east of an upper ridge axis and downstream at about 35N, 20W. The Atlantic targets propagate towards the European verification region for the later observation times.

Ensemble Spread: The maximum ensemble spread is in the eastern Pacific where there is a strong surface low beneath the upper trough and in the Atlantic were there are two surface cyclones almost directly beneath the upper lows. Significant ensemble spread occurs at two small areas of relative low pressure at 35N, 120 and 155W. There is another area of high ensemble spread at 70N, 20E that is not associated with any of the target regions.

Observations: The large area of high ensemble spread in the eastern Pacific probably exaggerates the importance of this area as a target. Equally strong ensemble spread is found in the Atlantic but these areas are not identified as targets until the verification region is in the vicinity or for short term forecast downstream. The Pacific targets would seem to be associated with the development of a surface low north of the Hawaiian islands while the Atlantic targets are related to a large strong surface cyclone that is forecast to develop in this region.

Signal Variance Evolution: ab

CASE #22

Date: 3-14-05

Requested Flight Track: 35

Verification region / event : (32N,79W) Cyclogenesis off the Southeast U.S. coast on 3-17

Maximum Signal Variance: The maximum signal variance decreases slightly from the day to the 2 day forecast then doubles for the 3 day forecast and continues to increase slightly for later forecasts to a maximum of 1.8. For the European verification region the signal variance doubles on day 2 to .6 increases then doubles again on day 4 to 1.8.

Maximum Ensemble Spread: The maximum ensemble spread is in the Pacific for all verification times and regions with a value of 140. This is fairly high in comparison with the other cases indicating sensitivity to observations.

Target Regions and Characteristics: There are three main target regions, two in the Pacific and one in the Atlantic. The Pacific regions are located at 35N, 160W at the center of a strong upper level low and at 30N, 125W in a pocket of low pressure of the western side of a broad upper level trough. The Atlantic target is located at 45N, 50W for verification region that include this target and eastward. This target focuses on the center and downstream side of a strong upper level low on the eastern edge of a broad trough. The target is modified slightly for different verification regions. For later observation times the easternmost Pacific target evolves so that it is on the downstream side of the trough that develops here and propagates eastward. The Atlantic target remains locked to the upper trough's east side as it deepens.

Ensemble Spread: The maximum ensemble spread is in the Pacific in a large area surrounding a developing surface low beneath the upper low at 35N, 160W and in a small area to the north. Secondary maximums occur in the Pacific at 35N, 120W where there is possibly a developing surface low, and at 50N, 180W on the downstream side of a large upper low. In the Atlantic there is large ensemble spread in a large area centered on the surface cyclone at 47N, 50W and 75N, 20E about 20 degrees northeast of a surface low.

Observations: The western Pacific and north Atlantic areas of high ensemble spread are not associated with target regions and would appear to have a dynamical connection with any of the target regions. The large western Pacific target is strong for the European verification region for all observation times but the associated upper level feature does not appear to propagate eastward. The Atlantic target is modified for different observation times as the upper level feature evolves. The eastern Pacific target begins in the verification region and remains strong for all verification regions and times. This target propagates eastward but it would still be expected to be west of the European verification region from the one day forecast position more that 50 degrees to the west. This target is expected to impact the forecast via the possible weakening of the pressure gradient on the south side of the Atlantic cyclone.

Signal Variance Evolution: ab

CASE #23

Date: 3-16-05

Requested Flight Track: 36

Verification region / event : Snow storm for mid-Atlantic on 3-18

Maximum Signal Variance: The maximum signal variance is constant between .6 and .8 for the 1-4 day forecasts then jumps to 1.3 for the 5 day forecast. The signal variance increases more smoothly for the European verification region. The maximum is found in the Pacific at all times.

Maximum Ensemble Spread: The maximum ensemble spread is also in the Pacific with a relatively high value of 180. The maximum decreases for late observation times.

Target Regions and Characteristics: There are four main target regions. The largest and strongest is centered on a strong upper level low over the Pacific at 35N, 150W. There is a much smaller target about 20 degrees to the northwest in a shortwave trough. Another large and fairly strong target is found at 35N, 90W along the downstream edge of an upper trough, with a signal maximum just west of the downstream ridge. A large target in the Atlantic is located from 35-65N and 50-20W near the center and along the downstream side of a strong upper trough. Small areas of maximum signal variance are found within this target region.

Ensemble Spread: The maximum ensemble spread is at the center and to the northwest of the Pacific cyclone at 28N, 150W. There is high ensemble spread in the Pacific at about 45N from 150 to 130 W with the start and end points on weak surface lows. There is another area to the north on the downstream side of the Aleutian low. In the Atlantic weaker areas of high ensemble spread are found on the south side of lows at 45N, 40W and 60N, 10W. There is relatively high ensemble spread for this case suggesting a high impact from supplemental observations.

Observations: Although the ensemble spread remains high the Pacific targets fizzle out for European forecast of less than 4 days. The Atlantic targets dominate for these verification times and propagate with the upper level features. The European targets would all seem to have a good dynamic connection to this verification region. The Pacific targets would be expected to impact the forecast of the developing surface cyclone in this region while the target in the Gulf region in related to an upper level feature that is expected to modify the Atlantic cyclone and impact forecasts for the eastern U.S. and European verification regions.

Signal Variance Evolution: ab

CASE #24

Date: 3-17-05

Requested Flight Track: 37

Verification region / event : (38N,122W) Heavy California rainfall also (35N,75W) East coast storm both on 3-19

Maximum Signal Variance: The maximum signal variance is constant at .6 until the 5 day forecast when it jumps to 1.3. For the European forecasts the SV increases from .3 to .6 for the 1-4 day forecast before increasing to 1.3. The maximums are found both in the Atlantic and Pacific region until it blows up in the Pacific at 5 days.

Maximum Ensemble Spread: The maximum ensemble spread is relatively high at 140 and is found in the Pacific but for some observation times the spread in the Atlantic is nearly as high, only in a much smaller area. For the one day European forecast the ensemble spread increases in the Pacific to 180.

Target Regions and Characteristics: The targets are similar to the previous case. The Pacific targets are unchanged while the northern Gulf of Mexico target has propagated about 20 degrees downstream so it is at 35N, 80W around South Carolina and the Atlantic target has spread out along the downstream side of the upper trough from 30-55N and 40-10 W with a maximum midway between the trough and the ridge. The Pacific target moves more eastward for the 4day European forecast then fizzles out for shorter forecast times. The Atlantic target is only strong for the European forecast although it is indicated for the Atlantic verification regions. The target on the U.S. east coast although much smaller appears to be as strong as the Pacific target while the Atlantic target is weaker for all verification regions for the 3-17 observation time.

Ensemble Spread: The maximum ensemble spread is in the Pacific centered on two surface cyclones at 45N, 170W and 30N, 145W that are directly beneath the upper level lows. For later observation times the more eastern area of high spread remains stationary while the surface cyclone propagates eastward. There is a band of high ensemble spread extending from northern Mexico arcing northeastward then following the Atlantic trough eastward over the downstream ridge with a relative maximum across the Northeast U.S. coast at 45N. The strong east U.S. target is at least 20 degrees south of the maximum ensemble spread in this region. The Atlantic regions of high ensemble spread stays locked to the surface cyclone but varies in intensity and position.

Observations: The ensemble spread is equally large in the two Pacific target regions but the two targets are not equally strong. The eastern U.S. and Atlantic targets propagate eastward for later observation times. The Pacific target splits into two parts, one that propagates and one that remains stationary only changing in shape as the upper level feature evolves. The propagating area is associated with a surface cyclone that develops over the Pacific and tracks northeastward. The Atlantic target is relevant to the development of a large surface cyclone while the eastern U.S. target tracks a shortwave that modifies the southern edge of this cyclone.

Signal Variance Evolution: ab

Summary

Target Regions:

Target regions were mainly located over the oceans and generally over much larger regions in the Pacific than in the Atlantic. The target regions were usually in the general area of the regions of high ensemble spread but differed in structure and intensity.The target regions are most often located on the downstream side of upper level troughs. in most cases the target extends to just west of the downstream ridge. In a few cases the targets extended over the ridge as well. There were also cases where the target is located in a small circular region midway between the trough and the ridge. There were a number of cases where there was a secondary target that extended southward along the ridge axis. In the cases where the upper level trough enclosed a cut off low the targets were often circular about the center of the low or centered on the low and in both instances usually stronger on the downstream side. When there was a surface cyclone present beneath the low, the target would generally be strongest in the area directly over the cyclone. Small targets in the central Pacific often appeared to be associated with short waves moving through the larger trough region or with local convective regions. Convection seemed to be the likely cause when Hawaii was targeted. There was heavy precipitation predicted for the region but no obvious upper level dynamic target. In these instances the target did not propagate for future ensemble dates, but remained stationary and eventually fizzled out. There were similar unexplained targets in the Baja region and northern Atlantic that may also be related to convection. The targets over the continental U.S. were generally in the south and in the regions where the flow of moist are from either the Gulf or Atlantic onshore would be expected. The convective targets over the U.S. did seem to propagate before dissipating. The Atlantic targets tended to be more localized at the center of closed lows in the central and eastern Atlantic although there were a significant number of cases where the Atlantic targets were found on the downstream side of troughs in the north Atlantic.

The European forecasts almost always strongly indicated the Pacific target region for the 4 and 5 day forecasts, a time period where information from observations in this region could possibly propagate into this verification region. However in many cases the Pacific was targeted for forecasts of 3 days or less when it would seem highly unlikely that information would be able to reach the verification region. In all but one of these cases the ensemble spread in the Pacific was either much greater than any of the more plausible target regions or else covered a much larger area. The one exception is the one day forecast for 2-17 ensembles where a strong ensemble spread in the Atlantic, just downstream from the target region is associated with a much weaker target than the one associated with smaller spread in the Pacific. In many cases, especially during the month of March, the Pacific target region was associated with a stationary low. In these instances it seems highly unlikely that information would propagate into the verification region even for the longer forecast periods. In all the European forecasts the targets remained locked to the upper level features, propagating at the same rate as these features with structural changes as the features evolved.

A breakdown of the targets follows. I did not distinguish between strong and weak targets. Only local areas of maximum signal variance were used to determine the major target characteristic.

Signal Variance:

The maximum signal variance ranges between .8 and 2.4 overall usually with smaller values for the short term forecasts. The signal variance generally starts out small, usually between .4 and .7 then remains constant or increases slightly with forecast time until day 3 or 4 when the value often doubles then continues to increase. In some cases the large increase at day 3 was partly due to the area of maximum variance being located in the Atlantic and not in scanned for targets before this time. In the rest of the cases the increase can be attributed to the growth in uncertainty of the forecasts with time. This time at which the the rate of increase of the signal variance maximum might also be related to the time at which non-linearities become apparent. The strong signal variance for 3-5 day forecasts suggests that supplementary observations might be most useful in improving these longer term forecasts. There is also the possibly of a strong connection between the Pacific target region and the verification region in the western Atlantic. Whether this connection is real or an artifact of the forecast model has not yet been determined.

The cases with the greatest maximum signal variance (1.8 or greater) all had maximums located in the Pacific but for cases with maximum signal variance between 1.6 and 1.8 the areas of maximum signal variance were equally divided between the Atlantic and Pacific. In all but one case (2/2/05) the area of maximum signal variance in the Pacific was much larger than in the Atlantic.

There were a few cases when the initial signal was relatively strong for the one day forecast and decreased on day 2 before beginning to increase again. The most notable instances is for case #4 where the initial signal variance maximum is 1.6. In all the cases with an initially high signal variance the target region is in the Pacific just upstream from the verification region.

Ensemble Spread vs Signal Variance:

The ensemble spread ranged from a low of 80 to a maximum of 220 with 140 being the most frequent spread. Unlike the signal variance the ensemble spread covers roughly equal areas in the Atlantic and Pacific. There is often large ensemble variance over the continental U.S. but it is usually much smaller than that over the oceans. The cases with high ensemble spread did not always equate with high maximum signal variance. For the roughly mean ensemble spread value of 140 the signal variance maximum ranged from .8 to 2 and the largest signal variance maximum occurred with the relatively low ensemble spread of 110. The most obvious explanation is that the signal variance is only great when the area of high ensemble spread is dynamically related to the verification region. This would also explain the instances where no target was associated with high ensemble spread. In cases where the ensemble spread was much greater in a region not dynamically connected to the verification region a spurious target was often produced. This was apparent for many of the European forecasts.

Signal Variance Evolution: