ALBERT

Description: The receiver operating characteristic (ROC) curve is a two-dimensional measure of classification performance. The area under the ROC curve (AUC) is a scalar measure gauging one facet of performance. In this short article, five idealized models are utilized to relate the shape of the ROC curve, and the area under it, to features of the underlying distribution of forecasts. This allows for an interpretation of the former in terms of the latter. The analysis is pedagogical in that many of the findings are already known in more general (and more realistic) settings; however, the simplicity of the models considered here allows for a clear exposition of the relation. For example, although in general there are many reasons for an asymmetric ROC curve, the models considered here clearly illustrate that an asymmetry in the ROC curve can be attributed to unequal widths of the distributions. Furthermore, it is shown that AUC discriminates well between “good” and “bad” models, but not between good models.

Description: Lake Okeechobee, located in southern Florida, is an important component in the regional hydrologic system. Currently, the Southeast River Forecast Center (SERFC) is setting up a forecasting scheme for Lake Okeechobee and its major inflows. An important aspect in calibrating the system is estimating the depth of direct precipitation over the water surface. Within this project, National Weather Service (NWS) and South Florida Water Management District (SFWMD) surface gauges, along with stage III multisensor precipitation estimates, are used to create time series of mean areal precipitation (MAP). The computed MAP values are compared in order to find the relative differences between them, and to determine the utility of using each data source for calibration and in future operations. It was found that the SFWMD gauge-based MAP was the most useful data source, because it had a suitable period of record and the SFWMD gauges had a better spatial sampling of precipitation over the lake surface. The radar-based stage III estimates were not found to be a useful source of data, despite the superior spatial sampling resolution, because they had too short a period of record and a number of changes in the processing algorithms made the associated MAP nonhomogeneous and inappropriate for model calibration.

Description: A prime challenge for ENSO seasonal forecast models is to predict boreal summer ENSO conditions at lead. August–September ENSO has a strong influence on Atlantic hurricane activity, Northwest Pacific typhoon activity, and tropical precipitation. However, summer ENSO skill is low due to the spring predictability barrier between March and May. A “consolidated” ENSO–climatology and persistence (CLIPER) seasonal prediction model is presented to address this issue with promising initial results. Consolidated CLIPER comprises the ensemble of 18 model variants of the statistical ENSO–CLIPER prediction model. Assessing August–September ENSO skill using deterministic and probabilistic skill measures applied to cross-validated hindcasts from 1952 to 2002, and using deterministic skill measures applied to replicated real-time forecasts from 1900 to 1950, shows that the consolidated CLIPER model consistently outperforms the standard CLIPER model at leads from 2 to 6 months for all the main ENSO indices (3, 3.4, and 4). The consolidated CLIPER August–September 1952–2002 hindcast skill is also positive to 97.5% confidence at leads out to 4 months (early April) for all ENSO indices. Optimization of the consolidated CLIPER model may lead to further skill improvements.

Description: A search of radar mosaics and level-II Weather Surveillance Radar-1988 Doppler (WSR-88D) data revealed 51 cold-season (October–April) bow echoes that occurred in the contiguous United States from 1997–98 to 2000–01. Proximity soundings indicated mean 0–2.5-, 0–5-, and 5–10-km shear values of 14, 23, and 19 m s−1, respectively. Mean CAPE was 1366 J kg−1. Most bow echoes developed from squall lines, groups of cells, or squall lines overtaking cells that originated in the path of the squall line. Overall, cell mergers occurred just prior to the development of 34 (67%) of the 51 bow echoes, and embedded supercells were present in the mature stage of 22 (43%) bow echoes. Nine severe, long-lived bow echoes (LBEs) were identified, and seven of these had damage paths that met derecho criteria. LBEs developed in strongly forced, dynamic synoptic patterns with low to moderate instability. As in previous observational studies, proximity soundings suggested that LBEs are possible within much wider ranges of sampled CAPE and shear than idealized numerical modeling studies have indicated. Cold-season bow echoes formed overwhelmingly (47 of 51) in southwesterly 500-mb flow. Twenty (39%) bow echoes formed in a Gulf coast synoptic pattern that produced strong shear and moderate instability over the southeastern United States. Nineteen (37%) and seven (14%) bow echoes, respectively, formed in the plains and east synoptic patterns, which resemble classic severe weather outbreak patterns. Four (8%) bow echoes developed in a northwest flow synoptic pattern that produced strong shear and moderate instability over the southern plains.

Description: Two banded, heavy snowstorms that occurred over the northern mid-Atlantic region are compared and contrasted. On 6–7 January 2002, a narrow, intense band of heavy snow was observed, along with several other weaker bands, embedded within a large area of moderate snow. On 19–20 January 2002, a single, broader band of heavy snow was observed, embedded within a broken area of light snow. The synoptic-scale settings associated with these two storms were strikingly dissimilar. In the first case, strong quasigeostrophic (QG) forcing for ascent was present just to the south of the heavy snowfall area. A highly amplified longwave trough was located over the Mississippi River valley, while a compact shortwave trough moved northward, up the east side of the longwave trough. The result was robust cyclogenesis off of the mid-Atlantic coast. In the second case, the relatively weaker QG forcing for ascent was located much farther southwest of the snowband. The flow aloft was much less amplified, with weaker cyclogenesis occurring off of the mid-Atlantic coast. Analysis of the frontal scale environments for both cases indicated that the snowbands were each associated with the collocation of midtropospheric frontogenesis and reduced stability. In the first case, evidence is shown that a layer of potential symmetric instability (PSI) was located just above a deep, sloping zone of frontogenesis, in the presence of deep near-saturated conditions. In the second case, evidence is shown that a layer of potential instability (PI), associated with rapidly decreasing relative humidity with height, was located just above a shallow, sloping zone of frontogenesis. In addition, it is shown that a particularly favorable thermal environment for snowflake growth and accumulation became collocated with the heavy snowband. It is hypothesized that the differences in the intensity and horizontal extent of the bands observed with these two events resulted from differing atmospheric responses associated with the areal extent of large-scale and frontogenetical forcing, moisture availability, degree of instability, and specific thermal profiles.

Description: The likelihood of simulated rainfall above a specified threshold being observed is evaluated as a function of the amounts predicted by two mesoscale models. Evaluations are performed for 20 warm-season mesoscale convective system events over the upper Midwest of the United States. Simulations were performed using 10-km versions of the National Centers for Environmental Prediction Eta Model and the Weather Research and Forecasting (WRF) model, with two different convective parameterizations tested in both models. It was found that, despite large differences in the biases of these different models and configurations, a robust relationship was present whereby a substantial increase in the likelihood of observed rainfall exceeding a specified threshold occurred in areas where the model runs forecast higher rainfall amounts. Rainfall was found to be less likely to occur in those areas where the models indicated no rainfall than it was elsewhere in the domain; it was more likely to occur in those regions where rainfall was predicted, especially where the predicted rainfall amounts were largest. The probability of rainfall relative-operating-characteristic and relative-operating-level curves showed that probabilistic forecasts determined from quantitative precipitation forecast values had skill comparable to the skill obtained using more traditional methods in which probabilities are based on the fraction of ensemble members experiencing rainfall. When the entire sample of cases was broken into training and test sets, the probability forecasts of the test sets evidenced good reliability. The relationship noted should provide some additional guidelines to operational forecasters. The results imply that the tested models are better able to indicate the regions where atmospheric processes are most favorable for convective rainfall (where the models generate enhanced amounts) than they are able to predict accurately the rainfall amounts that will be observed.

Description: The decision-making literature contains considerable information about how humans approach tasks involving uncertainty using heuristics. Although there is some reason to believe that weather forecasters are not identical in all respects to the typical subjects used in judgment and decision-making studies, there also is evidence that weather forecasters are not so different that the existing understanding of human cognition as it relates to making decisions is entirely inapplicable to weather forecasters. Accordingly, some aspects of cognition and decision making are reviewed and considered in terms of how they apply to human weather forecasters, including biases introduced by heuristics. Considerable insight into human forecasting could be gained by applying available studies of the cognitive psychology of decision making. What few studies exist that have used weather forecasters as subjects suggest that further work might well be productive in terms of helping to guide the improvement of weather forecasts by humans. It is concluded that a multidisciplinary approach, involving disciplines outside of meteorology, needs to be developed and supported if there is to be a future role for humans in forecasting the weather.

Description: A climatological and composite study of banded precipitation in the northeast United States during the cold season (October–April) is presented. Precipitation systems in the northeast United States in April 1995 and from October 1996 to April 2001 that exhibited greater than 25.4 mm (1 in.) of rainfall, or 12.7 mm (0.5 in.) liquid equivalent, were identified as cases for study. A total of 111 cases were identified during this period, of which 88 had available radar data. Of these cases, 75 exhibited banded structure whereas 13 did not. A band classification scheme was developed from a subset of study cases. Application of the classification scheme to the 88 cases revealed that banded cases can exhibit a variety of banded events during their evolution. Single-banded events were the most common (48), followed by transitory (40), narrow cold frontal (36), multi (29), and undefined (9). Further investigation of the single-banded events highlighted banded structure in the comma-head portion of storms, with 81% of these events exhibiting a majority of their length in the northwest quadrant of the surface cyclone. Composites were calculated for cases exhibiting single-banded events in the northwest quadrant of the surface cyclone and for nonbanded cases to distinguish synoptic and mesoscale flow regimes associated with banded events and nonbanded cases. The banded composite was marked by cyclogenesis and the development of a closed midlevel circulation. This flow configuration was associated with deformation and strong midlevel frontogenesis northwest of the surface cyclone center, which coincided with the mean band position. The nonbanded composite exhibited a much weaker cyclone located in the confluent entrance region of an upper-level jet. The absence of a closed midlevel circulation in the nonbanded composite limited deformation and associated frontogenesis northwest of the surface cyclone. Cross-section analysis through the respective composite frontogenesis maxima showed that the banded composite frontal zone exhibited stronger and deeper frontogenesis and weaker conditional stability than the nonbanded composite frontal zone. Case studies from the northeast United States confirm the composite results, highlighting the importance of deep-layer frontogenesis coincident with weak conditional stability. These results are in qualitative agreement with the Sawyer–Eliassen equation, which predicts that the frontogenetical response will be enhanced (reduced) in the presence of small (large) moist symmetric stability.

Description: In this study comparisons are made between Met Office mesoscale model boundary layer profiles, and radiosonde data collected in the central United Kingdom during three intensive boundary layer cloud experiments. Significant differences between forecast and observed profiles were found. An assessment of whether these differences are dominated by sonde random error or model error is performed. Results suggest that sonde random errors are insignificant for stratocumulus fields, but not for cumulus ones. Results show that model fields do not represent the fine vertical structure seen in the observations. The impact of this result for cloud prediction is discussed, and an estimate for the vertical resolution required to adequately parameterize the cloud field is provided.