Esperanza O. Cayanan, Bernardo M. Soriano, Jr.
Short range objective techniques of forecasting rainfall probability and visibility at three airport stations in the Philippines (Cagayan de Oro City, Roxas City and Dumaguete City) were developed using stepwise regression analysis. A 15-year data set (1978-1992) was used as developmental data and a three-year data set (1993-1995) was utilized for forecast skill verification. Both rainfall probability and visibility forecast equations for January, February and December (northeast monsoon months) for Cagayan de Oro City Airport can be used operationally at a significant level of accuracy. The forecast equations developed for Roxas City Airport gave insignificant results for most of the months. For Dumaguete City Airport, rainfall probability equations for April and November and visibility forecast equations for March and April attained good results.
Vicente B. Malano, Bernardo M. Soriano, Jr., Marino L. Mendoza
A numerical model is developed for storm surge prediction incorporating overland surges over the Gulf of Leyte. The model is formulated using the depth-integrated form of the transport equations applicable for storm surges. Movement of land/sea boundary is one dimensional. The forcing function of the model consists of surface wind generated by a symmetric cyclone model of Jelesnianski with inflow angle of 15 degrees. The model domain is divided into 30 x 41 grids with 1 km spacing. At each grid point, the maximum water height is computed during the 24-hour model time.
Results reveal that the maps of mean envelope of water can be used to predict storm surge over the Gulf of Leyte. With the aid of these maps, improvement of the model can be achieved by incorporating more theoretical typhoons with different characteristics, e.g. radius of maximum winds and typhoon intensity and movement. A similar study can be replicated in other surge-prone areas of the country.
Shirley J. David, Gaspar B. Salaguste
The study relates the behavior of rainfall intensity and water level during the passage of the tropical cyclone. The observation of 3-hourly rainfall data from the Science Garden Synoptic Station and the water level from the San Juan River Basin in Quezon City is considered from the time the tropical cyclone entered the Philippine Area of Responsibility (PAR) in order to have a longer series available for the estimation procedures.
The study made use of the regression technique, the Linear Regression with Autocorrelated Errors, to account for the serial correlation that is present in the data. The performance of the developed model is accounted for by using the Mean Absolute Percentage Error (MAPE).
Contrary to most hydrological literatures, rainfall intensity in this particular study was found to be an insignificant parameter to explain the water level in a river. This was attributed to the possibility of non-representativeness of the rainfall data considered in the development of the model. The results showed that lags 1 and 6 are the significant time that have bearing on the consequent behavior of water level.
It is suggested that more cases should be studied to come up with more conclusive results.
Nathaniel T. Servando
The behavior of two tropical cyclones (TC) due to mutual interaction are investigated through observational analysis and a series of numerical experiments using a non-divergent barotropic model. The basis of the observational analysis is the 45-year northwestern Pacific (NWP) TC data from 1952-1966. From these records, the so-called binary TC are picked based on the following criteria: TC which co-existed for at least 48 hours with separation distance of not more than 13 degrees lat. (1400 km) and attained at least storm intensity. Results indicate that the total number of binary TC reached 91 out of the 1,200 recorded number of storms or typhoons during the period or a percentage of 7.6 with an average of 2.02 cases per year. The orientation of the binary TC at the initial stage of interaction are generally E-W and NE-SW. Furthermore, decreasing separation distance with time was noted in about 50% of the total binary TC. Only four cases (4.4%) of the approaching binary TC were observed to have merged into one single TC. Results from the numerical experiments indicated that the relative motion of the cyclonic vortices are sensitive to the separation of the binary vortices. There is a critical separation distance at which a distinct bifurcation between merging and non-merging occurs. Below the critical separation distance merger takes place and above which, repulsion occurs. The critical separation distance varies with size of the binary vortex system and in particular, not affected by intensity changes if the initial separation distance is below 700 km. At larger separation distance, variations in intensity alter the critical separation distance, with the weaker binary systems merging at relatively higher separation distance than the stronger binary systems. On the interaction of two unequal vortices in terms of intensity, the stronger vortex dominates. The weaker vortex is destroyed by the strong tangential wind of the stronger vortex as they move closer. The presence of shear in the environmental flow has significant impact on the relative motion and attraction of the pair vortex. A cyclonic shear enhances the relative cyclonic rotation and favors merger. An anticyclonic shear reduces the relative cyclonic rotation and inhibits merger.
Felicidad V. Villareal
A model for simulating the radial distribution of tropical cyclone rainfall is developed. The model is primitive equation model in cylindrical coordinates with axial symmetry; a sigma coordinate is used. It incorporates microphysical cloud processes for explicit prediction of cloud water and rain water; these processes are evaluated by using the formulation of Kessler (1969). The sensitivity of the rainfall predictions of the model to the microphysical parameters involving autoconversion and accretion is examined. It is found that the model is fairly sensitive to the collection efficiency. On the other hand, the model is relatively insensitive to the auto conversion threshold value and autoconversion factor. The sensitivity of the model to the formulation of the drag coefficient is also examined. It is found that the model is quite sensitive to values of the drag coefficient. A constant value of drag coefficient equal to 0.003 is found to be a good alternative to using a variable drag coefficient, especially for stronger intensities of tropical cyclones (typhoon). On the basis of the results of the above-mentioned sensitivity tests, together with observational verification, the optimum values of the model parameters are determined. After determining these, the model is utilized to study the dependence of the rainfall distribution on the intensity and size of the cyclonic circulation. This is done by conducting a series of numerical simulations or numerical experiments with the model. Various sizes and intensities of tropical cyclone are specified by prescribing the appropriate value of horizontal diffusion coefficient. A total of fifteen (15) experiments have been made. The experiments show that, in general, the radial profile of rainfall rate is similar in shape to the corresponding radial profile of wind speed. In all intensity of tropical cyclones, the rainfall maximum is always located slightly closer to the cyclone center relative to the location of the wind maximum. Expressions which describe the radial profile of the simulated rainfall rate are formulated. These are based on equations which describe the radial wind profile of tropical cyclones. The equations has been used by Holland (1980). The two important parameters which are incorporated in the radial rainfall rate profile are the maximum rainfall rate and the radius where the maximum rate is located. If these two parameters are known, then one can determine the rainfall rate at any radius. An attempt is done to develop empirical equations or regression equations for determining these two parameters in terms of the wind profile. One regression equation specifies the maximum rainfall rate in terms of the maximum wind. The other regression equation specifies the radius of maximum rainfall rate in terms of radius of maximum wind. One can use these expressions for predicting the rainfall rate at any distance from the cyclone center from a forecast of the intensity and the size of the cyclone.
Bonifacio G. Pajuelas
The long period variations of rainfall in the Philippines are studied. This is done by first classifying the rainfall stations into climate types and then constructing time series of rainfall anomalies called rainfall anomaly indices (RAI) for each of the defined climate type. The RAI is the composite of standardized rainfall anomaly for a group of stations representing a region of quasi-homogeneous climate. An F-test suggests that the geographical variations of standardized rainfall anomaly is insignificant as compared to the temporal variations within each climate type, which mean that the RAI's are representative of each climatic region and can be used with confidence. In order to determine the longer period variations of the RAI time series, the shorter period variations are eliminated by very low frequency filter. Firstly, the rainfall stations were classified into climate types based on interstation correlations, the time averaged monthly standard deviations (SD), and maximum rainfall. five climate types emerged: the first two which is opposite to each other describe the dominant rain period and the remaining three types are intermediate between the first two. These intermediate types are further delineated into : (a) SD range < 100 (b) SD range from 100 to 147 (c) and SD range > 147. The minimum correlation coefficient within each climate type is 0.75. Secondly, the standardized rainfall anomaly (SRA) for every station in each climate type are compared, and the variance structure of the geographical and temporal variations of the standardized anomaly were computed to determine the significance of the RAI. In each 4-month period, plots of individual stations SRA generally respond to the prevailing climatic condition that influence the region. There are some inconsistencies in the sensitivity of the individual stations' SRA as shown in the time series of graphs. However, the geographical variations of these SRA's are statistically insignificant compared with the temporal variations, which imply high confidence on these RAI's. Thirdly, RAI graphs and statistical characteristics of each time series are analyzed. The time series graphs of RAI reveal the wet and dry periods in each climatic region which suggest the dominant phase of the Southern Oscillation. Chi-square test suggests that all unfiltered RAI samples do not significantly differ from the normal (Gaussian) distribution, except in western Luzon where samples have skewness and kurtosis which differ significantly from normal. The Von Neumann ratio showed that non-random elements are significantly present in all RAI samples. However, Mann-Kendall rank statistics indicated these trends are not significant in the unfiltered RAI series. On the other hand, 10-year moving average (10Y-MA) filtered RAI series trends are significant particularly over eastern Philippines, Mindanao region, and Central Visayas during the 4-month period of February, March, April and May (FMAM). The trend is further accentuated by the 30Y-MA filtered RAI's. Student's t-test suggested a significant decrease in the mean unfiltered RAI values of FMAM 1973-1996 series of type 3 climate. In the 10Y-MA filtred RAI, climate types 2, 4 and 5 display also significant change in the mean RAI values of FMAM 1973-1996.
Nathaniel A. Cruz
Early, normal and late onset of rainfall associated with the southwest monsoon are analyzed by compositing pentad (five-day) values of selected meteorological parameters. Results showed that the early onset of rainy season (May 2, 1984) was primarily due to the passage of a tail of a cold front over the northern part of the country. Although, there was an early start of the rains, the rainy condition was not sustained due to weak southwest flow. On the case of the normal start (May 28, 1988), the onset can be clearly attributed to the prevalence of the southwest monsoon. Associated with this condition is the normal northward migration of the Intertropical Convergence Zone as well as the northward movement of the ridge of the North Pacific high pressure area. One distinct characteristic of this type of onset is the continuous rainfall over the western part of the country for several pentads after the onset. The persistence of the ridge of the NPHPA over the country was the main reason for the late onset of the rains (June 23, 1993). Analysis showed that the unseasonal position of the ridge has a significance influence in the movement of the ITCZ as well as the monsoon trough. Rainfall conditions, like the case of the early start, experienced a significant decrease after the onset. Significant wind shift (associated with the start of the southwest monsoon) is likewise related to rainfall onset. Investigation showed that a significant change in wind direction (from easterlies to southwesterlies) at selected synoptic stations occurs before the onset of the rains except on instances when the onset was brought about by other systems like cold fronts. An index was developed to assist in the forecasting of monsoon and possibly rainfall onset in the future. Called the Brunei-Hongkong index, the average pressure gradient between the two stations several pentads before the onset, could serve as an indicator of possible start of the southwest monsoon as well as the rainy season.