Collection and processing system of hydrological data during flash floods on the River Yodo

Transcription

1 Collection and processing system of hydrological data during flash floods on the River Yodo Tojiro Ishihara, Yutaka Inada and Masao Hayashi Abstract. The data collection and processing system required to forecast and to issue warnings of flash floods on the Yodo River is now in operation. This paper describes some interesting and important items in the practical system which involves 40 raingauges and 49 flow stations. These include transformation of measured data to electric signals and their transmission by radio, the sampling interval for collection and processing of data, methods of checking and interpolating transmitted data, and methods of storing and computing. Système de collecte et de traitement des données hydrologiques des crues brutales du fleuve Yodo Résumé. Un système de collecte et de traitement des données hydrologiques nécessaires à la prevision et l'annonce des crues brutales du fleuve Yodo (Japon) est actuellement en service. La présente communication expose quelques points importants et intéressants de ce dispositif qui comporte 40 stations pluviometriques ainsi que 49 stations limnimetriques. Il s'agit de la transformation des données d'observation en signaux électriques, de la source d'énergie, de l'émission radio des signaux, de l'intervalle de temps pris en compte pour le rassemblement des données, et des procèdes de contrôle des données reçues, d'interpolation éventuelle pour combler les lacunes, et du traitement des données pour le stockage et le calcul. GENERAL DESCRIPTION The River Yodo which drains an area of 7300 km 2 runs through the cities of Kyoto and Osaka, draining into Osaka Bay as shown in Fig.l. Two thirds of the river basin is occupied by mountains. Typhoons cause heavy concentrated rainfall so that the river basin often experiences flash floods. Typical of the flash floods in recent years is the one caused by Typhoon 5 in 959. The hydrograph of this flood is shown in Fig.2. The time lag between the peak of the average rainfall over the river basin and that of the flood in the lower reach of the river is only 9 h. So far, five dams have been constructed in this river basin in order to control such flash floods. The Yodo River Dams Control Office was established in 969 to control these dams in an integrated manner. An automatic system of hydrological'data collection and processing is now operated by this office. As shown in Fig.3 this system consists of telemetering equipment, a terminal data controller (TDC) and an electronic computer. At present 89 items of hydrological data of which 40 are rainfall data and the remainder water stage data are collected and processed automatically every hour. TRANSFORMATION OF DATA TO ELECTRIC SIGNALS Figure 4 shows a block diagram of the telemetering equipment. The drum driven by the float movement in a water stage recorder or by a tipping bucket raingauge is provided with a binary contact point and also with an A D converter which outputs the water stage in centimetres or the rainfall in millimetres by means of binary coded electric signals. Each digit in the measured value is converted into binary code and a parity bit is added to the end of each binary unit for error checking purposes; for example,

2 4 Tojiro Ishihara, Yutaka Inada and Masao Hayashi 0,0 20 km L BORDER OF SUB-BASIN DAM CONSTRUCTED RAINFALL OBSERVATION STATION, \y / Y TAKAYAMANIAM J / U < ( \ ( MUE JAM ISS^'/" SHO&ENJI DAM ^ > \. FIGURE. Yodo River basin. HlRAKATA OBSERVATION STATION.V-^^^7,,,,!,,,,!,,,!,,,,,,,,, SEPT,, SEPT, FIGURE 2. Hydrograph of flood caused by Typhoon 5, 959. Discharge in cubic metres per second and precipitation in millimetres per hour.

3 Collection and processing system of hydrological data 5 TELEMETER CONTROL BOARD (TELEMETER SYSTEM) T D C (ARRANGEMENT OF DATA) CONTROL BOARD (COMPUTER UNIT) ZZ3 COMPUTER UNIT FACOM MAGNETIC DRUM MEMORY PUNCH OUT (PAPER TAPE) SYSTEM PUNCH OUT (PAPER TAPE) GRAPHIC PANEL (INDICATION OF DATA) OFF LINE.J FIGURE 3. Collection and processing system of hydrological data on the River Yodo (decimal system) (binary coded decimal system with odd number parity). TRANSMISSION OF DATA AND ELECTRIC POWER SUPPLY AT A REMOTE STATION The electric signal converted to binary code by the A D converter are transmitted by a 2 khz band subcarrier called an FS system in telemetering language. In that system and 0 are expressed by pulses of 20 milli-second and 40 milli-second respectively. The subcarrier frequency is modulated once again into a 70 MHz band microwave before transmission. The microwave is so used because of the possible sharing of the microwave network which is already being operated by the Ministry of Construction for emergency communications. A transmitted signal consists of a 7 digit number; 2 digits for the station code, 4 for the observation value and the remaining digit is used for monitoring the condition of the power source at the observing station. The telemetering equipment at an observing station begins to work only when it receives a special call signal. An instrument called a TDC interrogates automatically all the observation stations at regular intervals. All the equipment is transistorized, offering the advantages of reliability in operation, ease of maintenance and small demand from the (DC 2 V X 0.02 A) power source. ANTENNV^-" w T D C IN CONTROL OFFICE r RAIN RAIIMAKGE "* I" "J + 20 cm >CA TlPPING-BUCKET SOLAR BATTERY OR AC,COMMERCIAL WATER LEVEL FLOAT 30cm FIGURE 4. Block diagram of the observation station.

4 6 Tojiro Ishihara, Yutaka Inada and Masao Hayashi An alkali battery supplies the electric power and it is recharged by commercial electricity via a rectifier. However, at a station where commercial power is not available, such as a rainfall station in a mountainous area, the alkali battery is recharged from a solar battery which is installed on the roof of the observation station at right angles to the average direction of the sunbeam. With a receiving area of 30 cm X 40 cm its maximum power production is DC 2 V X 500 ma. A DC 2 V alkali battery of 40 A Xhour capacity can be used without recharging for 90 days despite long spells of rainy or cloudy weather. Moreover, if the capacity of the alkali battery falls below a certain level, an alarm is given by the telemetering equipment via the 7th digit. At present three rainfall observation stations have solar batteries for charging alkali batteries and all of them are working well. TIME INTERVAL FOR DATA COLLECTION AND PROCESSING For flash floods the choice of time interval for hydrological data collection and processing must take into account the time needed for the data collection, the capacity of data processing system and the usefulness of the data. For the River Yodo, forecasts and warning of flash floods are required every hour and therefore the sampling interval is one hour. The following division of time is allowed and the capacity of each piece of equipment is designed so that all parts finish work within a fixed time of one another. 0 min: time needed for the collection of hydrological data by telemeter 5 min: time needed for data processing by computer 30 min: time needed for flood runoff calculation by computer 5 min: time needed for judging the calculated result and giving revised operating instructions to dams Moreover, one hour is a convenient time interval for the calculation of rainfall intensity. Collection and processing of data at time intervals of 5 or 30 min is possible, but these data cannot be used directly for the flood runoff calculation. CHECKING TRANSMITTED DATA AND INTERPOLATING MISSING OR UNTRANSMITTED DATA Data transmitted by telemeter are checked by the TDC both on the correctness of the electric signal and on the parity information. If something is wrong with the data the TDC orders the telemetering station to repeat the transmission. If after three transmissions the data are still wrong the observation is regarded as missing and the TDC records a missing observation signal. The number of re-transmissions of a single item is decided in the light of the necessity of collecting data in a fixed time of 0 min. After finishing the data collection, the TDC adds clerical data such as the time and date and transmits them to the computer. The following processing is then performed by the computer: () transformation of measured rainfall values to accumulated rainfall since some reference time; (2) transformation of observed water stage values to those from a reference height; (3) transformation of water stage to discharge and calculation of storage capacity of dams; (5) searching for alarming values of rainfall, water level or discharge. Errors and missing observations are sometimes included in the data. In collecting and processing the data automatically as well as using them for the flood runoff calculation, the treatment of errors and missing observations is very important. Attention is

5 Collection and processing system of hydrological data 7 paid to the following points. Rainfall over 20 mm/h is very improbable and is regarded as an error. The maximum and minimum water level are stored and if the observed water stage value is outside the range it is regarded as an error. However, no restriction is imposed on the rate of rise of the water stage because rapid change in water level is possible due to dam operation. The flood runoff calculation of the Yodo River basin is based on the storage function methods, and the whole basin is divided into 20 sub-basins. Therefore, missing rainfall observations at each station must be accurately estimated at each observation time. The estimate is based on the regression equation with that station which shows the highest coefficient of correlation with the station concerned. The missing observation is estimated by the regression equation Y = ax+b in which Y is the interpolated value of rainfall intensity in millimetres per hour, X the observed value at the reference observation station in millimetres per hour and a and b the numerical constants obtained from statistical analysis of the past data as shown in Fig.5.,0 ÏÏ c > >- i c < 3 > t > AND 5 ^ 4 AND 6 A 7 AND 8 ' i H < C ( > «P! C > i i c ) t ' < < * < I i C ) 7 AND 8» AND 2 a 9 AND 0 AcOMJLATICN TIME c A o o A A AND 2 3 AND 4 A 5 AND 6 2 H H FIGURE 5. Several examples of correlation coefficients between two points rainfall accumulated during, 2, 3, 6, 9 and 2 h. (Numerals in the figure represent observation stations given in Fig.l.). As there is no suitable method for interpolating a missing water stage observation, the value from the previous hour is used. The observation stations are regularly inspected once a month to check for serious damage. At present the efficiency of the data collection system is 99.9 per cent at normal times. STORAGE AND COMPUTING Data processed by the computer are stored on a magnetic drum of M bits capacity. As the quantity of data received per hour is IK bits, it is possible to store 40 days' data. When the remaining capacity of the magnetic drum is small, the data are transferred to a magnetic tape for permanent storage. In days of flood, the time and date are keyed in manually to the computer by typewriter, and the data after that time are stored on a 5M bits magnetic disc pack as well as on the magnetic drum used for the flood runoff calculation. The results of the flood runoff calculations and some important numerical values in the calculation are also stored in that magnetic disc pack. As mentioned above, by storing the special flood data separately from the ordinary hydrological data, continuity of hydrological data collection is assured and the storage

6 8 Tojiro Ishihara, Yutaka Inada and Masao Hayashi capacity of the electronic computer can be used effectively. As the computer is of a common type, it is used for calculations other than the flood runoff calculation. However, the computer is designed so that the hydrological data collection and processing are given top priority even if in the midst of the other calculations. CONCLUSION The hydrological data collection and processing system for the River Yodo is made to work automatically as far as possible. Processed data can be used directly for the flood runoff calculation. Therefore, it is possible to forecast a flash flood and give an alarm sufficiently early. Such a system is a good example for other river basins suffering from flash floods.