Simplified Calculation And Forecasting Model Of ET₀in Qinghai Province

ET₀（Reference Corp evaportranspiration） is an assumption of cropevapotransipiration.,with appropriate coefficient, it can be used as the basis for irrigation. Itmeans a lot for requirement of crop water and the use of agricultural water, also can be usedas the behalf of the land evaporation in water circulation. Penman-Monteith equation,recommended by FAO, is verified to be the most closet calculation to actual evaporation inmany regions of the world. The only default is it has to use complete meteorological factors,the missing of factors can cause great derivation. In the area lack of information, the use ofP-M equation has its own difficulties. To solve this problem, we are trying to find out acalculate and prediction model which can easily get its information and easy to calculate aswell as guaranteed accuracy.Based on the information of56metorological stations of Qinghai province, we calculateET₀of each stations day by day, month by month and year by year from1958-2006withPenman-Monteith equation. By using it as the standard, we did the following work and gotthe results like this:（1） By using kriging method in GIS, the muti-year average daily ET₀were classified anddistribution characteristic of ET₀were analyzed. The result shows that the muti-year averageET₀verified from720mm to1099mm in Qinghai province. It decreased from northwest tosoutheast except a few stations showed difference because of elevation.（2） ET₀of the whole province were divided into6levels and each stations were selectedfrom the6levels. The picked out stations is Lenghu, Mangya, Dachaidan, Ledu, Zaduo,Mengyuan. Using this6stations as representative of the whole region, with wavelet analysisand wavelet variance, the maybe periodic cycle were analyzed. The result shows, there is nosame period of the whole station, each station has its own period. Mengyuan station’s35-yearperiod is very significant, the most intense cycle of Dachaidan is11-year, the first period oflenghu is27-year and the first period of Mangya is7-year, the only period that can passsignificant test in Zaduo is6-year and the first period in Ledu is49-year. By using correlationcoefficient and sensitive coefficient, the sensitive coefficient of each meteorological factorwere studied. The two correlation coefficient study shows the coefficient deceases like this: the maximum temperature, the per temperature, the minimum temperature, sunshine duration,wind speed and relative humidity. In a word, it deceases from temperature, sunshine duration,wind speed and relative humidity. The sensitive coefficient analysis shows the coefficientvaried like this: from December to next January, the absolute value of each meteorologicalfactor sensitivity coefficient decreased as follows: air temperature, sunshine hours, relativehumidity and wind speed, The temperature sensitivity coefficient and relative humiditysensitive coefficient are negative, sunshine duration and wind speed sensitive coefficient arepositive; From February to March, the absolute value of each meteorological factor sensitivitycoefficient decreased as follows: the sunshine duration, temperature, relative humidity, windspeed, the temperature sensitivity coefficient is still negative, but the absolute value is reduced,showing this negative impact on the ET₀becomes smaller. From April to October, thetemperature sensitivity coefficient turns from negative to positive, the weather factorssensitive coefficient absolute value decreased as follows: sunshine duration, air temperature,relative humidity and wind speed, to November, temperature sensitivity coefficient turnsnegative again, but in terms of absolute value, the most sunshine duration is still the biggest.（3） based on sensitive analysis,3emprical formulas were picked out, they areHargreaves formula based on temperature, Blanney-Criddle formula and Mass-transferformula. In the length of a surname, month and year, the application of each formula werestudied and the changes of error were also studied. It appears that the Hargreaves has theminimum error in each length, and the waveform also similar to the P-M standard. Thecalibration work is done and the study of changes of C value shows that the month’s Hformula’s coefficient C appears like a irregular “M”，the horns appears in March andAugust.The study of changes of C value also shows that the H formula of a surname’scoefficient C fluctuate rise from the1stto the21st, and peaked in21st, after24th,it fluctuatefalls.（4） Using BP neural network and RBF network to construct a simulation and predictionof ET₀network, the results show that, the precision of the model increases with the added ofinput factor （factor number limit for the number of factor P-M standard required）; when thenumber of factors are the same, the precision changes with the combination, the combinationof high sensitivity, selectivity factor can significantly improve the accuracy. Compared withBP neural network, RBF network has a better precision and the waveform is more similar toP-M standard.