气象要素时间分辨率对参考作物蒸散估算的影响

参考作物蒸散(reference evapotranspiration,ET0)的准确估算是农业水资源合理利用的重要环节。为了明确气象要素不同时间分辨率对参考作物蒸散估算的影响,该文基于寿县国家气候观象台2007-2013年观测资料,将1min时间分辨率数据平均值作为真实值,分析了10、20、30、40、60 min、4次/d(02:00、08:00、14:00、20:00)和3次/d(08:00、14:00、20:00)这7种不同时间分辨率对逐日气温、风速、太阳辐射、相对湿度和日、月及年参考作物蒸散(ET0)估算的误差情况。结果表明:ET0和气象要素的误差整体上随时间分辨率降低而增大。4个气象因子中,日平均风速估算受时间分辨率变化的影响最显著,误差最大;其次是太阳辐射。逐日ET0估算在7种时间分辨率的平均绝对相对误差(mean absolute relative error,MAPE)依次为0.53%、1.01%、1.38%、1.72%、2.46%、4.72%和6.14%,表明10至60min时间分辨率的估算效果相较3次/d和4次/d有明显改善。10至40 min的绝对误差超过95%都在-0.20~0.20 mm/d区间内,误差较小且集中度高;太阳辐射时间分辨率变化对ET0估算误差贡献最大,其次是风速,这主要是由于两个要素本身对分辨率较敏感且分别是ET0辐射项和动力项的主要组成因子。时间分辨率的变化对累计后长时间尺度ET0的影响较小,月和年ET0的误差明显小于逐日ET0,月ET0在7种时间分辨率的MAPE值依次为0.13%、0.21%、0.27%、0.40%、0.50%、1.18%和1.48%;各年ET0相对误差(relative error,PE)的绝对值多数均小于0.50%。

Effect of time resolution of meteorological variables on estimation of reference evapotranspiration

Abstract: The daily reference evapotranspiration (ET0) is generally demanded in the application of agriculture, meteorology, hydrology and other fields. The calculation of daily ET0 based on the data of different time resolutions is always a simplification because the change of ET0 is a continuous process in the time scale. It is valuable to discuss the influence of the simplification with different time resolutions on the accuracy of ET0 estimation. In this paper, based on the observed data of Shouxian National Climate Observatory from 2007 to 2013, the daily average values were calculated using the data with the resolution of 1 minute as the true values. The effects of the 7 different time resolutions (including 10 min, 20 min, 30 min, 40 min, 60 min, 4 times per day and 3 times per day) on the estimation of daily air temperature, wind speed, solar radiation, relative humidity and daily, monthly, yearly reference evapotranspiration (ET0) were analyzed by the error comparison. Results showed that the absolute values of the errors for ET0 and 4 climatic variables increased with the lower time resolution. Wind speed was most sensitive to the change in the time resolution, followed by solar radiation. The mean absolute relative errors (MAPE) of wind speed were 1.35%, 2.20%, 2.79%, 3.54%, 4.48%, 16.01% and 24.29% for the time resolution of 10 min, 20 min, 30 min, 40 min, 60 min, 4 times per day and 3 times per day, respectively. The change in the time resolution showed less influence on daily air temperature and relative humidity than the other 2 factors. The MAPE values of daily ET0 were 0.53%, 1.01%, 1.38%, 1.72%, 2.46%, 4.72% and 6.14% respectively for the 7 time resolutions, indicating that the accuracies of ET0 estimation for 3 and 4 times per day were significantly lower than the other 5 time resolutions. Over 95% of the absolute errors for daily ET0 with the time resolution from 10 to 40 min were in the range of -0.20-0.20 mm/d. These errors were so small and concentrated that the meteorological data with these 4 time resolutions were suitable for daily ET0 estimation. The mean bias errors (MBE) were almost equal to 0 for the time resolutions from 10 to 60 min, and the total deviation degree was very low. The MBE value for 4 times per day was -0.01 and the estimated ET0 was smaller than the true value, while the MBE value for 3 times per day was 0.02 and the estimated ET0 was larger. The change in the time resolution of the solar radiation led to the largest attribution to the error of estimated ET0, followed by wind speed, because these 2 climatic factors were more sensitive to the change in the time resolution and they were the key factors to the radiation item and dynamic item of ET0 respectively. The absolute values of the errors of monthly and yearly ET0 were significantly less than that of daily ET0. The MAPE values of monthly ET0 were 0.13%, 0.21%, 0.27%, 0.40%, 0.50%, 1.18% and 1.48% respectively for the 7 time resolutions. The absolute values of the relative error of yearly ET0 were mostly less than 0.50%. Our study demonstrated that the change in the time resolution of meteorological variables showed less impact on the estimation of the long-term integrated ET0 and it was important to improve observation time resolution for increasing the ET0 estimation accuracy.