水保所知识产出(1956---)知识库

小麦与玉米是黄土高原两大主要的粮食作物。黄土高原以旱作为主，作物用
水主要靠大气降水，但黄土高原地区年均降水量少，作物生长经常处于一种缺水
环境中。为了提高作物水分利用效率，增加粮食产量，需要制定和施行科学的农
田管理措施。科学管理制度的提出，一方面需要以大量的田间试验为基础，这些
试验往往比较费时，费力，而且试验成本较高。另一方面，与田间试验相结合，
作物模型通过模拟田间作物生长，可以再现田间试验的过程，其成本低，重复性
高，且节省时间。现有的大多数模型需要输入较多的参数，比较复杂，作物类型
与地理区域有一定限制，这在一定程度上增加了模型的校验难度，也限制了模型
的使用范围。世界粮农组织（FAO）发布的以水分为驱动力的 AquaCrop 模型需
要输入的参数少，模型精度高，使用简单，适用范围广，是一种较理想的作物模
型。本论文主要就黄土高原地区冬小麦与春玉米对 AquaCrop 模型进行了校验研
究，进而就作物产量与水分关系进行模拟分析，以期能为区域作物的田间管理提
供科学依据，并使 AquaCrop 模型的适用性得到延伸。主要研究结果如下：
(1). AquaCrop 模型可以较好地模拟黄土高原南部冬小麦产量、生物量、冠层
覆盖度及土壤含水量，其中冬小麦产量与冠层覆盖度的及模拟效果更好。均方根
误差（RMSE）及符合指数（d）分析表明，地上部生物量、冠层覆盖度、产量
及 1.8 米深土层的土壤含水量 RMSE 取值范围分别为 0.16-0.38 t/ha、1.87-4.15%、
0.50-1.44 t/ha、5.70-22.56 mm，d 取值范围地上部生物量、冠层覆盖度、土壤含
水量及产量依次为 0.22-0.89, 0.25-0.43, 0.36-0.62 和 0.95-0.98。(2). 在 AquaCrop 模型运行过程中，以 MATLAB 软件为平台，开发了冬小
麦冠层覆盖度计算程序，其结果与 Samplepoint 冠层分析结果的决定系数达到了
0.98。
(3). 以 Hsiao 的研究为基础，利用在黄土高原沟壑区的长武与丘陵沟壑区安
塞的春玉米田间试验结果，对 AquaCrop 模型进行了校正及验证分析。结果表明，
AquaCrop 模型对玉米冠层生长、土壤含水量的模拟，在生长衰老之前的结果较
好，之后模拟精度较差；模型对产量的模拟结果令人满意，WP*(标准化水分生
产率)取值 33；对于农田蒸散的模拟，时段较短时结果较差，整个生育期内农田
蒸散模拟结果总体好于短时段内的。对模型模拟结果进行均方根误差（RMSE）
及符合指数（d）分析，RMSE 值取值范围依次为 1.87-82.84mm、12.09-71.23mm、
0.12-1.03t/ha、1.22-8.3%。d 值取值范围依次为 0.69-0.99、0.68-0.84、0.59-0.97、
0.70-0.8。
(4). AquaCrop 模型可以较好地模拟冬小麦及春玉米的产量、冠层覆盖度以及
土壤含水量的变化，但对农田蒸散的模拟，除较短时段内模型模拟结果较差外，
对整个生育期农田蒸散的模拟，结果接近实测值。总体而言，AquaCrop 模型可
以较好模拟作物的生长，这也表明，模型调参比较准确，尤其是对水分生产率与
水分亏缺指数的确定，通过调参模型较好地表达了作物生长与水分消耗之间的关
系。
关键词：AquaCrop 模型；校正；验证；冬小麦；春玉米；黄土高原；影像分析

Wheat and corn are the two major crops at the Loess Plateau of China. In this
region, water requirements of crops are mainly met by rainfall. However, rainfall is
scarce in this area. In order to increase the water use efficiency and yields, scientific
field management is necessary. However, good field management schedules are
usually based on field experiments, which are expensive and laborious. On the
contrary, crop models proven may represent the actual growth process of crops
accurately, with the features of time-saving, highly repeatability, and low costs if
compared with field experiments. So far, major crop models are usually complicated
and need a lot of input parameters, and are not always feasible for diverse locations
and cultivars. These disadvantages have limited the range of model applications and
increased the difficulties in model calibration and validation. The water–driven
AquaCrop model, released by the Food and Agriculture Organization of the United
Nations (UN-FAO), requires less input parameters and is easy and robust to use. Thus,
the AquaCrop model is a practical model for usage. In this thesis, calibration and
validation of the AquaCrop model was conducted using data obtained from field
experiments with spring maize and winter wheat. Then, the relationship between crop
yields and soil water content on the Loess Plateau was analyzed in order to extend the
range of AquaCrop model applications and establish the principles for filed
management. Main results of this thesis are summarized as follows:
(1) AquaCrop model can simulate the yield, canopy cover, biomass, and soil
water content of winter wheat well at the Loess Plateau and the simulated results of yield and canopy were better than other output variables. The ranges of root mean
square errors (RMSE) were 0.16-0.38 t/ha, 1.87-4.15%, 0.50-1.44 t/ha, and 5.70-22.56
mm for biomass, canopy cover, yields, and soil water content at 1.8 m soil depth,
respectively. The index of agreement of Willmott (d) of biomass, canopy cover, soil
water content and yields were 0.22-0.89, 0.25-0.43, 0.36-0.62, and 0.95-0.98,
respectively. The results of statistic analyses confirmed that AquaCrop model is
applicable and the parameters chosen for the Loess Plateau were also accurate.
(2) A program of image analysis was designed with the MATLAB software for
analyzing canopy cover of winter wheat in the process of model operation. The
coefficient of determination (R 2 ) was 0.98 for the comparison between the results
analyzed with this program and the results analyzed with the Samplepoint software.
(3) The AquaCrop model was calibrated and validated with the experimental data
of spring maize obtained from the researches conducted at the gully region of
Changwu County and hilly and gully region of Ansai County at the Loess Plateau of
China. The results of model simulations demonstrated that the AquaCrop model can
simulate the expansion of canopy cover and soil water content well before the stage of
senescence and poorly after the senescence. For yields, the results of model
simulation were satisfactory when the parameter WP*（normalized biomass water
productivity）was 33. With respect of evapotranspiration (ET), the AquaCrop model
was not able to simulate the variations of ET accurately within short periods of time,
but it was able to simulate the variations of ET well for the whole growth season.
Using the RMSE and d analyzed the results of model simulation. The range of RMSE
was 1.87-82.84mm, 12.09-71.23mm, 0.12-1.03t/ha, 1.22-8.3% successively. And the
range of d was 0.69-0.99, 0.68-0.84, 0.59-0.97, 0.70-0.80 successively.  
(4) The canopy cover, yields and soil water content of winter wheat and spring
maize, which were simulated with AquaCrop model, agreed well with the
observations. But the model was not able to simulate accurately the ET values within
short time, but it can simulate the ET values well for the whole growth seasons of
maize. In general, the AquaCrop model can simulate the growth of crops and the
estimation of model parameters was accurate, especially for the parameters of WP* and p, at the Loess Plateau. Through simulations, the AquaCrop model accurately
elucidated the relationship between crop growth and water consumption.
Key words: AquaCrop model; calibration; validation; winter wheat; spring maize;
Loess Plateau; image analysis