黃健熙,賈世靈,馬鴻元,侯英雨,何 亮
(1. 中國農業大學信息與電氣工程學院,北京 100083;2. 國家氣象中心,北京 100081)
基于WOFOST模型的中國主產區冬小麥生長過程動態模擬
黃健熙1,賈世靈1,馬鴻元1,侯英雨2,何 亮2
(1. 中國農業大學信息與電氣工程學院,北京 100083;2. 國家氣象中心,北京 100081)
大區域尺度WOFOST(world food studies)模型的動態模擬是作物模型區域應用的重要基礎。該文以中國冬小麥主產區為研究對象,利用中國冬小麥主產區內 174個農業氣象站多年觀測數據以及氣象站點觀測數據,重點優化WOFOST模型中與品種相關的積溫參數,即出苗至開花有效積溫與開花至成熟有效積溫。在冬小麥主產區分區的基礎上,以2012—2015年氣象數據驅動WOFOST模型,在站點尺度進行冬小麥的物候期、葉面積指數(leaf area index,LAI)和單產動態模擬和精度分析。結果表明:WOFOST模型模擬出苗至開花天數的決定系數R2為0.89~0.94,均方根誤差RMSE為7.87~11.52 d,模型模擬開花至成熟天數的R2為0.63~0.77,RMSE為2.99~4.65 d; 模型模擬LAI的R2為0.70~0.83,RMSE為0.89~1.46 m2/m2;灌溉區WOFOST模擬的單產精度R2為0.45~0.59,RMSE為734~1 421 kg/hm2;雨養區WOFOST模擬的單產精度R2為0.48~0.61,RMSE為1 046~1 329 kg/hm2。結果表明,WOFOST模型在全國尺度取得了較高模擬精度,為區域尺度作物模型的農業應用提供了堅實的過程模型基礎。
模型;優化;溫度;WOFOST;冬小麥;參數標定;物候期;動態模擬
黃健熙,賈世靈,馬鴻元,侯英雨,何 亮. 基于 WOFOST模型的中國主產區冬小麥生長過程動態模擬[J]. 農業工程學報,2017,33(10):222-228. doi:10.11975/j.issn.1002-6819.2017.10.029 http://www.tcsae.org
Huang Jianxi, Jia Shiling, Ma Hongyuan, Hou Yingyu, He Liang. Dynamic simulation of growth process of winter wheat in main production areas of China based on WOFOST model[J]. Transactions of the Chinese Society of Agricultural Engineering(Transactions of the CSAE), 2017, 33(10): 222-228. (in Chinese with English abstract)
doi:10.11975/j.issn.1002-6819.2017.10.029 http://www.tcsae.org
冬小麥是中國的 3大糧食作物之一,主要分布在長城以南,長江以北,其種植面積占全國耕地總面積約18%~27%,糧食作物總面積的18%~24%。基于作物生長模型的方法是開展長勢監測與產量估測的重要技術手段之一。其區域尺度的應用極大依賴于作物模型的標定精度。
本文選擇WOFOST(world food studies)模型作為作物生長動態過程模型。WOFOST作物模型由世界糧食研究中心和瓦赫寧根農業大學共同研發,能夠以天為步長定量模擬氣象和其他環境因子影響下的作物生長過程[1]。在過去的幾十年里,WOFOST模型已經在諸多國家和地區的多個領域得到廣泛應用。例如產量風險分析、年際間產量變化分析、土壤狀況對產量的影響、氣象條件對產量的影響、不同作物品種與耕作制度對產量的影響、氣象條件對產量的影響等。且模型對過程的描述是通用的,可以通過改變參數模擬不同的地理位置上的不同作物,因此對作物模型的參數進行標定校準,使其適應于當地的指定作物,是進行模型區域應用的重要前提。WOFOST能夠模擬潛在生長、水分脅迫和養分脅迫三種水平[1]。
國內外學者在作物模型區域標定和模型應用做了許多研究與探索。Wit等[2]利用集合卡爾曼濾波方法將遙感的土壤水分估測值同化到WOFOST模型中,糾正該模型土壤水平衡誤差,對西班牙、法國、意大利和德國1992?2000年間的冬小麥和玉米進行產量估測,結果表明,數據同化明顯改善了 66%地區的冬小麥產量模擬和56%地區的玉米產量模擬。Ma等[3]在華北平原冬小麥標定與區域化WOFOST模型,通過葉面積指(leaf area index,LAI)耦合SAIL-PROSPECT模型來模擬土壤調節植被指數(soil adjust vegetation index, SAVI),最小化模擬與合成的SAVI之間的差異,重新初始化出苗日期,結果表明,該方法在將模擬應用到區域尺度方面具有應用潛力。Boogaard等[4]采用WOFOST模型評估歐盟25個國家的秋播小麥產量差距的優勢與限制,結果表明,WOFOST水分脅迫模式下估算秋播小麥的產量精度較高。張建平等[5-10]利用WOFOST模型分析了氣候變化與低溫冷害對東北地區春玉米產量的影響。張素青等[11]在河南省夏玉米主產區對WOFOST模型進行了校準與驗證。孫琳麗等
[12]在內蒙古河套灌區玉米種植區對WOFOST模型進行了適應性分析。陳思寧等[13]分析了PyWOFOST模型在東北玉米種植區的適應性。張建平等[14]選擇華北平原冬小麥為研究對象,在WOFOST模型區域適應性分析與檢驗的基礎上,利用區域化的WOFOST模型模擬結果實現干旱災害對作物影響的定量分析與動態評估。綜上所述,目前作物模型標定與驗證方面的工作主要集中于單個站點和若干站點的尺度上。尚未見中國主產區尺度WOFOST作物模型標定和適應性研究報道,其主要挑戰在于中國冬小麥主產區的WOFOST模型輸入參數和初始狀態的空間變異性。
本文基于中國冬小麥主產區內的 174個農業氣象站點觀測數據,在站點尺度,評估WOFOST模型生育期、葉面積指數和單產動態模擬與精度。評價WOFOST模型在全國冬小麥主產區的動態模擬的適應性。
1.1 研究區與數據
冬小麥主要分布在長城以南,長江以北,本文的研究區選擇中國主要冬小麥種植區,主要包括河北、山西、江蘇、安徽、山東、河南、湖北、重慶、四川、貴州、云南、陜西、甘肅、寧夏等省區。
研究區內共包括174個農業氣象站點,其中包括15個農業氣象試驗站點,簡稱試驗站,位置分布如圖 1所示。本研究收集了2011至2015年各站點的冬小麥生育期及單產的觀測數據。其中關鍵生育期用于積溫參數的計算,并對模擬生育期的驗證,單產數據用于對模擬結果精度的檢驗。此外,試驗站還提供不同生育期冬小麥根、莖、葉、貯存器官的干物質質量和實測LAI等生長率參數。

圖1 研究區冬小麥分區及農業氣象站點分布Fig.1 Winter wheat zone and spatial distribution of agro-meteorological stations in study area
WOFOST模型的氣象要素采用中科院青藏高原研究所生產的中國區域地面氣象數據集[15-16],主要包括 7個要素,近地面氣溫,地表氣壓,近地面空氣比濕,近地面全風速,向下短波輻射,向下長波輻射和降水率。空間分辨率為 0.1°,數據獲取網址為:“http://www. Tpedatabase.cn/portal/MetaDataInfo.jsp?MetaDataId=249369”。選擇2011—2015年的氣象數據,進行要素計算與格式轉換,獲得WOFOST模型所需6個氣象要素,包括逐日的輻射量、平均水汽壓、日最高溫、日最低溫、風速及降水。
1.2 WOFOST模型參數的校準方法
由于主產區內的冬小麥品種和種植模式存在差異,因此有必要將整個冬小麥主產區劃分為相對均質的子區域進行作物模型標定。選取了單產水平、土壤類型、氣象條件和種植結構為指標,采用空間聚類的方法,獲得冬小麥分區。
WOFOST模型中輸入參數包括氣象、作物、土壤和田間管理,參數較多,難以實現對每個參數的標定與校準。因此,需要根據WOFOST輸入參數的敏感性和物理含義進行分類標定與校準,對于不敏感的參數或低敏感性的參數,采用WOFOST模型默認值或通過文獻查閱確定;對于與品種有關,敏感性較高且空間變異性較大的參數,先通過觀測數據計算其取值范圍,再通過優化算法確定。
對于每個分區,通常還包含若干農業氣象站點。在站點尺度上,通過每個農氣站點記錄的生育期和鄰近氣象站點觀測的日平均溫度,計算出與品種相關的積溫參數,即出苗至開花有效積溫TSUM1與開花至成熟有效積溫TSUM2。同時,假設鄰近的2 a冬小麥種植的品種不發生變化,因此,把 2011—2014冬小麥生育期標定的TSUM1和TSUM2參數值,分別賦予到WOFOST模擬的2012—2015生育期。此外,WOFOST的出苗日期和重要土壤參數(田間持水量、凋萎系數和初始可利用水含量)是通過每個農業氣象站點觀測給定。對于每個試驗站點,比葉面積(specific leaf area,SLA)根據試驗站不同生育期的干物質質量和LAI計算。不同生育期的根、莖、葉、貯存器官的干物質量分配系數,也是通過觀測數據計算獲得。對于分區中缺失試驗站的情況,則采用最近鄰站點賦值的方法確定。其他參數值通過文獻查閱[17-27]確定或采用WOFOST默認值。表1為WOFOST模型中主要作物參數校準值。
1.3 模型參數的檢驗方法
模型模擬檢驗包括了 2個部分:散點圖比較以及選擇統計評價指標對模擬值和實測值進行定量評價。散點圖為出苗至開花天數、開花至成熟天數、單產及LAI模擬值與實測值的回歸分析圖。統計評價指標選擇了決定系數(R2)、一致性系數(D)、變異系數(coefficient of variation,CV)、均方根誤差(root mean square error,RMSE)、歸一化均方根誤差(normalized root mean square error,NRMSE)。其中R2和D反映了實測值與模擬值之間的一致性,越接近1表示模擬效果越好。CV反映了數值離散程度,值越大越能體現數據的空間變異性[28],可將其進行粗略分級:CV<10%,為弱變異性;10%≤CV≤100%;為中等變異性,CV>100%,為強變異性[29]。RMSE和 NRMSE反映了模擬值與實測值之間的相對誤差和絕對誤差[30],值越小表示模擬效果越好,其中,NRMSE≤10%,為極高精度;10%


式中i表示第i個樣本;Yi和Xi分別為第i個樣本模擬值和實測值;為全部樣本實測數據平均值;n為樣本數;SD為模擬結果的標準差,為全部樣本模擬結果的平均值。

表1 WOFOST模型中主要作物參數校準值范圍Table 1 Range of calibrated values of main crop parameters of WOFOST model
2.1 模型參數的校準結果
根據上述模型參數校準方法,進行模型的校準。表1為所有冬小麥分區的部分關鍵參數校準值范圍。
2.2 WOFOST模型檢驗
為驗證WOFOST模型在中國冬小麥主產區的動態模擬精度,在站點尺度,以2012—2015年當年的站點觀測出苗日期為模擬初始日期,以氣象、土壤、作物等參數驅動WOFOST模型進行冬小麥生長模擬,并對模型模擬的出苗至開花天數、開花至成熟的天數、LAI和單產進行模擬結果精度分析與驗證。
2.2.1 生育期驗證
開花期和成熟期分別是冬小麥營養生長和生殖生長階段的結束日期,是評價WOFOST模型模擬的重要生育期。該文分別選擇出苗期至開花期的天數和開花期至成熟期的天數來進行關鍵生育期的驗證。
2012—2015年,模型對生育期天數的模擬,具有較為相似的模擬精度。由表2可知,出苗至開花的R2在0.89以上,D在0.96以上,說明模擬值與實測值具有較好的一致性,NRMSE在7%以下,模擬誤差在7.87~11.52 d之間,表明WOFOST模型能準確模擬冬小麥開花期。開花至成熟天數的R2位于0.63與0.77之間,D在0.87~0.93之間,NRMSE在8%~12%之間,模擬誤差在2.99到4.65 d之間。不同熱量條件的地區,開花到成熟期的天數差異較大。模擬誤差主要依賴于開花期的誤差和開花到成熟期的積溫精度。以2012年為例(圖2),模型模擬出苗至開花天數、開花至成熟天數分別與實測值之間具有較好的相關性,各點均勻的分布在回歸線兩側。同時,出苗至開花天數與開花至成熟天數的 CV均在 10%以上,具有顯著的空間變異性,能充分解釋模擬冬小麥生育期的區域空間變異。

表2 不同年份WOFOST模擬生育期的驗證結果精度對比(2012—2015)Table 2 Comparison of simulated growth stages accuracies in different years (2012—2015)

圖2 WOFOST模型模擬出苗到開花期天數和開花到成熟期天數對比(2012)Fig.2 Comparison of simulated and measured days from emergence to anthesis and anthesis to maturity (2012)
2.2.2 LAI的驗證
由圖3可知,2012—2015年模擬LAI與實測值之間的R2在0.70~0.83之間,D在0.88~0.96之間,WOFOST模擬LAI值與實測值之間的一致性較好,RMSE在0.89~1.46 m2/m2之間,NRMSE在50%~63%之間。由敏感性分析結果可知,對LAI最大值敏感的參數主要有葉片最大CO2同化速率、SLATB、初始生物量和葉片衰老系數[19]。本研究中,TDWI和SPAN都采用了默認值,AMAXTB雖然根據參考文獻確定,但有一些冬小麥種植區缺少觀測數據,采取模型默認值,導致某些站點的WOFOST模擬誤差較大。

圖3 WOFOST模型模擬LAI值與實測值對比(2012—2015)Fig.3 Comparison of WOFOST simulated and field-measured LAI (2012—2015)
2.2.3 單產的驗證
考慮到冬小麥光熱和降雨條件的差異,將主產區劃分為灌溉區和雨養區分別進行WOFOST模擬,其中黃淮海設定灌溉區,采用WOFOST潛在模式,西北和西南地區設定為雨養區,采用WOFOST的水分脅迫模式。
灌溉區WOFOST模擬的單產精度R2為0.45~0.59,RMSE為734~1 421 kg/hm2;雨養區WOFOST模擬的單產精度R2為0.48~0.61,RMSE為1 046~1 329 kg/hm2。相比較而言,WOFOST模型模擬的灌溉區單產精度總體要高于雨養區,具有更低的RMSE值(表3)。

表3 灌溉區不同年份WOFOST模型模擬單產的驗證結果精度對比(2012—2015)Table 3 Comparison of simulated yield accuracies in multiple years in irrigation area (2012—2015)
對于某些年份單產偏低的可能原因是設定的品種相關的參數 TSUM1和 TSUM2等具有較大誤差,導致WOFOST模擬的生育期和產量誤差較大。由表3可知,灌溉區與雨養區的D位于0.73~0.99之間,說明模擬值與實測值具有很好的一致性。總體而言,而單產模擬的誤差主要在于,對單產敏感的參數難以獲得準確的空間分布值。

圖4 2012和2015年WOFOST模型模擬單產和實測單產相對誤差的空間分布圖Fig.4 Spatial distribution of relative error of WOFOST simulated and field-measured yield per unit in 2012 and 2015
從單產的空間分布差異來看(圖 4),模擬單產精度較高的站點主要分布于黃淮海灌溉區。而模擬單產相對誤差較大的點,主要集中在雨養區。可能原因是農業氣象站點分布稀疏和關鍵土壤參數難以準確標定。灌溉區模擬單產的 CV在 14%~22%之間,雨養區模擬單產的CV在25%~40%之間,能解釋模擬冬小麥單產的空間變異性,能解釋模擬冬小麥單產的空間變異性。
由于中國冬小麥主產區光熱條件和種植品種的差異,WOFOST模型模擬的生育期空間差異性較大,國家農氣站點記錄的觀測數據表明,主產區內冬小麥出苗到開花期歷時天數在89~200 d之間,開花到成熟期時天數在20~70 d之間,參數校準后的WOFOST模型能準確撲捉這一差異;WOFOST模型模擬LAI的誤差主要來源于對LAI敏感的WOFOST模型輸入參數(例如:TDWI和SPAN)未考慮參數的空間變異性;WOFOST單產模擬方面,我們考慮了雨養區和灌溉區的差異,其他影響因素未予考慮,例如:營養和病蟲害脅迫等。研究表明,合理的標定WOFOST模型的順序為生育期,LAI和單產。今后的研究將通過衛星遙感數據和作物模型數據同化獲得WOFOST的關鍵輸入參數的空間分布的優化值,從而進一步提高大區域作物模型的模擬能力。標定后的WOFOST模型將為區域尺度的溫度脅迫或水分脅迫對產量的影響提供動態過程模型。
本文以WOFOST為動態生長模型,中國冬小麥主產區為研究對象,在分區的基礎上,基于農業氣象站點觀測數據標定WOFOST模型的敏感參數,在站點尺度,動態模擬生育期、LAI和單產。驗證結果表明,模型模擬出苗-開花天數的NRMSE在4%~7%之間,模型模擬開花-成熟天數的NRMSE在8%~12%之間,具有較高的模擬精度,CV在14%~20%之間,具有空間變異性。模型模擬的LAI的NRMSE在50%~63%之間。模型模擬單產的NRMSE在11%~28%之間,CV在14%~40%之間,能較好地體現單產的空間差異性。總體來說,WOFOST模型取得了較為理想的模擬精度,具有較好的適應性。
[1]Diepen C A, Wolf J, Keulen H, et al. WOFOST: A simulation model of crop production[J]. Soil Use &Management, 1989, 5(1): 16-24.
[2]Wit A J W De, Diepen C A Van. Crop model data assimilation with the Ensemble Kalman filter for improving regional crop yield forecasts[J]. Agricultural & Forest Meteorology, 2007, 146(1-2): 38-56.
[3]Ma Yuping, Wang Shili, Zhang Li, et al. Monitoring winter wheat growth in North China by combining a crop model and remote sensing data[J]. International Journal of Applied Earth Observation & Geoinformation, 2008, 10(4): 426-437.
[4]Boogaard Hendrik, Wolf Joost, Supit Iwan, et al. A regional implementation of WOFOST for calculating yield gaps of autumn-sown wheat across the European Union[J]. Field Crops Research, 2013, 143(143): 130-142.
[5]張建平,王春乙,楊曉光,等. 溫度導致的我國東北三省玉米產量波動模擬[J]. 生態學報,2009,29(10):5516-5522.Zhang Jianping, Wang Chunyi, Yang Xiaoguang, et al.Simulation of yield fluctustion caused by the temperature in Northeast China[J]. Acta Ecologica Sinica, 2009, 29 (10):5516-5522. (in Chinese with English abstract)
[6]張建平,王春乙,趙艷霞,等. 基于作物模型的低溫冷害對我國東北三省玉米產量影響評估[J]. 生態學報,2012,32 (13):4132-4138.Zhang Jianping, Wang Chunyi, Zhao Yanxia, et al. Impact evaluation of low temperature to yields of maize in Northeast China based on crop growth model[J]. Acta Ecologica Sinica,2012, 32(13): 4132-4138. (in Chinese with English abstract)
[7]陳振林,張建平,王春乙,等. 應用 WOFOST模型模擬低溫與干旱對玉米產量的綜合影響[J]. 中國農業氣象,2007,28(4):440-442.Chen Zhenlin, Zhang Jianping, Wang Chunyi, et al.Application of WOFOST model in simulation of integrated impacts of low temperature and drought on maize yield[J].Chinese Journal of Agrometeorology, 2007, 28(4): 440-442.(in Chinese with English abstract)
[8]張建平,趙艷霞,王春乙,等. 氣候變化情景下東北地區玉米產量變化模擬[J]. 中國生態農業學報,2008,16(6):1448-1452.Zhang Jianping, Zhao Yanxia, Wang Chunyi, et al.Simulation of maize production under climate change scenario in Northeast China[J]. Chinese Journal of Eco-Agriculture, 2008, 16(6): 1448-1452. (in Chinese with English abstract)
[9]張建平,趙艷霞,王春乙,等. 不同時段低溫冷害對玉米灌漿和產量的影響模擬[J]. 西北農林科技大學學報(自然科學版),2012, 40(9):115-121.Zhang Jianping, Zhao Yanxia, Wang Chunyi, et al. Modeling the impact of low temperature disaster during different development stages on grain filling processes and yields of maize[J]. Journal of Northwest A&F University (Nat.Sci.Ed.),2012, 40(9): 115-121. (in Chinese with English abstract)
[10]張建平,趙艷霞,王春乙,等. 不同發育期低溫冷害對玉米灌漿和產量影響模擬[J]. 中國農學通報,2012, 28(36):176-182.Zhang Jianping, Zhao Yanxia, Wang Chunyi, et al. The simulation of the effects of low temperature disasters on maize growth and yield during different development stages[J]. Chinese Agricultural Science Bulletin, 2012,28(36): 176-182. (in Chinese with English abstract)
[11]張素青,張建濤,李繼蕊,等. WOFOST模型在河南省夏玉米主產區的校準與驗證[J]. 河南農業科學,2014, 43(8):152-156.Zhang Suqing, Zhang Jiantao, Li Jirui, et al. Calibration andvalidation of WOFOST in main maize-producing regions in Henan[J]. Journal of Henan Agricultural Sciences, 2014,43(8): 152-156. (in Chinese with English abstract)
[12]孫琳麗,侯瓊,馬玉平,等. WOFOST模型在內蒙古河套灌區模擬玉米生長全程的適應性[J]. 生態學雜志,2016,35(3):1-10.Sun Linli, Hou Qiong, Ma Yuping, et al. Adaptability of WOFOST model to simulate the whole growth period of maize in Hetao irrigation region of Inner Mongolia[J].Chinese Journal of Ecology, 2016, 35(3): 1-10. (in Chinese with English abstract)
[13]陳思寧,趙艷霞,申雙和,等. 基于 PyWOFOST作物模型的東北玉米估產及精度評估[J]. 中國農業科學,2013,46(14):2880-2893.Chen Sining, Zhao Yanxia, Shen Shuanghe, et al. Study on maize yield estimation and accuracy assessment based on PyWOFOST crop model in Northeast China[J]. Scientia Agricultura Sinica, 2013, 46(14): 2880-2893. (in Chinese with English abstract)
[14]張建平,趙艷霞,王春乙,等. 基于 WOFOST作物生長模型的冬小麥干旱影響評估技術[J]. 生態學報,2013,33(6):1762-1769.Zhang Jianping, Zhao Yanxia, Wang Chunyi, et al.Evaluation technology on drought disaster to yields of winter wheat based on WOFOST crop growth model[J]. Acta Ecologica Sinica, 2013, 33(6): 1762-1769. (in Chinese with English abstract)
[15]Chen Yingying, Yang Kun, He Jie, et al. Improving land surface temperature modeling for dry land of China[J].Journal of Geophysical Research Atmospheres, 2011,116(116): 999-1010.
[16]Yang Kun, He Jie, Tang Wenjun, et al. On downward shortwave and longwave radiations over high altitude regions:Observation and modeling in the Tibetan Plateau[J].Agricultural and Forest Meteorology, 2010, 150(1): 38-46.
[17]Huang Jianxi, Sedano Fernando, Huang Yanbo, et al.Assimilating a synthetic Kalman filter leaf area index series into the WOFOST model to improve regional winter wheat yield estimation[J]. Agricultural and Forest Meteorology,2016(216): 188-202.
[18]Huang Jianxi, Tian Liyan, Liang Shunlin, et al. Improving winter wheat yield estimation by assimilation of the leaf area index from Landsat TM and MODIS data into the WOFOST model[J]. Agricultural and Forest Meteorology, 2015(204):106-121.
[19]Ma Guannan, Huang Jianxi, Wu Wenbin, et al. Assimilation of MODIS-LAI into the WOFOST model for forecasting regional winter wheat yield[J]. Mathematical & Computer Modelling, 2013, 58(3–4): 634-643.
[20]Ma Hongyuan, Huang Jianxi, Zhu Dehai, et al. Estimating regional winter wheat yield by assimilation of time series of HJ-1 CCD NDVI into WOFOST-ACRM model with Ensemble Kalman filter[J]. Mathematical & Computer Modelling, 2013, 58(3–4): 759-770.
[21]Huang Jianxi, Ma Hongyuan, Su Wei, et al. Jointly assimilating MODIS LAI and ET products into the SWAP model to estimate winter wheat yield[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 20158(8): 4060-4071.
[22]Tian Liyan, Li Zhongxia, Huang Jianxi, et al. Comparison of two optimization algorithms for estimating regional winter wheat yield by integrating MODIS leaf area index and world food studies model[J]. Sensor Letters. 2013, 11(6–7): 1261-1268.
[23]何亮,侯英雨,趙剛,等. 基于全局敏感性分析和貝葉斯方法的 WOFOST作物模型參數優化[J]. 農業工程學報,2016,32(2):169-179.He Liang, Hou Yingyu, Zhao Gang, et al. Parameters optimization of WOFOST model by integration of global sensitivity analysis and Bayesian calibration method[J].Transactions of the Chinese Society of Agricultural Engineering(Transactions of the CSAE), 2016, 32(2): 169-179. (in Chinese with English abstract)
[24]黃健熙,馬鴻元,田麗燕,等. 基于時間序列 LAI和 ET同化的冬小麥遙感估產方法比較[J]. 農業工程學報,2015,31(4):197-203.Huang Jianxi, Ma Hongyuan, Tian Liyan, et al. Comparison of remote sensing yield estimation methods for winter wheat based on assimilating time-sequence LAI and ET[J].Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2015, 31(4): 197-203.(in Chinese with English abstract)
[25]黃健熙,武思杰,劉興權,等. 基于遙感信息與作物模型集合卡爾曼濾波同化的區域冬小麥產量預測[J]. 農業工程學報,2012(04):142-148.Huang Jianxi, Wu Sijie, Liu Xingquan, et al. Regional winter wheat yield forecasting based on assimilation of remote sensing data and crop growth model with Ensemble Kalman method[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE),2012(04): 142-148. (in Chinese with English abstract)
[26]邱美娟,宋迎波,王建林,等. 山東省冬小麥產量動態集成預報方法[J]. 應用氣象學報,2016, 27(2):191-200.Qiu Meijuan, Song Yingbo, Wang Jianlin, et al. Integrated technology of yield dynamic prediction of winter wheat in Shandong province[J]. Journal of applied meteorological science, 2016, 27(2): 191-200.(in Chinese with English abstract)
[27]Wu Dingrong, Yu Qiang, Lu Changhe, et al. Quantifying production potentials of winter wheat in the North China Plain[J]. European Journal of Agronomy, 2006, 24(3): 226-235.
[28]Cheng Zhiqiang, Meng Jihua, Wang Yiming. Improving spring maize yield estimation at field scale by assimilating time-series HJ-1 CCD data into the WOFOST model using a New Method with Fast Algorithms[J]. Remote Sensing, 2016,8(4): 303.
[29]Mallants D, Mohanty P B, Jacques D, et al. Spatial Variability Of Hydraulic Properties In A Multi-Layered Soil Profile[J]. Soil Science, 1996, 161(3): 167-181.
[30]戴彤,王靖,赫迪,等. APSIM模型在西南地區的適應性評價:以重慶冬小麥為例[J]. 應用生態學報,2015, 26(4):1237-1243.Dai Tong, Wang Jing, He Di, et al. Adaptability of APSIM model in Southwestern China: A case study of winter wheat in Chongqing City. Chinese Journal of Applied Ecology,2015, 26(4): 1237-1243. (in Chinese with English abstract)
Dynamic simulation of growth process of winter wheat in main production areas of China based on WOFOST model
Huang Jianxi1, Jia Shiling1, Ma Hongyuan1, Hou Yingyu2, He Liang2
(1.College of Information and Electrical Engineering, China Agricultural University, Beijing100083,China;2.National Meteorological Center, Beijing100081, China)
Crop model calibration and parameterization are essential for model evaluation and agricultural application. It is important for model application to accurately estimate the values of crop model parameters and further improve the performance of model prediction. WOFOST (world food studies) is a well-known, widely applied simulation model to analyze quantitatively the growth and production of field crops, which was originally developed for crops in European countries. It is the base model for Monitoring Agricultural Resources (MARS) Crop Growth Monitoring System (CGMS) in operational use for yield estimation in European Union. Dynamic simulation of WOFOST model in large regional scale is an important basis for regional crop modeling. In this study, we selected the main winter wheat production areas of China as the study area, and the data from 174 agricultural meteorological stations from 2011 to 2014 were used to calibrate several key WOFOST input parameters, especially 2 parameters related with variety, namely the effective accumulated temperature from emergence to flowering (TSUM1) and the effective accumulated temperature from flowering to maturity (TSUM2). On the basis of the zoning of the main winter wheat production areas, we used the meteorological data from 2012 to 2015 to drive the WOFOST model at a single-point scale, to simulate the winter wheat growth and dynamic development. The simulated phenology, LAI(leaf area index) and yield at the station level were evaluated with the field measured data. Results showed that the NRMSE(normalized root mean square error) of LAI ranged from 50% to 63%. The NRMSE of simulated days was 4%-7% from emergence to anthesis period and 8%-12% from anthesis to maturity period, and then CV (coefficient of variation) of the phenology was between 14% and 20%, which meant significant spatial variability. We simulated the yield respectively in irrigated area and rainfed area. And the NRMSE of simulated yield in irrigated area ranged from 11% to 23%, while the NRMSE of simulated yield in rain-fed area ranged from 22% and 28%, and the CV ranged from 14% to 22% for irrigated areas and from 25% to 40% for rain-fed areas, which exhibited significant spatial variability. The NRMSE of simulated LAI was between 50% and 63%, which could be explained that the LAI during different growth stages was all included into the accuracy analysis. Several important input parameters (such as TDWI (initial biomass) and SPAN (leaf senescence coefficient))could be optimized through assimilating remote sensing data into crop model, which could greatly improve the performance of crop model at the regional scale. Our results showed that the WOFOST model is of great potential for simulating the dynamic growth process of winter wheat in China. The calibrated WOFOST provides the dynamic model basis for regional applications,such as assimilating remote sensing data into crop model for crop yield estimation and climate change prediction with crop model.
models; optimization; temperature; WOFOST; winter wheat; parameter calibration; phendogy; dynamic simulation
10.11975/j.issn.1002-6819.2017.10.029
S127
A
1002-6819(2017)-10-0222-07
2016-10-07
2017-05-05
國家自然科學基金(41671418,41471342,41371326)
黃健熙,博士,博士生導師,主要從事農業定量遙感研究。北京中國農業大學,100083。Email:jxhuang@cau.edu.cn