徐 浩,張小虎,邱小雷,朱 艷,曹衛(wèi)星
·農業(yè)信息與電氣技術·
格網(wǎng)化小麥生長模擬預測系統(tǒng)設計與實現(xiàn)
徐浩,張小虎※,邱小雷,朱艷,曹衛(wèi)星
(南京農業(yè)大學國家信息農業(yè)工程技術中心,智慧農業(yè)教育部工程研究中心,農業(yè)農村部農作物系統(tǒng)分析與決策重點實驗室,江蘇省信息農業(yè)重點實驗室,江蘇現(xiàn)代作物生產協(xié)同創(chuàng)新中心,南京,210095)
利用作物生長模型模擬小麥區(qū)域生產力,分析氣候變化對農業(yè)生產的影響是研究糧食安全的熱點問題之一。擁有操作方便、計算快速特點的小麥區(qū)域生產力模擬系統(tǒng),可有效提高作物生長模型區(qū)域應用能力。該研究在分解小麥生長模型WheatGrow算法基礎上,利用Python語言構建了格網(wǎng)化小麥生長模型,實現(xiàn)了基于空間格網(wǎng)數(shù)據(jù)的小麥區(qū)域生產力模擬。驗證試驗結果表明:模擬產量的均方根誤差為1 070 kg/hm2,標準均方根誤差小于20%,系統(tǒng)所集成的WheatGrow模型具有較好的預測性;同時,結合格網(wǎng)數(shù)據(jù)分塊構建區(qū)域模擬的并行計算策略,優(yōu)化了區(qū)域模擬的性能。在此基礎上,采用GIS組件式開發(fā)模式,在.NET平臺下開發(fā)格網(wǎng)化小麥生長模擬預測系統(tǒng),實現(xiàn)作物生長模型與GIS耦合,為研究區(qū)域小麥產量潛力,評估氣候變化對小麥生長影響,制定農業(yè)決策提供軟件工具。
作物;模型;并行算法;GIS;格網(wǎng)數(shù)據(jù);系統(tǒng)開發(fā)
作物生長模型作為模擬作物生長、發(fā)育和產量形成的動態(tài)系統(tǒng)模型,可模擬作物生長關鍵物候期、光合物質生產及干物質分配、植株器官建成與產量品質形成等生理生態(tài)過程[1-3]。作物生長模型區(qū)域應用可反映區(qū)域作物生長水平,展示產量空間分布差異,量化生產力可增長幅度以及發(fā)現(xiàn)影響產量增長的限制因素,是評估糧食安全及優(yōu)化種植結構的有效工具[4-6]。隨世界土壤數(shù)據(jù)庫HWSD(Harmonized World Soil Database)[7],美國國家氣象數(shù)據(jù)中心的格網(wǎng)化氣象數(shù)據(jù)GHCN-M(Global Historical Climatology Network-Monthly)[8]、高空間分辨率的美國土壤剖面數(shù)據(jù)庫SSURGO(Soil Survey Geographic)[9],中國1:100萬的區(qū)域土壤數(shù)據(jù)庫[10]等地理格網(wǎng)數(shù)據(jù)集在不同時空范圍內可公開獲得,格網(wǎng)化區(qū)域生長模擬已成為作物生長模型和GIS耦合實現(xiàn)區(qū)域應用的主要手段,并在農業(yè)模型比較與改進項目(Agricultural Model Intercomparison and Improvement Project,AgMIP)的框架下,建立了多種全球格網(wǎng)作物生長模型(Global Gridded Crop Models,GGCMs)[5]。其核心策略是通過GIS集成作物生長模型與地理格網(wǎng)數(shù)據(jù)庫,實現(xiàn)區(qū)域模擬分析,并能反映生產力要素在空間上的變異特征。然而,中國自主的格網(wǎng)化作物生長模型及其軟件系統(tǒng)研究并未得到足夠重視。
目前,格網(wǎng)化區(qū)域生長模擬的實現(xiàn)可分為松耦合與緊耦合2種方法[1,4,11]。松耦合是指作物生長模型模塊與GIS模塊分別構建獨立系統(tǒng)。其中,GIS模塊用來管理格網(wǎng)數(shù)據(jù),作物生長模型系統(tǒng)用來模擬計算。這種方法通過GIS獲取模型輸入數(shù)據(jù),轉換成作物生長模型系統(tǒng)的文件格式提供給作物生長模擬模塊進行計算,并將計算結果賦值到每個模擬格網(wǎng)。兩者通過文本文件交互方式實現(xiàn)彼此調用,具有開發(fā)代碼量少,難度小的優(yōu)點,目前應用較為廣泛[4, 12-13]。但松耦合需要在GIS與作物生長模型之間進行切換并轉換數(shù)據(jù),容易導致數(shù)據(jù)出錯,分析效率較低。同時,此方法一般不提供可視化圖形操作界面,增加了非專業(yè)人員使用難度。緊耦合將GIS模塊與作物生長模型模塊集成到一起,是基于GIS的作物生長空間模擬系統(tǒng)。其中,GIS集成了數(shù)據(jù)管理、分析和交互的所有功能,用戶只需簡單點擊操作即可進行作物生長模型的數(shù)據(jù)選擇、區(qū)域模擬與結果展示,大大減少錯誤率,提高作物生長模型在區(qū)域上的可用性[14]。此過程需利用GIS的系統(tǒng)架構和實現(xiàn)方式對原作物生長模型進行重構、耦合,將區(qū)域生產模擬功能發(fā)布成可在桌面GIS、網(wǎng)絡GIS及移動GIS上都能調用的地理處理工具(Geo-processing工具)。這種方式實現(xiàn)困難,目前相關研究較少。同時,在緊耦合方式中,當格網(wǎng)數(shù)據(jù)空間分辨率較高時,依次遍歷每個格網(wǎng)執(zhí)行作物生長模型的串行方式效率低下,并需在內存中加載全部格網(wǎng)數(shù)據(jù),內存消耗較高。近些年基于MPI(Message Passing Interface)的并行計算編程技術,充分利用計算機CPU多核心資源,能夠實現(xiàn)格網(wǎng)數(shù)據(jù)并行處理,提高區(qū)域計算效率,并降低內存消耗。該技術已在GIS和遙感領域廣泛應用[15-18],但在作物生長模型區(qū)域模擬研究中應用較少。
綜上,本研究基于軟件工程的思想,采用GIS與作物生長模型緊耦合的方式,基于自主研發(fā)的小麥生長模型WheatGrow,設計實現(xiàn)格網(wǎng)化小麥生長模擬預測系統(tǒng),能夠便捷的實現(xiàn)區(qū)域小麥生產力的高效模擬預測。將解決2個關鍵問題:1)如何構建格網(wǎng)化小麥生長模型;2)如何實現(xiàn)格網(wǎng)化小麥生長模型的并行計算,提高計算效率。
根據(jù)格網(wǎng)化小麥生長模擬系統(tǒng)的應用目的及操作方式,本系統(tǒng)應實現(xiàn)以下功能:
1)數(shù)據(jù)管理。實現(xiàn)模型運行所需空間數(shù)據(jù)及其他數(shù)據(jù)的導入、導出、刪除更新等操作;
2)地圖管理。實現(xiàn)底圖加載及顯示,圖層放大、縮小、漫游等操作,并制作生成區(qū)域模擬結果專題地圖;
3)生長模擬。在保留傳統(tǒng)單個站點模擬功能的基礎上,實現(xiàn)利用格網(wǎng)化小麥生長模型的區(qū)域模擬和結果驗證;
4)生產力預測。實現(xiàn)不同模擬情景的小麥區(qū)域生產力模擬,包括光溫生產潛力、雨養(yǎng)生產潛力、氮素生產潛力,并可計算模擬產量與統(tǒng)計產量之間的產量差值;
5)統(tǒng)計分析。結合區(qū)劃數(shù)據(jù),利用分區(qū)統(tǒng)計對區(qū)域模擬結果均值、標準差以及年際間變化進行做圖分析。
根據(jù)功能需求分析,本系統(tǒng)主要包括數(shù)據(jù)管理、地圖管理、生長模擬、生產力預測與統(tǒng)計分析5個模塊。
本研究基于系統(tǒng)的需求分析,分解重構了南京農業(yè)大學自主研發(fā)的小麥生長模型WheatGrow算法[6],采用Python編程語言結合Numpy包和Multiprocessing包實現(xiàn)并行化的格網(wǎng)化WheatGrow模型,并在ESRI?ArcGIS的環(huán)境下,將格網(wǎng)化WheatGrow封裝成可調用的地理處理工具;同時,結合格網(wǎng)化區(qū)域小麥生產力模擬的數(shù)據(jù)需求構建空間數(shù)據(jù)庫;最終,在Microsoft?.Net環(huán)境下利用ESRI?ArcGIS Engine組件進行系統(tǒng)的開發(fā)實現(xiàn)(圖1)。
圖1 格網(wǎng)化小麥生長模擬預測系統(tǒng)開發(fā)技術路線圖
本系統(tǒng)基于Microsoft?.NET平臺,以ESRI?ArcGIS Engine組件式開發(fā)為基礎,使用C#開發(fā)的桌面端應用程序。主要包括基礎設施層、數(shù)據(jù)資源層、軟件構建層與客戶終端層(圖2)。
圖2 系統(tǒng)架構
基礎設施層包含系統(tǒng)開發(fā)、運行所需的基礎軟硬件環(huán)境,包括操作系統(tǒng)、存儲設備、數(shù)據(jù)庫軟件、服務器設備、安全設備等。
數(shù)據(jù)資源層存儲基礎地理數(shù)據(jù),以及模型計算所需區(qū)域氣象、土壤、品種參數(shù)及管理措施等農業(yè)空間數(shù)據(jù)。其中,農業(yè)空間數(shù)據(jù)在入庫前對數(shù)據(jù)進行幾何修復、規(guī)范命名、統(tǒng)一投影等預處理操作。
系統(tǒng)的數(shù)據(jù)包括空間數(shù)據(jù)集與屬性數(shù)據(jù)集2部分(表1),屬性數(shù)據(jù)集利用關系數(shù)據(jù)庫進行存儲,空間數(shù)據(jù)利用ArcGIS空間數(shù)據(jù)庫存儲[19]??臻g數(shù)據(jù)包括2部分,一是WheatGrow模擬所需氣象、土壤、品種和管理措施的格網(wǎng)數(shù)據(jù),且每種數(shù)據(jù)提供了10、20、30、…、190與200 km,共20種空間分辨率;二是觀測站點空間分布與區(qū)劃矢量數(shù)據(jù)。屬性數(shù)據(jù)包括氣象站點歷史監(jiān)測數(shù)據(jù),產量統(tǒng)計數(shù)據(jù),并通過設置關聯(lián)字段與矢量數(shù)據(jù)建立關聯(lián)。
表1 系統(tǒng)數(shù)據(jù)庫中主要數(shù)據(jù)列表
2.3.1 GIS模塊
GIS模塊實現(xiàn)空間數(shù)據(jù)讀取、顯示及操作,實現(xiàn)空間分析功能,包括站點數(shù)據(jù)的空間插值,利用區(qū)劃數(shù)據(jù)對格網(wǎng)模擬結果分區(qū)統(tǒng)計,并能實現(xiàn)區(qū)域生產力模擬結果的專題地圖制作。
2.3.2 WheatGrow模型及其校正
WheatGrow模型是以小麥生長發(fā)育及生產力形成的過程機理為基礎,以氣象、土壤、品種和管理技術為驅動變量而構建的小麥生長模擬模型。模型包括頂端發(fā)育與物候期、光合作用與干物質生產、器官發(fā)育與建成、同化物分配與產量形成、土壤水分與養(yǎng)分平衡5個子模型[20-22]。模型利用作物生理發(fā)育時間(Physiological Development Time,PDT)來劃分作物發(fā)育階段,可模擬小麥通過光合作用將CO2轉化成干物質積累及同化物分配,最終形成小麥收產量品質的過程。所有模擬的動態(tài)過程都受到土壤水分與養(yǎng)分平衡的共同影響(圖3)。WheatGrow模型可以模擬光溫生產潛力、水分限制下的生產潛力和氮素限制下的生產潛力限制3種等級水平下小麥生長發(fā)育狀況。
圖3 WheatGrow模型結構流程圖
為驗證模型的準確性,本研究從中國小麥主產省份山東、江蘇、安徽、湖北、河南、重慶、四川、陜西、山西、河北、天津選取共45個典型生態(tài)觀測站點2000年至2009年小麥生產記錄數(shù)據(jù)對WheatGrow模型進行驗證(圖4)。數(shù)據(jù)記錄了京東8號、魯麥21、川麥107、鄭麥9023等10個品種的生育期、產量等觀測數(shù)據(jù)。結果表明:均方根誤差RMSE為1 070 kg/hm2,標準均方根誤差NRMSE小于20%,表明系統(tǒng)所集成的WheatGrow模型具有較好的預測性[23]。
2.3.3 格網(wǎng)化WheatGrow模型及實現(xiàn)
1)格網(wǎng)化WheatGrow模型區(qū)域模擬
為解決單點尺度的WheatGrow模型在區(qū)域生產力模擬的升尺度問題,研究采用基于空間插值的升尺度策略將WheatGrow模型從單點模擬擴展到區(qū)域應用[5]。該策略利用空間插值這種“由點到面”的GIS空間分析方法獲取區(qū)域模型輸入格網(wǎng)數(shù)據(jù);利用GIS柵格代數(shù)方法重構WheatGrow模型,從而實現(xiàn)逐格網(wǎng)計算從而獲取區(qū)域模擬結果的目的[6]。格網(wǎng)模擬結果也可利用區(qū)域均值統(tǒng)計等方法進一步獲取區(qū)域平均產量水平。
圖4 基于WheatGrow模型的小麥籽粒產量模擬值與觀測值比較
2)模型輸入格網(wǎng)數(shù)據(jù)獲取及管理
格網(wǎng)化WheatGrow模型所需的氣象、土壤、品種和管理措施具有不同的數(shù)據(jù)特征,研究采用了不同插值方法生產相應的格網(wǎng)數(shù)據(jù)。其中,歷史和未來氣候情景下的氣象數(shù)據(jù)和土壤數(shù)據(jù)分別采用了ANUSPLIN和克里金插值[6,24-25]。品種參數(shù)為冬麥區(qū)各副區(qū)的代表性品種,因此利用矢量轉格網(wǎng)的方式獲取格網(wǎng)數(shù)據(jù)。所有輸入?yún)?shù)的格網(wǎng)數(shù)據(jù)按照數(shù)據(jù)資源層的規(guī)范和要求利用空間數(shù)據(jù)庫統(tǒng)一管理,便于系統(tǒng)的高效安全訪問(表2)。
表2 Gridded WheatGrow模型輸入數(shù)據(jù)
3)格網(wǎng)化WheatGrow模型計算優(yōu)化
當格網(wǎng)數(shù)據(jù)空間分辨率較高時,格網(wǎng)數(shù)目較多,以逐格網(wǎng)串行任務計算方式耗時較長;采用格網(wǎng)數(shù)據(jù)分塊的策略將串行任務分解成功能獨立的子任務,可將數(shù)據(jù)并行能力最大化,數(shù)據(jù)通訊與同步開銷最小化[26]。所以針對格網(wǎng)數(shù)據(jù)特征,本系統(tǒng)從數(shù)據(jù)并行的角度,通過判斷計算機CPU數(shù)目,將格網(wǎng)數(shù)據(jù)劃分成指定大小的塊,使用Python Multiprocessing包將每塊格網(wǎng)數(shù)據(jù)交給不同CPU核心單元獨立進行計算,確保CPU滿載運行,提高運算效率。最終,通過將不同CPU計算結果進行合并,從而獲取整個區(qū)域格網(wǎng)化模擬結果。
一般格網(wǎng)數(shù)據(jù)在計算機中以二維矩陣的形式進行組織,其數(shù)據(jù)劃分方式可分為一維行劃分、一維列劃分,二維劃分與不規(guī)則劃分等多種方式[27]。由于二維數(shù)組在物理層中按行存儲,所以本研究采用一維行劃分對格網(wǎng)數(shù)據(jù)進行劃分。利用格網(wǎng)數(shù)據(jù)的行數(shù)除以計算機CPU數(shù)目,得出并行計算所需分塊數(shù)目,從而實現(xiàn)并行計算中的負載均衡。
4)格網(wǎng)化WheatGrow模型封裝
本研究利用ESRI?ArcMap?將格網(wǎng)化WheatGrow模型封裝成ArcToolbox地理處理工具,本工具兼容ArcMap?10.0-10.7版本。通過預先設定好模型數(shù)據(jù)讀取路徑,只需在前端輸入格網(wǎng)數(shù)據(jù)路徑、起始年份、終止年份、模擬情景、格網(wǎng)數(shù)據(jù)的空間分辨率、模擬結果保存路徑,即可實現(xiàn)區(qū)域生產力模擬。
客戶終端層包含實現(xiàn)和顯示用戶界面的組件,管理客戶對信息的請求,通過用戶與Windows窗體的交互,實現(xiàn)地圖操作及調用作物生長模型相關業(yè)務邏輯操作。
研究采用C#和Python 2.7為開發(fā)語言,基于Microsoft?Visual Studio 2010?的開發(fā)環(huán)境,以ESRI?ArcGIS Engine?10.2為GIS組件,并結合Microsoft?SQL Server 2008?R2與ESRI?的空間數(shù)據(jù)引擎ArcSDE構建空間數(shù)據(jù)庫,進而開發(fā)實現(xiàn)了格網(wǎng)化小麥生長模擬預測系統(tǒng)。用戶可通過界面點擊選擇與輸入的交互方式實現(xiàn)包括單點模擬、區(qū)域模擬、統(tǒng)計分析、專題制圖等主要功能,為相關研究和應用工作提供了友好的軟件平臺。
光溫生產潛力指在理想條件下作物可以達到的最高理論產量,區(qū)域光溫生產潛力模擬對合理開發(fā)利用農業(yè)氣候資源具有重要意義[28]。所以,本研究以2013年中國冬麥區(qū)光溫生產潛力為模擬情景,對系統(tǒng)進行案例應用分析(圖5)。案例選擇了5、10、40、50、100、200 km,6個不同的空間分辨率。軟件的運行環(huán)境為32核Intel?Xeon(R)?CPU E5-2667 v3 @ 3.20 GHz,64位Microsoft?Windows 7 Professional?,內存64 GB。
運行結果表明本研究設計構建的系統(tǒng)完全實現(xiàn)了格網(wǎng)化的小麥區(qū)域生長模型,能夠高效便捷的模擬區(qū)域小麥生產力,并實現(xiàn)專題制圖。從運行效率上看,所提出的并行優(yōu)化策略顯著提升了系統(tǒng)模擬速度(表3),數(shù)據(jù)空間分辨率為5 km時,串行計算消耗時間超過7 d,而并行加速后模擬時間約為1.6 h,模擬效率顯著提高。此外,系統(tǒng)采用并行優(yōu)化后,模擬1 a數(shù)據(jù)僅需占用100 MB以內的計算機內存,提高了內存使用效率。
圖5 區(qū)域光溫生產潛力模擬系統(tǒng)界面
表3 不同空間分辨率數(shù)據(jù)集下的模型運行所需時間與內存
從專題圖上看,系統(tǒng)能夠根據(jù)用戶的需求,選擇不同空間分辨率的格網(wǎng)數(shù)據(jù)實現(xiàn)區(qū)域光溫生產潛力的模擬(圖6)。模擬結果的空間分布趨勢基本一致,但是格網(wǎng)輸入數(shù)據(jù)的分辨率對區(qū)域光溫生產潛力模擬結果具有顯著影響,隨空間分辨率逐漸增大,模擬結果空間分布細節(jié)特征逐漸消失。這種可視化分析結果也表明了本研究所構建的系統(tǒng)能夠為未來區(qū)域生產力模擬的進一步優(yōu)化提供數(shù)據(jù)管理和可視化分析的支持。
圖6 2013年中國冬麥區(qū)不同空間分辨率下光溫生產潛力模擬
本研究針對小麥生長模型區(qū)域應用中的技術需求,構建格網(wǎng)化小麥生長模擬預測系統(tǒng),實現(xiàn)了小麥區(qū)域生產力的模擬預測及分析制圖,并取得如下結論:
1)提出的網(wǎng)格化小麥生長模擬預測方案,能夠較好的拓展WheatGrow模型的區(qū)域模擬功能,實現(xiàn)區(qū)域小麥生產力的高效、準確模擬預測。
2)實現(xiàn)的格網(wǎng)化小麥生長模擬預測系統(tǒng),提供了高效的格網(wǎng)化小麥生長模擬模型和較完善的模擬數(shù)據(jù)庫;系統(tǒng)也可基于運行環(huán)境的CPU數(shù)目實現(xiàn)并行計算,有效提高了區(qū)域模擬效率。同時,所封裝的格網(wǎng)化小麥生長模型ArcGIS Toolbox工具,便于有專業(yè)基礎的用戶在ESRI?ArcGIS環(huán)境下實現(xiàn)定制化的需求。
[1] Rosenzweig C, Elliott J, Deryng D, et al. Assessing agricultural risks of climate change in the 21st century in a global gridded crop model intercomparison[J]. Proceedings of the National Academy of Sciences, 2014, 111(9): 3268-3273.
[2] Maiorano A, Martre P, Asseng S, et al. Crop model improvement reduces the uncertainty of the response to temperature of multi-model ensembles[J]. Field Crops Research, 2017, 202: 5-20.
[3] 孫寧,馮利平. 利用冬小麥作物生長模型對產量氣候風險的評估[J]. 農業(yè)工程學報,2005,21(2):106-110.
Sun Ning, Feng Liping. Assessing the climatic risk to crop yield of winter wheat using crop growth models[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2005, 21(2): 106-110. (in Chinese with English abstract)
[4] Resop J P, Fleisher D H, Wang Q, et al. Combining explanatory crop models with geospatial data for regional analyses of crop yield using field-scale modeling units[J]. Computers and Electronics in Agriculture, 2012, 89: 51-61.
[5] Mller C, Elliott J, Chryssanthacopoulos J, et al. Global gridded crop model evaluation: Benchmarking, skills, deficiencies and implications[J]. Geoscientific Model Development, 2017, 10: 1403-1422.
[6] Zhang X, Xu H, Jiang L, et al. Selection of appropriate spatial resolution for the meteorological data for regional winter wheat potential productivity simulation in China based on WheatGrow model[J]. Agronomy, 2018, 8(10): 198.
[7] Tifafi M, Guenet B, Hatté C. Large differences in global and regional total soil carbon stock estimates based on SoilGrids, HWSD and NCSCD: Intercomparison and evaluation based on field data from USA, England, Wales and France[J]. Global Biogeochemical Cycles, 2018, 32: 42-56.
[8] Lawrimore J H, Menne M J, Gleason B E, et al. An overview of the Global Historical Climatology Network monthly mean temperature data set, version 3[J]. Journal of Geophysical Research: Atmospheres, 2011, 116: 19121.
[9] Di L M, Arnold J G, Srinivasan R. Integration of SSURGO maps and soil parameters within a geographic information system and nonpoint source pollution model system[J]. Journal of Soil and Water Conservation, 2004, 59(4): 123-133.
[10] 張定祥,史學正,于東升,等. 中國1:100萬土壤數(shù)據(jù)庫建設的基礎[J]. 地理學報,2002,57(Z1):82-86.
Zhang Dingxiang, Shi Xuezheng, Yu Dongsheng, et al. The basis for establishing China's 1:1,000,000 soil database[J]. Acta Geographica Sinica, 2002, 57(Z1): 82-86. (in Chinese with English abstract)
[11] Villoria N B, Elliott J, Müller C, et al. Rapid aggregation of global gridded crop model outputs to facilitate cross-disciplinary analysis of climate change impacts in agriculture[J]. Environmental Modelling and Software, 2016, 75: 193-201.
[12] Shelia V, Hansen J, Sharda V, et al. A multi-scale and multi-model gridded framework for forecasting crop production, risk analysis, and climate change impact studies[J]. Environmental Modelling and Software, 2019, 115: 144-154.
[13] Kim S H, Kim J, Walko R, et al. Climate change impacts on maize-yield potential in the Southwestern United States[J]. Procedia Environmental Sciences, 2015, 29: 279-280.
[14] 王志強, 甘國輝, 王健, 等. 基于Web服務和GIS的作物生長模擬系統(tǒng)及應用[J]. 農業(yè)工程學報, 2008, 24(1): 179-182.
Wang Zhiqiang, Gan Guohui, Wang Jian, et al. Crop growth simulation system based on web services and GIS and its application[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2008, 24(1): 179-182. (in Chinese with English abstract)
[15] Rashmi C, Chaluvaiah S, Kumar G H. An efficient parallel block processing approach for K-means algorithm for high resolution orthoimagery satellite images[J]. Procedia Computer Science, 2016, 89: 623-631.
[16] 姜海燕,尹言,彭川陽,等. 作物生長模型分布式并行調度方案的比較[J]. 農業(yè)工程學報,2011,27(6):247-253.
Jiang Haiyan, Yin Yan, Peng Chuanyang, et al. Comparison of distributed parallel scheduling schemes for crop growth model[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2011, 27(6): 247-253. (in Chinese with English abstract)
[17] 趙青松,陳林,孫波,等. 基于Hadoop的云環(huán)境下作物生長模型算法的實現(xiàn)與測試[J]. 農業(yè)工程學報,2013,29(8):179-186.
Zhao Qingsong, Chen Lin, Sun Bo, et al. Algorithm implementation and tested of crop growth model based on hadoop of cloud computing[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2013, 29(8): 179-186. (in Chinese with English abstract)
[18] 楊靜靜,馬駿. 遙感影像區(qū)域面積快速計算并行算法研究[J]. 計算機時代,2020(6):8-12.
Yang Jingjing, Ma Jun. Research on parallel algorithm for fast calculation of area in remote sensing image[J]. Computer Era, 2020(6): 8-12. (in Chinese with English abstract)
[19] 李德元,姚文龍,楊二龍,等. 基于ArcSDE文件地理數(shù)據(jù)庫存儲和設計的應用研究[J]. 測繪與空間地理信息,2016,39(2):82-84.
Li Deyuan, Yao Wenlong, Yang Erlong, et al. The application of file geodatabase memory and design based on ArcSDE[J]. Geomatics & Spatial Information Technology, 2016, 39(2): 82-84. (in Chinese with English abstract)
[20] Liu T, Cao W, Luo W. Calculation of physiological development time and prediction of development stages after heading[J]. Acta Tritical Crops, 2000, 20: 29-34.
[21] Lv Z, Liu X, Cao W, et al. Climate change impacts on regional winter wheat production in main wheat production regions of China[J]. Agricultural & Forest Meteorology, 2013, 171-172(Complete): 234-248.
[22] 嚴美春,曹衛(wèi)星,羅衛(wèi)紅,等. 小麥發(fā)育過程及生育期機理模型的研究I.建模的基本設想與模型的描述[J]. 應用生態(tài)學報,2000,11(3): 355-359.
Yan Meichun, Cao Weixing, Luo Weihong, et al. A mechanistic model of phasic and phenological development of wheat. I. Assumption and description of the model[J]. Chinese Journal of Applied Ecology, 2000, 11(3): 355-359. (in Chinese with English abstract)
[23] Xu H, Huang F, Zuo W, et al. Impacts of spatial zonation schemes on yield potential estimates at the regional scale[J]. Agronomy, 2020, 10(5): 631.
[24] Zhang X, Zuo W, Zhao S, et al. Uncertainty in upscaling in situ soil moisture observations to multiscale pixel estimations with kriging at the field level[J]. International Journal of Geo-Information, 2018, 7(1): 33.
[25] 陳鋒銳,秦奮,李熙,等. 基于多元地統(tǒng)計的土壤有機質含量空間格局反演[J]. 農業(yè)工程學報,2012,28(20):188-194.
Chen Fengrui, Qin Fen,Li Xi, et al. Inversion for spatial distribution of soil organic matter content based on multivariate geostatistics[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2012,28(20):188-194. (in Chinese with English abstract)
[26] 姜海燕,彭川陽,尹言,等. 作物生育期模擬并行調度算法的研究與設計[J]. 江蘇農業(yè)學報,2010,26(6):1210-1216.
Jiang Haiyan, Peng Chuanyang, Yin Yan, et al. Design of parallel scheduling algorithm for crop development simulation[J]. Jiangsu Journal of Agricultural Sciences, 2010, 26(6): 1210-1216. (in Chinese with English abstract)
[27] 江嶺,湯國安,劉凱,等.局部型地形因子并行計算方法研究[J]. 地球信息科學學報,2012,14(6):761-767.
Jiang Ling, Tang Guoan, Liu Kai, et al. Study on parallel calculation method of local terrain parameters[J]. Journal of Geo-Information Science, 2012, 14(6): 761-767. (in Chinese with English abstract)
[28] 周治國,曹衛(wèi)星,王紹華,等. 基于GIS的區(qū)域作物生產系統(tǒng)潛力分析[J]. 農業(yè)工程學報,2003,19(1):124-128.
Zhou Zhiguo, Cao Weixing, Wang Shaohua, et al. GIS-Based potential productivity analysis of regional crop production system[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2003, 19(1): 124-128. (in Chinese with English abstract)
Design and implementation of gridded simulation and prediction system for wheat growth
Xu Hao, Zhang Xiaohu※, Qiu Xiaolei, Zhu Yan, Cao Weixing
(,,210095,)
To simulate the regional wheat productivity can be an essential way to evaluate the impacts of climate change on food security. Generally, a crop growth model is available to the wheat productivity prediction at the regional and national scales for decision-making. However, a convenient and fast software system is necessary to improve the ability of the crop growth model, thereby to efficiently calculate regional wheat productivity. In this research, a simulation platform was designed to implement a gridded wheat growth model (Gridded WheatGrow) for regional wheat forecast, combining with the observed weather data. The Gridded WheatGrow model is derived from the WheatGrow model that invented by Nanjing Agricultural University, particularly for a process-based wheat growth simulation. The Gridded WheatGrow model can be used to integrate the gridded data and the simulation model within a geographical information system (GIS). The modified model can facilitate the acquisition of the input data, such as the daily meteorological data and soil data with a high spatial resolution, for the season forecasts, wheat productivity simulation, and other application of most previous open grid databases. The Gridded WheatGrow model can be a core component of a GIS. This is because the simulation system, not just an independent software, was designed based on the close integration of GIS and crop model. Furthermore, this design can simplify the data preparation, further to make it friendly to a non-professional user. A parallel computing method was adopted with the strategy of grid data partition based on Message Passing Interface (MPI), in order to solve the time-consuming and inconvenient problems caused from the modeling and calculation of grid data in the application of the regional productivity simulation. As such, the grid data can be dynamically segmented into a certain number of blocks, according to the number of the CPU cores in a computer and the size of the original grid data. Therefore, the computation of regional productivity simulation can efficiently utilize the full capacity of CPU in the computer, while reduce the consumption of stored physical memory. In the case of high efficiency, a normal personal computer can also be used to develop a gridded simulation system of wheat productivity. The proposed system was implemented based on the component geographic information system in development mode using the Microsoft?platform, together with net developer platform, C# and Python programming language. The Gridded WheatGrow model was also served as a specific geoprocessing tool for ESRI?ArcGIS. In ArcMap?module of the system, the customized code can be used to simulate the regional wheat productivity on the specific purpose. The proposed system was verified by the field data collected from the winter wheat area in China, and the root mean square error (RSME) and normalized root mean square error (NRSME) are1 070 kg/hm2and less than 20%, respectively, showing an excellent performance. The typical system can be used to simulate the regional wheat productivity with a friendly user interface, while to reduce time and consumption of physical memory. Combining with the fundamental functions of GIS, the simulated data can be easily visualized and mapping for the later public use. All these features of the proposed system can prove that the Gridded WheatGrow simulation platform is an useful and reliable software on regional wheat productivity forecasts, and thereby it can be expected to evaluate the impacts of climate change on food security and decision making in modern agriculture.
crops; models; parallel algorithms; GIS; grid data; system development
徐浩,張小虎,邱小雷,等. 格網(wǎng)化小麥生長模擬預測系統(tǒng)設計與實現(xiàn)[J]. 農業(yè)工程學報,2020,36(15):167-172.doi:10.11975/j.issn.1002-6819.2020.15.021 http://www.tcsae.org
Xu Hao, Zhang Xiaohu, Qiu Xiaolei, et al. Design and implementation of gridded simulation and prediction system for wheat growth[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2020, 36(15): 167-172. (in Chinese with English abstract) doi:10.11975/j.issn.1002-6819.2020.15.021 http://www.tcsae.org
2020-04-27
2020-07-28
國家重點研發(fā)項目(2016YFD0300607);國家自然科學基金國際合作與交流項目(41961124008)
徐浩,博士生,研究方向為農業(yè)空間數(shù)據(jù)分析與建模。Email:haoxu1989@hotmail.com
張小虎,博士,副教授,研究方向為農業(yè)時空大數(shù)據(jù)智能分析。Email:zhangxiaohu@njau.edu.cn
10.11975/j.issn.1002-6819.2020.15.021
S24; TP311.5
A
1002-6819(2020)-15-0167-06