石磊
摘 要:隨著新型農(nóng)村居民區(qū)建設(shè)的不斷發(fā)展,農(nóng)民搬離原有居住區(qū)或遺棄村中舊房,導(dǎo)致村內(nèi)大量舊宅基地閑置,荒草叢生,形成了大量廢棄地坑窯洞。宅基地復(fù)墾改變了原有地形地貌,對(duì)已經(jīng)熟化的土壤耕層進(jìn)行了翻動(dòng)。但目前,對(duì)生土在耕作后磷、鉀等營(yíng)養(yǎng)元素的含量變化研究較少。由此,本研究對(duì)原始反射光譜進(jìn)行NOR、MSC、SNV處理后,分別進(jìn)行一、二階微分、反射率倒數(shù)對(duì)數(shù)變換,并采用偏最小二乘回歸法分別建立了P、K兩種重金屬元素的最優(yōu)高光譜估算模型。研究表明,研究區(qū)土壤中P較為豐富,K為豐富;進(jìn)行微分變換可以提高土壤中養(yǎng)分元素與反射光譜的相關(guān)性,使用其形成相關(guān)性較高的組合波段,使模型的穩(wěn)定性和預(yù)測(cè)能力提高。
關(guān)鍵詞:宅基地復(fù)墾;Vis-NIR;養(yǎng)分元素;反射光譜
中圖分類號(hào):S151.9 文獻(xiàn)標(biāo)識(shí)碼:A 文章編號(hào):1003-5168(2019)23-0144-03
Estimation of Soil Nutrient Element Content in Reclamation Area
of Homestead Based on Reflectance Spectrum
SHI Lei1,2,3,4
(1.Shaanxi Institute of Land Engineering Construction Co. Ltd.,Xi'an Shaanxi 710075;2.Institute of Shaanxi Land Engineering and Technology Co., Ltd.,Xi'an Shaanxi 710075;3.Key Laboratory of Degraded and Unused Land Consolidation Engineering, Ministry of Land and Resources,Xi'an Shaanxi 710075;4.Shaanxi Provincial Land Consolidation Engineering Technology Research Center,Xi'an Shaanxi 710075)
Abstract: With the continuous development of new rural residential area construction, farmers move out of the original residential area or abandon the old houses in the village, resulting in a large number of old residential land idle, overgrown with weeds, forming a large number of abandoned pits and caves. Homestead reclamation has changed the original topography and topography, turning the ripened soil tillage. However, there are few studies on the changes of phosphorus, potassium and other nutrient elements in raw soil after tillage. In this study, NOR, MSC and SNV were used to process the original reflectance spectra, and the first and second order differential and reciprocal logarithmic transformation of reflectance were performed respectively. The optimal Hyperspectral Estimation Models of P and K heavy metals were established by partial least squares regression. The results show that P is abundant and K is abundant in the soil of the study area. Differential transformation can improve the correlation between soil nutrient elements and reflectance spectra, and use it to form a combination band with high correlation, so as to improve the stability and prediction ability of the model.
Keywords: homestead reclamation;Vis-NIR;nutrient elements;reflective spectrum
宅基地復(fù)墾是指依據(jù)土地利用總體規(guī)劃以及土地整治開發(fā)規(guī)劃方案,對(duì)荒廢或者利用率極低的宅基地,復(fù)墾為耕地的行為[1]。在空心村土地整理實(shí)施過程中,改變了原有地形地貌,對(duì)已經(jīng)熟化的土壤耕層進(jìn)行了翻動(dòng)。但目前,對(duì)生土在耕作后磷、鉀等營(yíng)養(yǎng)元素的含量變化研究較少[2]。目前,人們主要通過對(duì)待測(cè)樣品進(jìn)行化學(xué)實(shí)驗(yàn),并通過計(jì)算得出土壤P、K含量[3,4]。傳統(tǒng)檢測(cè)方法中的化學(xué)實(shí)驗(yàn)較為煩瑣,且成本較高,不同的元素需要不同的前處理與實(shí)驗(yàn)儀器,在大規(guī)模的土壤質(zhì)量調(diào)查研究中效率較低。而高光譜技術(shù)可檢測(cè)的信息豐富、方便快捷,且不破壞供試土壤樣品的理化結(jié)構(gòu),因此,被廣泛應(yīng)用于土壤氮、磷、鉀等元素含量的預(yù)測(cè)[5,6]。近年來,國(guó)內(nèi)外學(xué)者在利用高光譜反演土壤有機(jī)質(zhì)、氮、磷、鉀等方面進(jìn)行了大量研究,為本研究提供了諸多借鑒[7]。Vis-NIR光譜和化學(xué)計(jì)量學(xué)的結(jié)合為準(zhǔn)確、快速監(jiān)測(cè)土壤在處理過程中的性質(zhì)(磷、鉀元素)變化提供了一種理想的方法,而無(wú)需化學(xué)分析。目的是建立澄城縣空心村土壤質(zhì)量定量模型,評(píng)價(jià)土壤光譜對(duì)土壤理化性質(zhì)的預(yù)測(cè)能力。
1 材料與方法
1.1 土樣采集
采用“S”布點(diǎn)法進(jìn)行土壤樣品采集,去除表層雜質(zhì),使用反射探頭對(duì)土壤進(jìn)行反射光譜測(cè)量,總計(jì)49個(gè)土壤樣品。將土壤樣品去除其他雜質(zhì),在自然條件下風(fēng)干后混合均勻,取300g樣品并過孔徑0.149mm篩用于室內(nèi)氮、磷、鉀含量測(cè)定。磷(P)、鉀(K)元素統(tǒng)計(jì)特征如表1所示。
1.2 光譜數(shù)據(jù)測(cè)定
選定測(cè)試樣點(diǎn)位置,使用便攜式ASD HR地物光譜儀對(duì)樣點(diǎn)進(jìn)行土壤反射光譜測(cè)定。地物光譜儀波長(zhǎng)范圍為300 ~2 500nm,采樣帶寬為1.3nm(350~ 1 000nm)和2nm(1 000~2 500nm),重采樣間隔為1nm,前端2cm的視野區(qū)可以避開土壤中的雜質(zhì)干擾反射。
2 數(shù)據(jù)分析
2.1 光譜微分變換
對(duì)原始反射光譜進(jìn)行反射率倒數(shù)對(duì)數(shù)、一階、二階微分變換,分析土壤光譜反射率,對(duì)其做3種變換,用于尋找適宜于不同元素的響應(yīng)區(qū)域。一階、二階微分變換增加反射率與測(cè)試元素間的相關(guān)性,也可限制或消除部分線性、接近線性的背景的影響。以一階微分[ρ'λi]、二階微分[ρ''λi]分別表示土壤原始光譜反射率,并進(jìn)行變換。計(jì)算公式為:
[ρλi=ρλi+1-ρλi-1/Δλ]? ? ? ? ? ? ? ? ? ? (1)
[ρλi=ρλi+1-ρλi-1/Δλ]? ? ? ? ? ? ? ? ? ? ? ? ?(2)
式中,[λi]為波長(zhǎng)[i]nm的波段,[Δλ=λi+1-λi=10 nm],[i]=400,410,…,2 450nm。
對(duì)數(shù)變換不但增強(qiáng)了可見光光譜的差異性,而且減少了光照條件下乘性因素的影響,對(duì)光譜數(shù)據(jù)做倒數(shù)的對(duì)數(shù)變換[log1ρλi]。
2.2 數(shù)據(jù)建模與驗(yàn)證
TQ Analyst根據(jù)馬氏距離隨機(jī)將全部數(shù)據(jù)集分為建模集和驗(yàn)證集,并采用PLS法建立預(yù)測(cè)模型。模型結(jié)果采用決定系數(shù)[R2]、均方根誤差RMSE進(jìn)行驗(yàn)證。計(jì)算公式如下:
[R2=i=1nyi-yi2i=1nyi-yi2]? ? ? ? ? ? ? ? ? ? ? ? ?(3)
[RMSE=1ni=1nyi-yi2]? ? ? ? ? ? ? ? ? ? ?(4)
式中,[yi]和[yi]分別為檢驗(yàn)樣本的觀測(cè)值和預(yù)測(cè)值;[yi]為樣本觀測(cè)值的平均值;[m]為校正集的樣品數(shù)量;[n]為驗(yàn)證樣本數(shù)量。
3 結(jié)果與討論
3.1 土壤養(yǎng)分元素含量分析
供試土壤類型為土,共采集土壤樣品49個(gè)。其中,土壤pH值介于7.59~8.29,屬于堿性土壤。土壤中P的含量介于44.78~172.64mg·kg-1,根據(jù)農(nóng)用地質(zhì)量分等標(biāo)準(zhǔn),表現(xiàn)為較豐富。土壤中的K的含量介于9.23~84.62mg·kg-1,根據(jù)農(nóng)用地質(zhì)量分等標(biāo)準(zhǔn),表現(xiàn)為豐富。
3.2 模型的建立
原始反射光譜作為對(duì)照組,對(duì)反射光譜數(shù)據(jù)經(jīng)過SNV、MSC、NOR處理,分別進(jìn)行一、二階微分和反射率倒數(shù)對(duì)數(shù)變換后均進(jìn)行Savitzky–Golay平滑,采用PLSR建立對(duì)應(yīng)的估算模型,并使用決定系數(shù)和均方根誤差進(jìn)行檢驗(yàn)。
針對(duì)不同光譜指標(biāo)R、R+SG、R+SG+FD、R+SG+SD和R+SG+MSC,采用偏最小二乘回歸法分別建立估算模型。通過與參照組對(duì)比:經(jīng)過預(yù)處理及微分變換后建立的回歸模型,無(wú)論是建模精度還是預(yù)測(cè)精度,均比基于原始數(shù)據(jù)建立的模型效果要好。
4 結(jié)論
本研究對(duì)原始反射光譜進(jìn)行NOR、MSC、SNV處理后,分別進(jìn)行一、二階微分、反射率倒數(shù)對(duì)數(shù)變換,并采用偏最小二乘回歸法分別建立了P、K兩種重金屬元素的最優(yōu)高光譜估算模型。通過模型模擬計(jì)算,得到以下結(jié)論。
①研究區(qū)土壤中P較為豐富,K為豐富,對(duì)區(qū)域內(nèi)的農(nóng)作物生長(zhǎng)有較大促進(jìn)作用。
②對(duì)反射光譜數(shù)據(jù)經(jīng)過SNV、MSC、NOR處理,分別進(jìn)行一、二階微分和反射率倒數(shù)對(duì)數(shù)變換,降低了土壤顆粒表面散射、粒徑大小不均勻等因素對(duì)光譜反射的影響。進(jìn)行微分變換可以提高土壤中養(yǎng)分元素與反射光譜的相關(guān)性,使用其形成相關(guān)性較高的組合波段,使模型的穩(wěn)定性和預(yù)測(cè)能力提高。
參考文獻(xiàn):
[1]Yang Z, Liu Y, Wu W, et al. Effects of rural–urban development transformation on energy consumption and CO2 emissions: A regional analysis in China[J]. Renewable & Sustainable Energy Reviews,2015(53):863-875.
[2]Liu Y. Revitalize the world's countryside[J].Nature,2017(7667):275-277.
[3]周清,周斌,張楊珠,等.成土母質(zhì)對(duì)水稻土高光譜特性及其有機(jī)質(zhì)含量光譜參數(shù)模型影響的研究初探[J].土壤學(xué)報(bào),2004(6):905-911.
[4]陳玉福,孫虎,劉彥隨.中國(guó)典型農(nóng)區(qū)空心村綜合整治模式[J].地理學(xué)報(bào),2010(6):727-735.
[5]Li C, Wu K. Driving forces of the villages hollowing based on geographically weighted regression model: a case study of Longde County, the Ningxia Hui Autonomous Region, China[J]. Natural Hazards,2017(3):1059-1079.
[6]Al-abbas A H, Swain P H, BAaumgardner M F. Relating organic matter and clay content to the multispectral radiance of soils[J]. Soil Science,1972(6):477-485.
[7] Siebielec G, Mccarty G W, Stuczynski T I. Near- and mid-infrared diffuse reflectance spectroscopy for measuring soil metal content[J].Journal of Environmental Quality,2004(6):2056-2069.