本文選題：黃河三角洲 + 鹽堿地分級(jí)��； 參考：《山東師范大學(xué)》2015年碩士論文

【摘要】：一直以來，鹽堿地問題都是全球性的問題和難題。鹽堿地破壞作物賴以生存的土壤環(huán)境，影響作物的正常生長(zhǎng)，甚至導(dǎo)致作物死亡，阻礙了農(nóng)業(yè)生產(chǎn)的正常進(jìn)行和生態(tài)環(huán)境的平衡發(fā)展。據(jù)國(guó)際糧農(nóng)組織及聯(lián)合國(guó)教科文組織的相關(guān)統(tǒng)計(jì)，目前全球鹽堿地的總面積約9.54億hm2，其中我國(guó)所占面積為9900余萬hm2。鹽堿地問題已經(jīng)成為制約我國(guó)農(nóng)業(yè)經(jīng)濟(jì)發(fā)展的重要因素。 黃河三角洲地區(qū)的鹽堿地問題較為突出，已經(jīng)嚴(yán)重制約該地區(qū)農(nóng)業(yè)的正常發(fā)展，加強(qiáng)治理刻不容緩。所謂“知己知彼，百戰(zhàn)不殆”，了解該地區(qū)的鹽堿地分布狀況與鹽堿化程度是治理的首要任務(wù)。 本文在黃河三角洲地區(qū)進(jìn)行實(shí)地土壤采樣，并對(duì)采樣樣本進(jìn)行實(shí)驗(yàn)，測(cè)得采樣點(diǎn)土壤樣本的土壤含鹽量。在ArcGIS10.1平臺(tái)上生成采樣點(diǎn)的點(diǎn)要素?cái)?shù)據(jù)，并存儲(chǔ)各采樣點(diǎn)的土壤含鹽量。 對(duì)環(huán)境與災(zāi)害監(jiān)測(cè)預(yù)報(bào)小衛(wèi)星星座B星（HJ-1B星）的同期遙感影像進(jìn)行預(yù)處理，并提取采樣點(diǎn)位置各個(gè)波段的遙感影像反射率值。在SPSS平臺(tái)上對(duì)影像各個(gè)波段反射率值與相應(yīng)的土壤含鹽量進(jìn)行Pearson相關(guān)系數(shù)的運(yùn)算，并計(jì)算各個(gè)波段的標(biāo)準(zhǔn)差，從而得到各個(gè)波段的診斷系數(shù)，以此來表明各個(gè)波段對(duì)土壤含鹽量的敏感程度。結(jié)果表明，Band1、Band2和Band3波段的土壤含鹽量敏感性較為統(tǒng)一，且明顯高于Band4波段，此三波段可以用來進(jìn)行土壤含鹽量的遙感定量反演運(yùn)算。 在SPSS平臺(tái)上，以Band1、Band2和Band3三個(gè)波段為自變量，使用逐步回歸分析的方法進(jìn)行多元線性回歸分析運(yùn)算，以建立影像反射率值與土壤含鹽量之間的關(guān)系模型。結(jié)果顯示，Band2波段自變量被剔除，Band1和Band3波段自變量作為保留自變量進(jìn)行了回歸運(yùn)算，運(yùn)算結(jié)果顯著。 BP神經(jīng)網(wǎng)絡(luò)以其較強(qiáng)的非線性回歸映射能力和突出的自學(xué)習(xí)自適應(yīng)能力而被廣泛使用。將BP神經(jīng)網(wǎng)絡(luò)引入到土壤含鹽量的遙感反演中來，可以充分利用其自身特點(diǎn)，為遙感定量反演提供一種全新的方法和途徑。本文在MATLAB平臺(tái)上構(gòu)建了一個(gè)包含輸入層、隱含層和輸出層的三層BP神經(jīng)網(wǎng)絡(luò)模型，使用2-16-1結(jié)構(gòu)，即輸入層2個(gè)節(jié)點(diǎn)，隱含層16個(gè)節(jié)點(diǎn)，輸出層2個(gè)節(jié)點(diǎn)。將Band1和Band3作為輸入節(jié)點(diǎn)，使用Log-Sigmoid作為訓(xùn)練函數(shù)，進(jìn)行網(wǎng)絡(luò)模型訓(xùn)練。通過精度檢驗(yàn)對(duì)比BP神經(jīng)網(wǎng)絡(luò)模型和多元線性回歸模型的預(yù)測(cè)能力，，結(jié)果顯示，前者在大部分情況下要強(qiáng)于后者。因此，使用訓(xùn)練得到的神經(jīng)網(wǎng)絡(luò)模型對(duì)黃河三角洲地區(qū)的土壤含鹽量進(jìn)行仿真。 結(jié)果表明，黃河三角洲地區(qū)的鹽堿地鹽堿化程度較為嚴(yán)重，其中以重度鹽堿地和鹽田為代表的原生鹽堿地比重占到了近七成，且集中分布在黃河以及大片水域集中的地區(qū)附近。由于水勢(shì)高、水量大等因素導(dǎo)致這些地區(qū)的地下水埋深變淺，從而引起鹽堿地的鹽堿化程度加深。 最后，本文描述了黃河三角洲地區(qū)鹽堿地的成因，并對(duì)治理提出了相關(guān)建議。盡管人為措施無法真正阻止該地區(qū)鹽堿地的形成，但通過科學(xué)合理的方法可以達(dá)到緩解鹽堿地鹽堿化程度的目的，從而減輕鹽堿地對(duì)農(nóng)業(yè)發(fā)展的制約。
[Abstract]:The problem of saline - alkali land has always been a global problem and difficult problem . The saline - alkali land destroys the soil environment that crops depend on , affects the normal growth of crops , and has hindered the normal development of agricultural production and the balanced development of the ecological environment . According to the relevant statistics of FAO and UNESCO , the total area of the global salt - alkali land is about 900 million hm2 . The problem of saline - alkali land has become an important factor restricting the development of agricultural economy in China .

The problem of saline - alkali land in the Yellow River Delta region is more prominent , which has seriously restricted the normal development of agriculture in the region and strengthened its governance .

In this paper , the soil samples were sampled in the Yellow River Delta region , and the samples were tested to measure the soil salinity of soil samples . The point element data of the sampling points were generated on the ArcGIS 10.1 platform , and the soil salinity of each sampling point was stored .

The results show that the sensitivity of each wavelength band to soil salinity is higher than that of Band1 , Band2 and Band3 bands , and the three wavelength bands can be used for remote sensing quantitative inversion of soil salinity .

Based on the SPSS platform , using three bands of Band1 , Band2 and Band3 as independent variables , multiple linear regression analysis was performed using stepwise regression analysis to establish the relationship model between the image reflectivity value and the soil salinity . The results show that the Band2 band independent variable is eliminated , the Band1 and Band3 band independent variables are used as the reserved independent variables , and the operation result is remarkable .

The BP neural network is widely used in remote sensing inversion with strong nonlinear regression mapping ability and prominent self - adaptive ability . The BP neural network is introduced into the remote sensing inversion of soil salinity , and a new method and approach for remote sensing quantitative inversion can be fully utilized . In this paper , a three - layer BP neural network model including input layer , hidden layer and output layer can be fully utilized .

The results show that the salinity of the saline - alkali land in the Yellow River Delta region is more serious , and the proportion of the native saline - alkali land represented by the severe saline - alkali land and the salt field accounts for nearly 70 % , and the concentration of the salt - alkali land in the Yellow River is close to the area concentrated in the Yellow River and the large area .

Finally , this paper describes the genesis of saline - alkali land in the Yellow River Delta region , and puts forward some suggestions for the treatment . Although artificial measures cannot really prevent the formation of saline - alkali land in the region , the aim of relieving the salinity of saline - alkali land can be achieved by scientific and reasonable method , so as to reduce the restriction of saline - alkali land to agricultural development .
【學(xué)位授予單位】：山東師范大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：S156.4

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 李彬;王志春;遲春明;;LEACHC模型及其在土壤鹽堿化預(yù)測(cè)預(yù)報(bào)研究中的應(yīng)用[J];安徽農(nóng)業(yè)科學(xué);2005年12期

2 翁永玲;宮鵬;;土壤鹽漬化遙感應(yīng)用研究進(jìn)展[J];地理科學(xué);2006年03期

3 關(guān)元秀,劉高煥,王勁峰;基于GIS的黃河三角洲鹽堿地改良分區(qū)[J];地理學(xué)報(bào);2001年02期

4 連煜;王新功;黃

本文編號(hào)：1988817