基于无人机热红外的水分胁迫指数与土壤含水率关系研究

为了实时快速监测作物根系活动层的土壤含水率,利用低空无人机搭载的热红外相机获取经4种水分处理的棉花花铃期一天中5个时刻的冠层温度,并连续观测5 d,应用水分胁迫指数(CWSI)的理论模式、简化模式、定义模式计算得到3种CWSI,与棉花根系不同土壤深度含水率建立模型。研究表明:3种胁迫指数与土壤含水率具有幂函数关系,其中理论模式与土壤含水率的相关性最佳,定义模式次之,简化模式最差;在一天中不同监测时间点上,3种CWSI的监测精度在13∶00最高,9∶00和17∶00最差;在监测深度上,3种胁迫指数与0～60 cm处的土壤含水率关系最为紧密,0～30 cm次之,0～15 cm最差。该研究可大面积获取作物根系层土壤含水率,提高作物根系层土壤含水率的反演精度。     

基于作物水分胁迫指数的表层土壤含水率遥感估算

【目的】探究作物水分胁迫指数(CWSI)与表层土壤含水率的空间分布特征,并分析不同下垫面(葵花、夏玉米、春小麦和甜椒)表层土壤含水率的遥感估算精度。【方法】利用MOD16A2遥感数据和气象数据,结合Penman-Monteith(P-M)模型,基于CWSI反演河套灌区解放闸灌域表层土壤含水率,并对不同下垫面的表层土壤含水率进行验证。【结果】CWSI的空间分布与表层土壤含水率相反,CWSI大的区域,表层土壤含水率小;春小麦下垫面遥感估算的表层土壤含水率效果较好,决定系数(R²)为0.748,其次为葵花,R²为0.357;灌溉次数较多的夏玉米和甜椒的表层土壤含水率估算精度较差,可见基于CWSI的表层土壤含水率遥感估算方法对土壤干旱较为敏感。【结论】基于CWSI的表层土壤含水率遥感估算方法更适用于灌水较少且耐旱作物下垫面的表层土壤含水率估算。     

基于LightGBM的温室番茄冠层CWSI预测模型研究

为研究温室内番茄冠层作物水分胁迫指数（CWSI）问题,通过布设多参数传感器,实时获取温室内外各环境参数。利用灰度关联分析,计算各环境参数与番茄冠层CWSI的关联度,根据关联度对环境参数进行排序,同时考虑对模型精度的影响,最终从9个环境参数中选取7个作为模型输入,建立基于LightGBM的温室番茄冠层CWSI预测模型。结合贝叶斯算法优化其中的关键参数,将模型预测结果与通过Jones经验公式计算出的CWSI做相关性分析,在相同的运算环境下,分别与GBRT和SVR模型对比。试验结果表明,基于贝叶斯优化LightGBM模型的决定系数（R2）、平均绝对误差（MAE）、均方根误差（RMSE）和运算时间分别为0.9601、0.0218、0.0314和0.0518s,与GBRT和SVR模型相比,其R2分别提高2.14%和14.05%,MAE分别降低0.0093和0.0612,RMSE分别降低0.0097和0.0591,时间分别缩短0.0459s和0.0612s。表明本研究提出的LightGBM模型性能更有效地提高了温室番茄冠层CWSI的预测精度,为实现温室番茄按需灌溉提供了参考。     

大田玉米水分胁迫指数CWSI基线建立方法试验研究

作物水分胁迫指数(Crop water stress index,CWSI)经验模型的建立与气候和种植条件密切相关。本文以内蒙古自治区鄂尔多斯市达拉特旗大田玉米为对象,研究CWSI的最优经验模型。玉米在营养生长阶段(Vegetative stage,V期)、生殖期(Reproductive stage,R期)和成熟期(Maturation stage,M期)分别进行不同灌溉水平的处理,采用红外测温传感器采集玉米冠层温度。分别结合田间和实验地旁标准气象站空气温湿度数据建立了CWSI经验模型的无水分胁迫基线。基于2种无水分胁迫基线,分别利用饱和水汽压梯度获取的无蒸腾作用基线和5℃无蒸腾作用基线建立了4种CWSI经验模型,得出反映大田玉米水分胁迫状况的关系曲线,并进行对比。结果表明,基于实验地旁标准气象站空气温湿度数据建立的CWSI经验模型具有很大的波动性,并不能很好反映玉米的水分胁迫状况,其值常常超出正常范围0～1。而基于田间空气温湿度数据建立的CWSI经验模型则可以很好地监测内蒙古自治区大田玉米水分胁迫状况,M期3种不同水分处理100%、52%和28%具有较好的CWSI数值梯度,分别为0.04、0.14和0.32。相比于基于饱和水汽压梯度获取的无蒸腾作用基线,以5℃作为无蒸腾作用基线时得到的CWSI数值较小,可以较好地反映水分胁迫状况,对应上述M期3种不同水分处理CWSI值分别为0.02、0.10和0.22,具有较为合理的梯度。经过初步检验和分析,认为基于田间空气温湿度数据建立的CWSI经验模型较为合理,可以有效监测大田玉米水分胁迫状况。     

基于冠层温度的温室葡萄CWSI模型试验研究

探讨并建立了适合于镇江丘陵地区温室葡萄水分状况监测的作物水分胁迫指数(CWSI)模型.通过田间实验和观测,得到了适合温室葡萄的CWSI经验模型中的经验关系.初步的检验和分析表明,这一模型是合理的,可以用于温室内葡萄基于冠层温度信息的水分状况监测.     

大田玉米水分胁迫指数经验模型建立方法

作物水分胁迫指数(Crop water stress index,CWSI)经验模型的建立与气候和种植条件密切相关。以内蒙古自治区鄂尔多斯市达拉特旗大田玉米为对象,研究了CWSI的最优经验模型。玉米在营养生长阶段(Vegetative stage,V期)、生殖期(Reproductive stage,R期)和成熟期(Maturation stage,M期)分别进行不同灌溉水平的处理,采用红外测温传感器采集玉米冠层温度。分别结合田间和实验地旁标准气象站空气温湿度数据建立了CWSI经验模型的无水分胁迫基线。基于2种无水分胁迫基线,分别利用饱和水汽压梯度获取的无蒸腾作用基线和5℃无蒸腾作用基线建立了4种CWSI经验模型,得出反映大田玉米水分胁迫状况的关系曲线,并进行对比。结果表明,基于实验地旁标准气象站空气温湿度数据建立的CWSI经验模型具有很大的波动性,并不能很好反映玉米的水分胁迫状况,其值常常超出正常范围(0~1)。而基于田间空气温湿度数据建立的CWSI经验模型则可以很好地监测内蒙古自治区大田玉米水分胁迫状况,M期3种不同水分处理100%、52%和28%具有较好的CWSI数值梯度,分别为0.03、0.14和0.32。相比于基于饱和水汽压梯度获取的无蒸腾作用基线,以5℃作为无蒸腾作用基线时得到的CWSI数值较小,可以较好地反映水分胁迫状况,对应上述M期3种不同水分处理CWSI值分别为0.02、0.10和0.22,具有较为合理的梯度。经过初步检验和分析,认为基于田间空气温湿度数据建立的CWSI经验模型较为合理,可以有效监测大田玉米水分胁迫状况。     

无人机热红外遥感反演玉米根域土壤含水率方法研究

为了减少土壤背景带来的干扰,更加准确、高效的获取无人机热红外图像中的玉米冠层温度,进而快速反演玉米地土壤含水率,以4种水分梯度处理的拔节期玉米为研究对象,借助无人机可见光和热红外图像,采用RGRI指数法、Otsu阈值法和不剔除土壤背景3种处理方法提取热红外图像中玉米冠层温度信息,计算作物水分胁迫指数(Crop water stress index,CWSI)并用于反演不同水分梯度处理下玉米地不同深度的土壤含水率,基于3种方法获得的CWSI分别记为CWSIRGRI、CWSIOtsu、CWSIsc.结果表明:①基于RGRI指数法获取的玉米冠层温度与实测冠层温度的相关性最高(R2均大于0.8;RMSE均小于1℃),Otsu方法次之,不剔除土壤背景方法效果最差.②在整个拔节期,CWSIRGRI反演土壤含水率效果最好(R2均大于0.5,P<0.01;效果显著),CWSIOtsu次之、CWSIsc反演效果最差.③选取CWSIRGRI为最优CWSI指标,其在玉米拔节期5个土壤深度内的R2呈现先上升后下降的趋势且都在0～30 cm深度内达到最大值.因此,基于RGRI指数法建立的CWSIRGRI可以作为反演玉米地土壤含水率的有效指标.     

基于无人机热红外遥感的冬小麦水分胁迫研究

为探究水分胁迫对冬小麦生长的影响,以不同水分处理的冬小麦为试验对象,利用无人机搭载热红外传感器,通过采集其不同生育期中一天不同时刻(11∶00,13∶00)的冠层热红外图像,提取其冠层温度信息,同时测定小麦叶片的气孔导度(Gs)、蒸腾速率(Tr)和田间土壤体积含水率(SWC)等信息。分别研究不同水分胁迫指数(CWSI、I_G、ICWSI)与各参数之间的关系,同时使用一元线性模型和多元线性回归模型进行建模并验证。结果表明:CWSI、I_G和ICWSI与Gs、Tr和SWC之间存在着显著的相关关系,在一元模型中,SWC对不同水分胁迫指数的预测效果更好,验证R~2均在0.800以上,相对分析误差均在2.0以上,在多元模型中,CWSI的预测效果最好,验证R~2为0.928,相对分析误差为3.041,同时多元模型的预测效果均优于一元模型。该研究可快速获取大量作物信息,为利用无人机热红外遥感探究冬小麦的水分胁迫状况提供了一条新途径。     

基于无人机热红外遥感的玉米地土壤含水率诊断方法

为使热红外遥感诊断土壤含水率更加准确、高效,以不同水分处理的大田玉米为研究对象,借助无人机可见光图像,对热红外图像进行植土分离,并提取玉米冠层温度和地表土壤温度。通过剔除温度直方图两端1%的温度像元对温度信息进行优化,进而计算作物水分胁迫指数(Crop water stress index,CWSI)、冠层相对温差(Canopy relative temperature difference,CRTD)、地表相对温差(Surface relative temperature difference,SRTD),利用三者之和求得水分-温度综合指数(Water-temperature composite index,WTCI),并用于诊断不同深度的土壤含水率。结果表明,剔除温度直方图两端1%温度像元的玉米冠层温度与实测冠层温度的相关性更高(4次试验的R2由0. 823、0. 886、0. 899、0. 876提高至0. 906、0. 938、0. 944、0. 922),剔除温度直方图前端1%温度像元的地表土壤温度与实测地表温度的相关性也更高(2次试验的R2由0. 841、0. 875提高至0. 908、0. 925),即通过直方图法优化的温度更接近实测温度;在拔节前期,CWSI、WTCI诊断0～20 cm土壤含水率效果较优,而拔节后期、抽雄吐丝期、乳熟期诊断0～40 cm土壤含水率效果较优;在半覆盖条件下,包含冠层温度信息(CWSI、CRTD)和土壤温度信息(SRTD)的WTCI1与土壤含水率的相关性更高(0～40 cm:决定系数为0. 500、0. 821,高于0. 463、0. 748);在全覆盖状态下,包含冠层相对温差(CRTD)的WTCI2与土壤含水率的相关性更高(0～40 cm:决定系数为0. 809、0. 729,高于0. 721、0. 656),表明WTCI是诊断土壤含水率效果较优的指标。     

基于高光谱成像技术的生菜冠层含水率检测

为实现作物含水率的无损检测,以6种水分胁迫水平的生菜为研究对象,利用高光谱成像技术和特征波长选取方法对生菜冠层含水率进行检测研究。采用掩模法去除高光谱图像的背景噪声,并对生菜冠层光谱图像进行光强校正。利用标准正态变量变换法(SNV)去除原始平均光谱数据的噪声,采用蒙特卡罗无信息变量消除法(MCUVE)剔除无关变量,结合基于最小绝对收缩和选择算法(LASSO)、连续投影法(SPA)、LASSO与SPA算法组合(LASSO SPA)筛选特征变量,对数据进行降维处理,采用偏最小二乘法(PLS)建立5个生菜冠层含水率检测模型。经对比发现,全光谱中存在很多冗余信息变量和无关变量,采用全光谱建立的PLS模型复杂度最高,且预测能力最差;以MCUVE LASSO SPA筛选变量后的PLS模型效果最优,其中建模集相关系数R c和预测集相关系数R p分别为0.8827和0.9015,均方根误差分别为1.0662和0.9287。择优选取MCUVE LASSO SPA PLS模型计算生菜冠层每个像素点的干基含水率,生成可视化分布图,实现了生菜冠层叶片干基含水率可视化检测。本研究可为生菜冠层含水率快速无损检测提供参考。     

油菜氮素光谱定量分析中水分胁迫与光照影响及修

研究了油菜氮素和水分胁迫在光谱检测中的相互作用,以及光照的变化对作物光谱检测的影响.为了克服光照因素对光谱检测的影响,针对氮素和水分的特征波长分别建立了基于光谱反射率变化率的光照修正模型;为了实现对氮素和水分相互作用的解耦,针对不同含水率水平的植株分别建立了全氮含量光谱特征的主成分回归模型.利用氮素光谱组合分析模型对植株全氮含量进行预测,结果表明,实测值与预测值的相关系数R为0.92,均方根误差(RMSE)为0.53,优于前期采用6特征波长变量和主成分回归法所建立的预测模型.     

冠层温度定量诊断覆膜作物水分状况试验研究

作物缺水指标 CWSI( Crop Water Strese Index)和冠层 -空气温差 ( Tc-Ta)是利用冠层温度评价作物水分状况的重要方法。 1 998和 1 999年在新疆乌兰乌苏农业气象站试验田内开展了对覆膜棉花和玉米的研究。结果表明 :CWSI能够指示作物根系层的水分状况 ,而 ( Tc-Ta)受到环境因素 (太阳辐射、空气饱和差 )的较大影响。另外 ,对用标准化的冠层 -空气温差法 NDT( Normalized Difference of Temperature)定量诊断作物水分状况的可行性进行了研究。结果表明该方法在很大程度上能消除环境因素的影响 ,直接指示作物根系层的水分状况 ,并提出了覆膜棉花和玉米各生育阶段需灌溉的临界 CWSI及 NDT值     

Infrared thermometry: A remote sensing technique for predicting yield in water-stressed cotton

A crop water stress index (CWSI) was derived from air temperatures, air vapor pressure deficits and the midday radiant leaf temperatures of cotton plants that were exposed to different early-season irrigation treatments at Phoenix, AZ, U.S.A. To calculate the CWSI, an infrared thermometer was used to measure leaf temperatures which were then scaled relative to minimum and maximum temperatures expected for no-stress (CWSI=0) and extreme drought-stress conditions (CWSI=1). Results showed the CWSI behaved as expected, dropping to low levels following an irrigation and increasing gradually as the cotton plants depleted soil moisture reserves. The final yield of seed cotton was significantly inversely correlated with the average CWSI observed over the interval from the appearance of the first square until two weeks following the final irrigation.     

Energy balance and crop water stress in winter maize under phenology-based irrigation scheduling

In eastern India, cultivation of winter maize is getting popular after rainy season rice and farmers practice irrigation scheduling of this crop based on critical phenological stages. In this study, crop water stress index of winter maize at different critical stages wase determined to investigate if phenology-based irrigation scheduling could be optimized further. The components of the energy budget of the crop stand were computed. The stressed and non-stressed base lines were also developed (between canopy temperature and vapor pressure deficit) and with the help of base line equation, [(T _c − T _a) = −1.102 VPD − 3.772], crop water stress index (CWSI) was determined from the canopy-air temperature data collected frequently throughout the growing season. The values of CWSI (varied between 0.42 and 0.67) were noted just before the irrigations were applied at critical phenological stages. The soil moisture depletion was also measured throughout the crop growing period and plotted with CWSI at different stages. Study revealed that at one stage (silking), CWSI was much lower (0.42–0.48) than that of recommended CWSI (0.60) for irrigation scheduling. Therefore, more research is required to further optimize the phenology-based irrigation scheduling of winter maize in the region. This method is being used now by local producers. The intercepted photosynthetically active radiation and normalized difference vegetation index over the canopy of the crop were also measured and were found to correlate better with leaf area index.     

油菜钵苗钵体的力学特性

针对油菜钵苗移栽过程中钵体受损影响钵苗成活率的问题,研究油菜钵苗钵体在顶苗夹具作用下的运动及力学特性,探究在不同含水率下钵体顶出力、脱离位移和承压力之间的变化规律,为油菜移栽机顶苗取苗机构参数设计提供依据。结果表明:钵体顶出力随钵体含水率的提高而增大;当钵体含水率由20.44%提高至31.02%时,脱离位移的范围由3.04~4.23 mm增长至5.02~5.44 mm,钵体承压力由5.61 N增长至7.90 N;当钵体含水率由31.02%提高至40.84%时,脱离位移的范围基本不变,钵体承压力由7.90 N减少至4.83 N。即钵体承压力随含水率的提高先增大后减少,呈非线性变化,且在相同含水率下,钵体的承压力基本不变;综合得出钵体含水率在30%左右时,有利于油菜钵苗被顶出,且钵体受损较小。      

Application of a new method to evaluate crop water stress index

Optimum water management and irrigation require timely detection of crop water condition. Usually crop water condition can be indicated by crop water stress index (CWSI), which can be estimated based on the measurements of either soil water or plant status. Estimation of CWSI by canopy temperature is one of them and has the potential to be widely applied because of its quick response and remotely measurable features. To calculate CWSI, the conventional canopy-temperature-based model (Jackson’s model) requires the measurement or estimation of the canopy temperature, the maximum canopy temperature (T _cu), and the minimum canopy temperature (T _cl). Because extensive measurements are necessary to estimate T _cu and T _cl, its application is limited. In this study, by introducing the temperature of an imitation leaf (a leaf without transpiration, T _p) and based on the principles of energy balance, we studied the possibility to replace T _cu by T _p and reduce the included parameters for CWSI calculation. Field experiments were carried out in a winter wheat (Triticum aestivum L.) field in Luancheng area, Hebei Province, the main production area of winter wheat in China. Six irrigation treatments were established and soil water content, leaf water potential, soil evaporation rate, plant transpiration rate, biomass, yield, and regular meteorological variables of each treatment were measured. Results indicate that the values of T _cu agree with the values of T _p with a regression coefficient r=0.988. While the values of CWSI estimated by the use of T _p are in agreement with CWSI by Jackson’s method, with a regression coefficient r=0.999. Furthermore, CWSI estimated by the use of T _p has good relations with soil water content and leaf water potential, showing that the estimated CWSI by T _p is a good indicator of soil water and plant status. Therefore, it is concluded that T _cu can be replaced by T _p and the included parameters for CWSI calculation can be significantly reduced by this replacement.     

基于CWSI诊断温室草皮水分胁迫的实验研究

通过观测夏季温室不同灌溉条件下草皮的冠层温度、气温、大气湿度以及土壤含水量等因素,利用Isdo经验模式确定了冠气温差的下限方程。通过观察不同水分处理条件下草皮CWSI的日变化,得出了利用CWSI诊断草皮水分状况的最佳时机。研究分析了作物水分胁迫指数与其他一些反映作物水分状况的指标,包括土壤含水量、叶片蒸腾速率以及叶片含水量之间的关系,CWSI验理论模式与上述这些指标关系良好,表明其很好地反映了作物的水分胁迫特征。     

Evaluating canopy temperature-based indices for irrigation scheduling

Summary Since the development of commercial versions of infrared sensors, they have been increasingly used to determine canopy temperature and schedule irrigations. However, some shortcomings of the technique have been identified, among them the sensitivity of canopy temperature measurements to weather fluctuations. Based on field and computer simulated data, an analysis of the suitability of crop water stress indices (CWSI's) developed from canopy temperature under variable weather conditions was done. Important day to day fluctuations of CWSI values determined using an empirical baseline (empirical CWSI) appeared common for nonstressed crops, particularly under low vapor pressure deficit conditions. These fluctuations generate uncertainty in the use of this empirical index to determine needs for irrigation. The use of an improved index (theoretical CWSI) requiring measurements of net radiation, soil heat flux and wind speed, and estimates of aerodynamic and canopy resistances reduced but did not eliminate these fluctuations. Results using a simulation model showed that the empirical CWSI provided late indication of irrigation needs, after some water stress has developed, which may limit its application for crops sensitive to water stress. These simulations also indicated that the theoretical CWSI was able to track the development of water stress and provide reasonable indication of irrigation needs. However, this result may not be fully realized in field applications where the determination of CWSI may be affected by various sources of variability which are not accounted for by the model.     

Relations between relative evapotranspiration and predawn leaf water potential in soybean grown in several locations

Summary The measurement of water consumption in the field is normally restricted to research purposes, although the development of practical field criteria for timing water application is required to improve crop productivity. To develop such criteria irrigation experiments on Soybean were conducted from flowering to grain filling at four locations which differed in their soil properties and the convective contribution of their climates to potential evapotranspiration. The energy balance, predawn leaf water potential (PLWP), soil moisture depletion, and a crop water stress index (CWSI) based on foliage temperature were measured. The range of soil, atmospheric, phenological and irrigation conditions, produced a common, linear relation between relative evapotranspiration (rET) and the logarithm of -PLWP. Correlation with the temperature based CWSI was weak. A similar relation with PLWP for other C₃ plants was also derived from data in the literature. This relation could be helpful for irrigation scheduling once the critical values of rET for crop productivity are known.     

基于PCA_SVR的油菜氮素光谱特征定量分析模型

研究了采用光谱分析技术对油菜植株全氮进行定量分析的方法.采用逐步回归法对氮素的光谱特征波长进行选择,为克服光谱变量间多重共线性的影响,对变量进行了主成分分析(PCA),为提高模型的拟合优度,应用支持向量机回归(SVR)建立油菜氮素的定量分析模型.对不同氮素水平的油菜冠层光谱数据进行分析,结果表明,406、460、556、634、662、675nm的光谱反射率与油菜含氮量呈极显著相关.植株全氮SVR模型预测值与实测值的相关系数为0.89,模型的检验误差(RMSE)为2.51.