Pre-harvest wheat yield prediction using agromet-spectral-trend-yield models for hoshiarpur and rupnagar districts of punjab

Wheat yield prediction using different agrometeorological indices, spectral index (NDVI, Normalized Difference Vegetation Index) and trend predicted yield (TPY) were developed in Hoshiarpur and Rupnagar districts of Punjab. On the basis of examination of Correlation Coefficients (R), Standard Error of Estimate (SEOE) and Relative Deviation (RD) values resulted from different agromet models, the best agromet subset were selected as Minimum Temperature (T_min), Maximum Temperature (T_max) and accumulated Heliothermal Units (HTU) in case of Hoshiarpur district and Minimum Temperature (T--_min), accumulated Temperature Difference (TD) and accumulated Pan Evaporation (E) for Rupnagar district at reproductive stage (2^nd week of March) of wheat. It was found that Agromet-Spectral-Trend-Yield model could explain 96 % (SEOE = 87 kg/ha) and 91 % (SEOE = 146 kg/ha) of wheat yield variations for Hoshiarpur and Rupnagar districts, respectively.     

Estimating aboveground and organ biomass of plant canopies across the entire season of rice growth with terrestrial laser scanning

Non-destructive and accurate estimation of crop biomass is crucial for the quantitative diagnosis of growth status and timely prediction of grain yield. As an active remote sensing technique, terrestrial laser scanning (TLS) has become increasingly available in crop monitoring for its advantages in recording structural properties. Some researchers have attempted to use TLS data in the estimation of crop aboveground biomass, but only for part of the growing season. Previous studies rarely investigated the estimation of biomass for individual organs, such as the panicles in rice canopies, which led to the poor understanding of TLS technology in monitoring biomass partitioning among organs. The objective of this study was to investigate the potential of TLS in estimating the biomass for individual organs and aboveground biomass of rice and to examine the feasibility of developing universal models for the entire growing season. The field plots experiments were conducted in 2017 and 2018 and involved different nitrogen (N) rates, planting techniques and rice varieties. Three regression approaches, stepwise multiple linear regression (SMLR), random forest regression (RF) and linear mixed-effects (LME) modeling, were evaluated in estimating biomass with extensive TLS and biomass data collected at multiple phenological stages of rice growth across the entire season. The models were calibrated with the 2017 dataset and validated independently with the 2018 dataset.The results demonstrated that growth stage in LME modeling was selected as the most significant random effect on rice growth among the three candidates, which were rice variety, growth stage and planting technique. The LME models grouped by growth stage exhibited higher validation accuracies for all biomass variables over the entire season to varying degrees than SMLR models and RF models. The most pronounced improvement with a LME model was obtained for panicle biomass, with an increase of 0.74 in R² (LME: R² = 0.90, SMLR: R² = 0.16) and a decrease of 1.15 t/ha in RMSE (LME: RMSE =0.79 t/ha, SMLR: RMSE =2.94 t/ha). Compared to SMLR and RF, LME modeling yielded similar estimation accuracies of aboveground biomass for pre-heading stages, but significantly higher accuracies for post-heading stages (LME: R² = 0.63, RMSE =2.27 t/ha; SMLR: R² = 0.42, RMSE =2.42 t/ha; RF: R² = 0.57, RMSE =2.80 t/ha). These findings implied that SMLR was only suitable for the estimation of biomass at pre-heading stages and LME modeling performed remarkably well across all growth stages, especially for post-heading. The results suggest coupling TLS with LME modeling is a promising approach to monitoring rice biomass at post-heading stages at high accuracy and to overcoming the saturation of canopy reflectance signals encountered in optical remote sensing. It also has great potential in the monitoring of other crops in cloud-cover conditions and the instantaneous prediction of grain yield any time before harvest.     

Yield Forecasting of Spring Maize Using Remote Sensing and Crop Modeling in Faisalabad-Punjab Pakistan

Real time, accurate and reliable estimation of maize yield is valuable to policy makers in decision making. The current study was planned for yield estimation of spring maize using remote sensing and crop modeling. In crop modeling, the CERES-Maize model was calibrated and evaluated with the field experiment data and after calibration and evaluation, this model was used to forecast maize yield. A Field survey of 64 farm was also conducted in Faisalabad to collect data on initial field conditions and crop management data. These data were used to forecast maize yield using crop model at farmers’ field. While in remote sensing, peak season Landsat 8 images were classified for landcover classification using machine learning algorithm. After classification, time series normalized difference vegetation index (NDVI) and land surface temperature (LST) of the surveyed 64 farms were calculated. Principle component analysis were run to correlate the indicators with maize yield. The selected LSTs and NDVIs were used to develop yield forecasting equations using least absolute shrinkage and selection operator (LASSO) regression. Calibrated and evaluated results of CERES-Maize showed the mean absolute % error (MAPE) of 0.35–6.71% for all recorded variables. In remote sensing all machine learning algorithms showed the accuracy greater the 90%, however support vector machine (SVM-radial basis) showed the higher accuracy of 97%, that was used for classification of maize area. The accuracy of area estimated through SVM-radial basis was 91%, when validated with crop reporting service. Yield forecasting results of crop model were precise with RMSE of 255 kg ha^?1, while remote sensing showed the RMSE of 397 kg ha^?1. Overall strength of relationship between estimated and actual grain yields were good with R² of 0.94 in both techniques. For regional yield forecasting remote sensing could be used due greater advantages of less input dataset and if focus is to assess specific stress, and interaction of plant genetics to soil and environmental conditions than crop model is very useful tool.     

Estimation of rice grain yield from dual-polarization Radarsat-2 SAR data by integrating a rice canopy scattering model and a genetic algorithm

Fast and accurate estimation of rice yield plays a role in forecasting rice productivity for ensuring regional or national food security. Microwave synthetic aperture radar (SAR) data has been proved to have a great potential for rice monitoring and parameters retrieval. In this study, a rice canopy scattering model (RCSM) was revised and then was applied to simulate the backscatter of rice canopy. The combination of RCSM and genetic algorithm (GA) was proposed for retrieving two important rice parameters relating to grain yield, ear length and ear number density, from a C-band, dual-polarization (HH and HV) Radarsat-2 SAR data. The stability of retrieved results of GA inversion was also evaluated by changing various parameter configurations.Results show that RCSM can effectively simulate backscattering coefficients of rice canopy at HH and HV mode with an error of <1 dB. Reasonable selection of GA’s parameters is essential for stability and efficiency of rice parameter retrieval. Two rice parameters are retrieved by the proposed RCSM-GA technology with better accuracy. The rice ear length are estimated with error of <1.5 cm, and ear number density with error of <23 #/m². Rice grain yields are effectively estimated and mapped by the retrieved ear length and number density via a simple yield regression equation. This study further illustrates the capability of C-band Radarsat-2 SAR data on retrieval of rice ear parameters and the practicability of radar remote sensing technology for operational yield estimation.     

Large-area rice yield forecasting using satellite imageries

Ability to make large-area yield prediction before harvest is important in many aspects of agricultural decision-making. In this study, canopy reflectance band ratios (NIR/RED, NIR/GRN) of paddy rice (Oryza sativa L.) at booting stage, from field measurements conducted from 1999 to 2005, were correlated with the corresponding yield data to derive regression-type yield prediction models for the first and second season crop, respectively. These yield models were then validated with ground truth measurements conducted in 2007 and 2008 at eight sites, of different soil properties, climatic conditions, and various treatments in cultivars planted and N application rates, using surface reflectance retrieved from atmospherically corrected SPOT imageries. These validation tests indicated that root mean square error of predicting grain yields per unit area by the proposed models were less than 0.7 T ha⁻¹ for both cropping seasons. Since village is the basic unit for national rice yield census statistics in Taiwan, the yield models were further used to forecast average regional yields for 14 selected villages and compared with officially reported data. Results indicate that the average yield per unit area at village scale can be forecasted with a root mean square error of 1.1 T ha⁻¹ provided no damaging weather occurred during the final month before actual harvest. The methodology can be applied to other optical sensors with similar spectral bands in the visible/near-infrared and to different geographical regions provided that the relation between yield and spectral index is established.     

Application of partial least squares regression and WetSpa model to determine factors controlling sediment yield in Chardavol watershed,Iran

This paper examines the effects of watershed complexity in terms of physiography and land use on the specific sediment yield of the Chardavol watershed (1012.946 km²) in Iran. First, specific sediment yield was simulated using spatially distributed hydrological WetSpa model, then the influential factors such as morphometric variables, land-use composition and pattern and soil properties of the watershed were calculated at the sub-watershed scale. Due to the inter-reliant of these watershed characteristics, a partial least squares regression (PLSR) was used to illustrate the relationship between the specific sediment yield and data of 15 selected watershed characteristics. The results showed that the land-use composition and soil properties had the maximum effects on the specific sediment yield and clarified 79% of the variation in the specific sediment yield. Regarding the availability of digital spatial database over the watershed, this simple PLSR procedure could be applied in different watersheds.     

Development of a wheat yield model for Punjab using remotely sensed data and historical yield trends

Spectral yield models for Punjab were updated by incorporating the latest data set on district-wise Normalised Difference Vegetation Index (NDVI) and wheat yields. In order to improve the model, historical yield trend for the past ten years was used to derive a linear regression relation for each district. Yield predicted by these linear trend relations was evaluated for validation of the approach. Finally a multiple regression relation incorporating both NDVI and trend-predicted yield was developed. This model shows better prediction capability as seen from yield forecasts of 1991–92 season.     

Integration of maximum crop response with machine learning regression model to timely estimate crop yield

Timely and reliable estimation of regional crop yield is a vital component of food security assessment, especially in developing regions. The traditional crop forecasting methods need ample time and labor to collect and process field data to release official yield reports. Satellite remote sensing data is considered a cost-effective and accurate way of predicting crop yield at pixel-level. In this study, maximum Enhanced Vegetation Index (EVI) during the crop-growing season was integrated with Machine Learning Regression (MLR) models to estimate wheat and rice yields in Pakistan's Punjab province. Five MLR models were compared using a fivefold cross-validation method for their predictive accuracy. The study results revealed that the regression model based on the Gaussian process outperformed over other models. The best performing model attained coefficient of determination (R2), Root Mean Square Error (RMSE, t/ ha), and Mean Absolute Error (MAE, t/ha) of 0.75, 0.281, and 0.236 for wheat; 0.68, 0.112, and 0.091 for rice, respectively. The proposed method made it feasible to predict wheat and rice 6– 8 weeks before the harvest. The early prediction of crop yield and its spatial distribution in the region can help formulate efficient agricultural policies for sustainable social, environmental, and economic progress.     

Evaluation of machine learning models for rice dry biomass estimation and mapping using quad-source optical imagery

ABSTRACT

Several machine learning regression models have been advanced for the estimation of crop biophysical parameters with optical satellite imagery. However, literature on the comparative performances of such models is still limited in range and scope, especially under multiple data sources, despite the potential of multi-source imagery to improving crop monitoring in cloudy areas. To fill in this knowledge gap, this study explored the synergistic use of Landsat-8, Sentinel-2A, China’s environment and disaster monitoring and forecasting satellites (HJ-1 A and B) and Gaofen-1 (GF-1) data to evaluate four machine learning regression models that include Random Forest (RF), Support Vector Machine (SVM), k-Nearest Neighbor (k-NN), and Gradient Boosting Decision Tree (GBDT), for rice dry biomass estimation and mapping. Taking a major rice cultivation area in southeast China as case study during the 2016 and 2017 growing seasons, a cross-calibrated time series of the Enhanced Vegetation Index (EVI) was obtained from the quad-source optical imagery and on which the aforementioned models were applied, respectively. Results indicate that in the before rice heading scenario, the most accurate dry biomass estimates were obtained by the GBDT model (R² of 0.82 and RMSE of 191.8 g/m²) followed by the RF model (R² of 0.79 and RMSE of 197.8 g/m²). After heading, the k-NN model performed best (R² of 0.43 and RMSE of 452.1 g/m²) followed by the RF model (R² of 0.42 and RMSE of 464.7 g/m²). Whist the k-NN model performed least in the before heading scenario, SVM performed least in the after heading scenario. These findings may suggest that machine learning regression models based on an ensemble of decision trees (RF and GBDT) are more suitable for the estimation of rice dry biomass, at least with optical satellite imagery. Studies that would extend the evaluation of these machine learning models, to other parameters like leaf area index, and to microwave imagery, are hereby recommended.     

Monitoring canopy growth and grain yield of paddy rice in South Korea by using the GRAMI model and high spatial resolution imagery

Monitoring crop conditions and forecasting crop yields are both important for assessing crop production and for determining appropriate agricultural management practices; however, remote sensing is limited by the resolution, timing, and coverage of satellite images, and crop modeling is limited in its application at regional scales. To resolve these issues, the Gramineae (GRAMI)-rice model, which utilizes remote sensing data, was used in an effort to combine the complementary techniques of remote sensing and crop modeling. The model was then investigated for its capability to monitor canopy growth and estimate the grain yield of rice (Oryza sativa), at both the field and the regional scales, by using remote sensing images with high spatial resolution. The field scale investigation was performed using unmanned aerial vehicle (UAV) images, and the regional-scale investigation was performed using RapidEye satellite images. Simulated grain yields at the field scale were not significantly different (p = 0.45, p = 0.27, and p = 0.52) from the corresponding measured grain yields according to paired t-tests (α = 0.05). The model’s projections of grain yield at the regional scale represented the spatial grain yield variation of the corresponding field conditions to within ±1 standard deviation. Therefore, based on mapping the growth and grain yield of rice at both field and regional scales of interest within coverages of a UAV or the RapidEye satellite, our results demonstrate the applicability of the GRAMI-rice model to the monitoring and prediction of rice growth and grain yield at different spatial scales. In addition, the GRAMI-rice model is capable of reproducing seasonal variations in rice growth and grain yield at different spatial scales.     

Remote Sensing-Based Yield Forecasting for Sugarcane (<Emphasis Type="Italic">Saccharum officinarum</Emphasis> L.) Crop in India

Early yield assessment at local, regional and national scales is a major requirement for various users such as agriculture planners, policy makers, crop insurance companies and researchers. This current study explored a remote sensing-based approach of predicting sugarcane yield, at district level, using Vegetation Condition Index (VCI), under the FASAL programme of the Ministry of Agriculture & Farmers’ Welfare. 13-years’ historical database (2003–2015) of NDVI was used to derive the VCI. NDVI products (MOD-13A2) of MODIS instrument on board Terra satellite at 16-day interval from first fortnight of June to second fortnight of October (peak growing period) were used to calculate the VCI. Stepwise regression technique was used to develop empirical models between VCI and historical yield of sugarcane over 52 major sugarcane-growing districts in five states of India. For all the districts, the empirical models were found to be statistically significant. A large number of statistical parameters were computed to evaluate the performance of VCI-based models in predicting district-level sugarcane yield. Though there was variation in model performance in different states, overall, the study showed the usefulness of VCI, which can be used as an input for operational sugarcane yield forecasting.     

水稻冠层氮素含量光谱反演的随机森林算法及区域应用

利用地面实测数据构建高精度的水稻冠层氮素含量光谱反演点模型并将其进行尺度转换,实现了水稻冠层氮素含量准实时、大区域监测。以氮素光谱敏感指数作为输入变量,冠层氮素含量数据为输出变量,利用随机森林算法构建水稻冠层氮素含量高光谱反演模型,并用苏州市水稻农田验证区数据,检验模型的普适性和有效性;利用准同步的Hyperion数据,采用对输入、输出变量进行线性变换的简单尺度转换方法实现了点模型的区域应用。结果表明:基于随机森林算法的水稻冠层氮素含量高光谱反演模型可解释、所需样本少、不会过拟合、精度高(模型在实验区的预测精度为R2=0.82,验证区检验精度为R2=0.73)且具有普适性;点模型基于高光谱遥感卫星影像和尺度转换进行区域应用,精度较高(R2=0.81)。     

Automatic Detection of Rice Disease Using Near Infrared Spectra Technologies

The rice disease is one of the most serious injurious factors that cause major loss of rice production and subsequent economy in agricultural industry. This study explored a new method for obtaining information of the rice disease in a short term through model regression methods. The spectrum characteristics of rice leaves under different disease damage were firstly analyzed for its relationship with rice disease level. The sensitive bands of the spectrum for accurately supervising rice diseases were selected with principal component analysis (PCA). The stepwise regression method and BP neural network were both used to establish the spectrum-based models for recognizing rice diseases. Results showed that five major characteristic bands were determined by PCA (990, 1850, 660, 1921, and 1933 nm) for monitoring foliar rice diseases, among which the edge area for red light had the best correlation with rice disease level was also selected as the parameter to establish the model. Specifically, the composite reflectivity of wavelengths between 990 and 1933 nm was negatively related to rice brown spot diseases stress, which was then used to establish the model. Parameters of the red edge area and the ranged reflectivity between 660 and 990 nm were used to establish models for monitoring rice sheath blight diseases. Totally, there were 60 samples employed to build models for identifying the two diseases by the stepwise regression method and the BP neural network method, and the rest 41 ones were used for further model verification. Compared with the stepwise regression analysis, BP neural network was evaluated to perform better with characteristic bands at 660, 990, and 1933 nm. In conclusion, the establishment of the function model in our study can be implemented to monitor rice diseases, which provided a theoretical basis for indirect and rapid monitoring rice diseases.     

一种多因子加权平均温度模型的构建方法

针对在地基GNSS水汽反演的过程中,天顶湿延迟转换为大气可降水量时如何建立精确的大气加权平均温度(Tm)模型的问题,该文在建立Tm模型前全面考虑了对Tm有显著影响的变量并选择最优回归子集。但分析发现,最优回归子集中各变量之间存在较强的相关性,这将会导致变量之间存在多重共线性,从而影响模型的稳定性和可靠性。选择2013—2015年相关气象数据作为变量并应用岭回归的方法削弱变量之间的多重共线性,建立稳定的多因子Tm回归模型。并利用该模型分别预测2016年1—12月、2019年1—7月的Tm,均方根误差分别为2.3 K和2.0 K,预测精度较高,这将为高精度的水汽反演奠定较好的数据基础。     

Sentinel-2卫星落叶松林龄信息反演

林龄结构信息能够有效反映区域森林群落不同生长阶段的固碳能力,对于评估森林生态系统的健康状况具有重要意义。本研究以中国温带典型优势树种落叶松林为研究对象,分别选择其芽萌动期、展叶期和落叶期时段的Sentinel-2影像,采用多元线性回归（MLR）、随机森林（RF）、支持向量机回归（SVR）、前馈反向传播神经网络（BP）以及多元自适应回归样条（MARS）等5种方法依次构建落叶松林龄反演模型。通过相关性分析首先确定最佳遥感反演物候期,并在此基础上根据相关性差异筛选出5个最优特征变量用于模型反演,分别为冠层含水量（CWC）,归一化水体指数（NDWI）,叶面积指数（LAI）,光合有效辐射吸收率（FAPAR）和植被覆盖度（FVC）。研究结果表明,展叶期为落叶松林最佳遥感反演物候期。除植被衰减指数（PSRI）以及落叶期的NDVI、RVI外,落叶松林龄与各指标之间均呈负相关关系,其中与冠层含水量（CWC）的相关性最高,pearson相关系数达到-0.74（p<0.01）。此外,不同模型反演结果表明,随机森林模型（RF）为最佳落叶松林龄估测模型,其平均决定系数R²和平均均方根误差RMSE分别为0.89和2.91 a;多元线性回归模型（MLR）的林龄估测结果最差,其平均决定系数R²和平均均方根误差RMSE仅为0.57和5.69 a,非线性模型能更好的解释林龄与建模变量之间的关系。     

Mapping seasonal rice cropland extent and area in the high cropping intensity environment of Bangladesh using MODIS 500 m data for the year 2010

Rice is the most consumed staple food in the world and a key crop for food security. Much of the world’s rice is produced and consumed in Asia where cropping intensity is often greater than 100% (more than one crop per year), yet this intensity is not sufficiently represented in many land use products. Agricultural practices and investments vary by season due to the different challenges faced, such as drought, salinity, or flooding, and the different requirements such as varietal choice, water source, inputs, and crop establishment methods. Thus, spatial and temporal information on the seasonal extent of rice is an important input to decision making related to increased agricultural productivity and the sustainable use of limited natural resources. The goal of this study was to demonstrate that hyper temporal moderate-resolution imaging spectroradiometer (MODIS) data can be used to map the spatial distribution of the seasonal rice crop extent and area. The study was conducted in Bangladesh where rice can be cropped once, twice, or three times a year.MODIS normalized difference vegetation index (NDVI) maximum value composite (MVC) data at 500 m resolution along with seasonal field-plot information from year 2010 were used to map rice crop extent and area for three seasons, boro (December/January–April), aus (April/May–June/July), and aman (July/August–November/December), in Bangladesh. A subset of the field-plot information was used to assess the pixel-level accuracy of the MODIS-derived rice area. Seasonal district-level rice area statistics were used to assess the accuracy of the rice area estimates. When compared to field-plot data, the maps of rice versus non-rice exceeded 90% accuracy in all three seasons and the accuracy of the five rice classes varied from 78% to 90% across the three seasons. On average, the MODIS-derived rice area estimates were 6% higher than the sub-national statistics during boro, 7% higher during aus, and 3% higher during the aman season. The MODIS-derived sub-national areas explained (R² values) 96%, 93%, and 96% of the variability at the district level for boro, aus, and aman seasons, respectively.The results demonstrated that the methods we applied for analysing and interpreting moderate spatial and high temporal resolution imagery can accurately capture the seasonal variability in rice crop extent and area. We discuss the robustness of the approach and highlight issues that must be addressed before similar methods are used across other areas of Asia where a mix of rainfed, irrigated, or supplemental irrigation permits single, double, and triple cropping in a single calendar year.     

基于冠层反射光谱的水稻产量预测模型

基于地面实测的水稻冠层反射光谱,计算了常用的8个植被指数,并在产量形成生理特征的基础上,系统分析了水稻籽粒产量及其构成因素与各植被指数之间的关系。结果表明,通过单一生育时期或某个生育阶段的光谱植被指数来直接估测产量精度较低。发现叶面积氮指数(叶片氮百分含量与叶面积指数的乘积)的变化趋势很好地反映了产量的形成过程,且与光谱植被指数极显著正相关,基于此建立了水稻的光谱植被指数-累积叶面积氮指数-产量估测模型(VICLANIYieldModel)。并将其与LAD-产量模型、多生育期复合估产模型进行了比较,表明本模型预测精度最高。     

Empirical regression models using NDVI, rainfall and temperature data for the early prediction of wheat grain yields in Morocco

In Morocco, no operational system actually exists for the early prediction of the grain yields of wheat (Triticum aestivum L.). This study proposes empirical ordinary least squares regression models to forecast the yields at provincial and national levels. The predictions were based on dekadal (10-daily) NDVI/AVHRR, dekadal rainfall sums and average monthly air temperatures. The Global Land Cover raster map (GLC2000) was used to select only the NDVI pixels that are related to agricultural land. Provincial wheat yields were assessed with errors varying from 80 to 762 kg ha⁻¹, depending on the province. At national level, wheat yield was predicted at the third dekad of April with 73 kg ha⁻¹ error, using NDVI and rainfall. However, earlier forecasts are possible, starting from the second dekad of March with 84 kg ha⁻¹ error, at least 1 month before harvest. At the provincial and national levels, most of the yield variation was accounted for by NDVI. The proposed models can be used in an operational context to early forecast wheat yields in Morocco.     

Modelling monthly soil losses and sediment yields in Cyprus

The aim of this study was to map soil erosion on the Mediterranean island of Cyprus. The G2 model, an empirical model for month-time step erosion assessments, was used. Soil losses in Cyprus were mapped at a 100?m cell size, while sediment yields at a sub-basin scale of 0.62?km² mean size. The results indicated a mean annual erosion rate of 11.75?t?ha^?1?y^?1, with October and November being the most erosive months. The 34% of the island's surface was found to exceed non-sustainable erosion rates (>10?t?ha^?1?y^?1), with sclerophyllous vegetation, coniferous forests, and non-irrigated arable land being the most extensive non-sustainable erosive land covers. The mean sediment delivery ratio (SDR) was found to be 0.26, while the mean annual specific sediment yield (SSY) value for Cyprus was found to be 3.32?t?ha^?1?y^?1. The annual sediment yield of the entire island was found to be 2.746?Mt?y^?1. This study was the first to provide complete and detailed erosion figures for Cyprus at a country scale. The geodatabase and all information records of the study are available at the European Soil Data Centre (ESDAC) of the Joint Research Centre (JRC).