Evaluation of MODIS and Landsat multiband vegetation indices used for wheat yield estimation in irrigated Indus Basin

Long short term memory deep net performance on fused Planet-Scope and Sentinel-2 imagery for detection of agricultural crop

In view of the challenges faced by organizations and departments concerned with agricultural capacity observations, we collected In-Situ data consisting of diverse crops (More than 11 consumable vegetation types) in our pilot region of Harichand Charsadda, Khyber Pakhtunkhwa (KP), Pakistan. Our proposed Long Short-Term Memory based Deep Neural network model was trained for land cover land use statistics generation using the acquired ground truth data, for a synergy between Planet-Scope Dove and European Space Agency's Sentinel-2. Total of 4 bands from both sentinel-2 and planet scope including Red, Green, Near-Infrared (NIR) and Normalised Difference Vegetation Index (NDVI) were used for classification purpose. Using short temporal frame of Sentinel-2 comprising 5 date images, we propose an realistic and implementable procedure for generating accurate crop statistics using remote sensing. Our self collected data-set consists of a total number of 107,899 pixels which was further split into 70% and 30% for training and testing purpose of the model respectively. The collected data is in the shape of field parcels, which has been further split for training, validation and test sets, to avoid spatial auto-correlation. To ensure the quality and accuracy 15% of the training data was left out for validation purpose, and 15% for testing. Prediction was also performed on our trained model and visual analysis of the area from the image showed significant results. Further more a comparison between Sentinel-2 time series is performed separately from the fused Planet-Scope and Sentinel-2 time-series data sets. The results achieved shows a weighted average of 93% for Sentinel-2 time series and 97% for fused Planet-Scope and Sentinel-2 time series.

Novel Water Retention and Nutrient Management Technologies and Strategies Supporting Agricultural Water Management in Continental, Pannonian and Boreal Regions

Urgent water and food security challenges, particularly in continental and boreal regions, need to be addressed by initiatives such as the Horizon 2020-funded project WATer retention and nutrient recycling in soils and streams for improved AGRIcultural production (WATERAGRI). A new methodological framework for the sustainable management of various solutions resilient to climate change has been developed. The results indicate that the effect of the climate scenario is significantly different for peatlands and constructed wetlands. The findings also highlight that remote-sensing-based yield prediction models developed from vegetation indices have the potential to provide quantitative and timely information on crops for large regions or even at the local farm scale. Verification of remotely sensed data is one of the prerequisites for the proper utilization and understanding of data. Research shows that current serious game applications fall short due to challenges such as not clarifying the decision problem, the lack of use of decision quality indicators and limited use of gaming. Overall, WATERAGRI solutions improve water and food security by adapting agriculture to climate change, recycling nutrients and providing educational tools to the farming community. Farmers in small agricultural catchments benefit directly from WATERAGRI, but over the long-term, the general public does as well.

Functional Evaluation of Digital Soil Hydraulic Property Maps through Comparison of Simulated and Remotely Sensed Maize Canopy Cover

Soil maps can usefully serve in data scarce regions, for example for yield (gap) assessments using a crop simulation model. The soil property estimates’ contribution to inaccuracy and uncertainty can be functionally evaluated by comparing model results using the estimates as input against independent observations. We conducted a functional evaluation of digital maps of soil hydraulic properties of the Zambezi River Basin using a crop growth model AquaCrop. AquaCrop was run, alimented with local meteorological data, and with soil hydraulic properties derived from the digital maps of digital soil mapping (DSM) techniques, as opposed to estimations from the widely used Saxton and Rawls pedotransfer functions. The two simulated time series of canopy cover (CC) (AquaCrop-CC-DSM and AquaCrop-CC-Saxton), which were compared against canopy cover data derived from the remotely sensed Leaf Area Index (LAI) from the MODIS archive (MODIS-CC). A pairwise comparison of the time series resulted in a root mean squared error (RMSE) of 0.07 and a co-efficient of determination (R2) of 0.93 for AquaCrop-CC-DSM versus MODIS-CC, and an RMSE of 0.08 and R2 of 0.88 for AquaCrop-CC-Saxton versus MODIS-CC. In dry years, the AquaCrop-CC-DSM deviated less from the MODIS-CC than the AquaCrop-CC-Saxton (p < 0.001), although this difference was not significant in wet years. The functional evaluation showed that soil hydraulic property estimates based on digital soil mapping outperformed those based on Saxton and Rawls when used for simulating crop growth in dry years in the Zambezi River Basin. This study also shows the value of conducting a functional evaluation of estimated (static) soil hydraulic properties in terms of dynamic model output.

Improving the Forecasting of Winter Wheat Yields in Northern China with Machine Learning–Dynamical Hybrid Subseasonal-to-Seasonal Ensemble Prediction

Subseasonal-to-seasonal (S2S) prediction of winter wheat yields is crucial for farmers and decision-makers to reduce yield losses and ensure food security. Recently, numerous researchers have utilized machine learning (ML) methods to predict crop yield, using observational climate variables and satellite data. Meanwhile, some studies also illustrated the potential of state-of-the-art dynamical atmospheric prediction in crop yield forecasting. However, the potential of coupling both methods has not been fully explored. Herein, we aimed to establish a skilled ML–dynamical hybrid model for crop yield forecasting (MHCF v1.0), which hybridizes ML and a global dynamical atmospheric prediction system, and applied it to northern China at the S2S time scale. In this study, we adopted three mainstream machining learning algorithms (XGBoost, RF, and SVR) and the multiple linear regression (MLR) model, and three major datasets, including satellite data from MOD13C1, observational climate data from CRU, and S2S atmospheric prediction data from IAP CAS, used to predict winter wheat yield from 2005 to 2014, at the grid level. We found that, among the four models examined in this work, XGBoost reached the highest skill with the S2S prediction as inputs, scoring R2 of 0.85 and RMSE of 0.78 t/ha 3–4 months, leading the winter wheat harvest. Moreover, the results demonstrated that crop yield forecasting with S2S dynamical predictions generally outperforms that with observational climate data. Our findings highlighted that the coupling of ML and S2S dynamical atmospheric prediction provided a useful tool for yield forecasting, which could guide agricultural practices, policy-making and agricultural insurance.

Effects of fallow tillage on winter wheat yield and predictions under different precipitation types

In northern China, precipitation that is primarily concentrated during the fallow period is insufficient for the growth stage, creates a moisture shortage, and leads to low, unstable yields. Yield prediction in the early growth stages significantly informs field management decisions for winter wheat (Triticum aestivum L.). A 10-year field experiment carried out in the Loess Plateau area tested how three tillage practices (deep ploughing (DP), subsoiling (SS), and no tillage (NT)) influenced cultivation and yield across different fallow periods. The experiment used the random forest (RF) algorithm to construct a prediction model of yields and yield components. Our results revealed that tillage during the fallow period was more effective than NT in improving yield in dryland wheat. Under drought condition, DP during the fallow period achieved a higher yield than SS, especially in drought years; DP was 16% higher than SS. RF was deemed fit for yield prediction across different precipitation years. An RF model was developed using meteorological factors for fixed variables and soil water storage after tillage during a fallow period for a control variable. Small error values existed in the prediction yield, spike number, and grains number per spike. Additionally, the relative error of crop yield under fallow tillage (5.24%) was smaller than that of NT (6.49%). The prediction error of relative meteorological yield was minimum and optimal, indicating that the model is suitable to explain the influence of meteorological factors on yield.

Rice Yield Estimation Based on Vegetation Index and Florescence Spectral Information from UAV Hyperspectral Remote Sensing

Rice floret number per unit area as one of the key yield structure parameters is directly related to the final yield of rice. Previous studies paid little attention to the effect of the variations in vegetation indices (VIs) caused by rice flowering on rice yield estimation. Unmanned aerial vehicles (UAV) equipped with hyperspectral cameras can provide high spatial and temporal resolution remote sensing data about the rice canopy, providing possibilities for flowering monitoring. In this study, two consecutive years of rice field experiments were conducted to explore the performance of florescence spectral information in improving the accuracy of VIs-based models for yield estimates. First, the florescence ratio reflectance and florescence difference reflectance, as well as their first derivative reflectance, were defined and then their correlations with rice yield were evaluated. It was found that the florescence spectral information at the seventh day of rice flowering showed the highest correlation with the yield. The sensitive bands to yield were centered at 590 nm, 690 nm and 736 nm–748 nm, 760 nm–768 nm for the first derivative florescence difference reflectance, and 704 nm–760 nm for the first derivative florescence ratio reflectance. The florescence ratio index (FRI) and florescence difference index (FDI) were developed and their abilities to improve the estimation accuracy of models basing on vegetation indices at single-, two- and three-growth stages were tested. With the introduction of florescence spectral information, the single-growth VI-based model produced the most obvious improvement in estimation accuracy, with the coefficient of determination (R2) increasing from 0.748 to 0.799, and the mean absolute percentage error (MAPE) and the root mean squared error (RMSE) decreasing by 11.8% and 10.7%, respectively. Optimized by flowering information, the two-growth stage VIs-based model gave the best performance (R2 = 0.869, MAPE = 3.98%, RMSE = 396.02 kg/ha). These results showed that introducing florescence spectral information at the flowering stage into conventional VIs-based yield estimation models is helpful in improving rice yield estimation accuracy. The usefulness of florescence spectral information for yield estimation provides a new idea for the further development and improvement of the crop yield estimation method.

Estimating Wheat Grain Yield Using Sentinel-2 Imagery and Exploring Topographic Features and Rainfall Effects on Wheat Performance in Navarre, Spain

Reliable methods for estimating wheat grain yield before harvest could help improve farm management and, if applied on a regional level, also help identify spatial factors that influence yield. Regional grain yield can be estimated using conventional methods, but the typical process is complex and labor-intensive. Here we describe the development of a streamlined approach using publicly accessible agricultural data, field-level yield, and remote sensing data from Sentinel-2 satellite to estimate regional wheat grain yield. We validated our method on wheat croplands in Navarre in northern Spain, which features heterogeneous topography and rainfall. First, this study developed stepwise multilinear equations to estimate grain yield based on various vegetation indices, which were measured at various phenological stages in order to determine the optimal timings. Second, the most suitable model was used to estimate grain yield in wheat parcels mapped from Sentinel-2 satellite images. We used a supervised pixel-based random forest classification and the estimates were compared to government-published post-harvest yield statistics. When tested, the model achieved an R2 of 0.83 in predicting grain yield at field level. The wheat parcels were mapped with an accuracy close to 86% for both overall accuracy and compared to official statistics. Third, the validated model was used to explore potential relationships of the mapped per-parcel grain yield estimation with topographic features and rainfall by using geographically weighted regressions. Topographic features and rainfall together accounted for an average for 11 to 20% of the observed spatial variation in grain yield in Navarre. These results highlight the ability of our method for estimating wheat grain yield before harvest and determining spatial factors that influence yield at the regional scale.

Improvement in Land Cover and Crop Classification based on Temporal Features Learning from Sentinel-2 Data Using Recurrent-Convolutional Neural Network (R-CNN)

Understanding the use of current land cover, along with monitoring change over time, is vital for agronomists and agricultural agencies responsible for land management. The increasing spatial and temporal resolution of globally available satellite images, such as provided by Sentinel-2, creates new possibilities for researchers to use freely available multi-spectral optical images, with decametric spatial resolution and more frequent revisits for remote sensing applications such as land cover and crop classification (LC&CC), agricultural monitoring and management, environment monitoring. Existing solutions dedicated to cropland mapping can be categorized based on per-pixel based and object-based. However, it is still challenging when more classes of agricultural crops are considered at a massive scale. In this paper, a novel and optimal deep learning model for pixel-based LC&CC is developed and implemented based on Recurrent Neural Networks (RNN) in combination with Convolutional Neural Networks (CNN) using multi-temporal sentinel-2 imagery of central north part of Italy, which has diverse agricultural system dominated by economic crop types. The proposed methodology is capable of automated feature extraction by learning time correlation of multiple images, which reduces manual feature engineering and modeling crop phenological stages. Fifteen classes, including major agricultural crops, were considered in this study. We also tested other widely used traditional machine learning algorithms for comparison such as support vector machine SVM, random forest (RF), Kernal SVM, and gradient boosting machine, also called XGBoost. The overall accuracy achieved by our proposed Pixel R-CNN was 96.5%, which showed considerable improvements in comparison with existing mainstream methods. This study showed that Pixel R-CNN based model offers a highly accurate way to assess and employ time-series data for multi-temporal classification tasks.

Spatial Heterogeneity of Winter Wheat Yield and Its Determinants in the Yellow River Delta, China

Understanding spatial differences of crop yields and quantitatively exploring the relationship between crop yields and influencing factors are of great significance in increasing regional crop yields, promoting sustainable development of regional agriculture and ensuring regional food security. This study investigates spatial heterogeneity of winter wheat yield and its determinants in the Yellow River Delta (YRD) region. The spatial pattern of winter wheat in 2015 was mapped through time series similarity analysis. Winter wheat yield was estimated by integrating phenological information into yield model, and cross-validation was performed using actual yield data. The geographical detector method was used to analyze determinants influencing winter wheat yield. This study concluded that the overall classification accuracy for winter wheat is 88.09%. The estimated yield agreed with actual yield, with R2 value of 0.74 and root mean square error (RMSE) of 1.02 t ha−1. Cumulative temperature, soil salinity and their interactions were key determinants affecting winter wheat yield. Several measures are recommended to ensure sustainable crop production in the YRD region, including improving irrigation and drainage systems to reduce soil salinity, selecting salt-tolerant winter wheat varieties, and improving agronomy techniques to extend effective cumulative temperature.

Assessment of the vegetation indices on Sentinel-2A images for predicting the soil productivity potential in Bursa, Turkey

Although field surveys represent an essential method for determining soil productivity, the use of remote sensing techniques has become a popular option over recent years due to its economic and practical applications. The fundamental basis of this approach is the estimation of soil productivity by using the vegetation indices as an indicator, with reference to the yield. In this study, it is aimed to estimate the productivity potential of the agriculture areas from biomass density in case of limited pedological and parcel-based data. For this purpose, relationships between the FAO Soil Productivity Rating (SPR) and different vegetation indices were investigated. The indices NDVI, RE-OSAVI, and REMCARI were used with Sentinel-2A images. Wheat was selected as an indicator plant to estimate the yield because it is the most occupied (27.47%) cultigen in the field. The study was conducted at the Karacabey State Farm with an area of 87 km² and is located in Bursa province, Turkey. The research showed a positive relationship between SPR and 2018 yield values (r² = 0.616). During the tillering period, the r² for RE-OSAVI was 0.629. In the heading stage, the r² for NDVI was 0.577. The index REMCARI provided yield estimations with low accuracy coefficient (0.216 ≤ r² ≤ 0.258) during all vegetation periods. These findings can be interpreted as the monitoring of the land quality with multispectral satellite images via NDVI and RE-OSAVI. In this way, we could decide the time to re-definition of soil properties with land surveys for determination of soil productivity when the detection of a decrease using the indices during some vegetation periods. However, further investigations are needed in controlled trial patterns with differential reference plants, although the findings obtained from the study are promising for the use of spectral vegetation indices to prediction and/or monitoring of soil productivity. Thus, the possibilities of using spectral indices in different ecologies and different plant species can be evaluated from a broad perspective. It was also suggested that Sentinel-2A images may be used for similar studies due to their spectral capabilities with the ESA-SNAP tool.

Determination of Appropriate Remote Sensing Indices for Spring Wheat Yield Estimation in Mongolia

In Mongolia, the monitoring and estimation of spring wheat yield at the regional and national levels are key issues for the agricultural policy and food management as well as for the economy and society as a whole. The remote sensing data and technique have been widely used for the estimation of crop yield and production in the world. For the current research, nine remote sensing indices were tested that include normalized difference drought index (NDDI), normalized difference water index (NDWI), vegetation condition index (VCI), temperature condition index (TCI), vegetation health index (VHI), normalized multi-band drought index (NMDI), visible and shortwave infrared drought index (VSDI), and vegetation supply water index (VSWI). These nine indices derived from MODIS/Terra satellite have so far not been used for crop yield prediction in Mongolia. The primary objective of this study was to determine the best remote sensing indices in order to develop an estimation model for spring wheat yield using correlation and regression method. The spring wheat yield data from the ground measurements of eight meteorological stations in Darkhan and Selenge provinces from 2000 to 2017 have been used. The data were collected during the period of the growing season (June–August). Based on the analysis, we constructed six models for spring wheat yield estimation. The results showed that the range of the root-mean-square error (RMSE) values of estimated spring wheat yield was between 4.1 (100 kg ha−1) to 4.8 (100 kg ha−1), respectively. The range of the mean absolute error (MAE) values was between 3.3 to 3.8 and the index of agreement (d) values was between 0.74 to 0.84, respectively. The conclusion was that the best model would be (R2 = 0.55) based on NDWI, VSDI, and NDVI out of the nine indices and could serve as the most effective predictor and reliable remote sensing indices for monitoring the spring wheat yield in the northern part of Mongolia. Our results showed that the best timing of yield prediction for spring wheat was around the end of June and the beginning of July, which is the flowering stage of spring wheat in this study area. This means an accurate yield prediction for spring wheat can be achieved two months before the harvest time using the regression model.

Field-Scale Rice Yield Estimation Using Sentinel-1A Synthetic Aperture Radar (SAR) Data in Coastal Saline Region of Jiangsu Province, China

In recent years, a large number of salterns have been converted into rice fields in the coastal region of Jiangsu Province, Eastern China. The high spatial heterogeneity of soil salinity has caused large within-field variabilities in grain yield of rice. The identification of low-yield areas within a field is an important initial step for precision farming. While optical satellite remote sensing can provide valuable information on crop growth and yield potential, the availability of cloud-free optical image data is often hampered by unfavorable weather conditions. Synthetic aperture radar (SAR) offers an alternative due to its nearly day-and-night and all-weather capability in data acquisition. Given the free data access of the Sentinels, this study aimed at developing a Sentinel-1A-based SAR index for rice yield estimation. The proposed SAR simple difference (SSD) index uses the change of the Sentinel-1A backscatter in vertical-horizontal (VH) polarization between the end of the tillering stage and the end of grain filling stage (SSDVH). A strong exponential relationship has been identified between the SSDVH and rice yield, producing accurate yield estimation with a root mean square error (RMSE) of 0.74 t ha−1 and a relative error (RE) of 7.93%.