Precise yield predictions are useful for implementing precision agriculture technologies and making better decisions in crop management. Convolutional neural networks (CNNs) have recently been used to predict crop yields in unmanned aerial vehicle (UAV)-based remote sensing studies, but weather data have not been considered in modeling. The aim of this study was to explore the potential of multimodal deep learning on rice yield prediction accuracy using UAV multispectral images at the heading stage, along with weather data. The effects of the CNN architectures, layer depths, and weather data integration methods on the prediction accuracy were evaluated. Overall, the multimodal deep learning model integrating UAV-based multispectral imagery and weather data had the potential to develop more precise rice yield predictions. The best models were those trained with weekly weather data. A simple CNN feature extractor for UAV-based multispectral image input data might be sufficient to predict crop yields accurately. However, the spatial patterns of the predicted yield maps differed from model to model, although the prediction accuracy was almost the same. The results indicated that not only the prediction accuracies, but also the robustness of within-field yield predictions, should be assessed in further studies.

Timely and nondestructive monitoring of leaf area index (LAI) using remote sensing techniques is crucial for precise and efficient management of crops. In this paper, a new spectral index (SI) for estimating LAI of winter wheat (Triticum aestivum L.) is proposed on the basis of field hyperspectral measurements. A simple index based on the empirical relationships between LAIs and SIs of all available two-waveband combinations from hyperspectral data is developed by considering the difference between reflectance values at 760 and 739 nm (DSI_R760–R739 = R₇₆₀ – R₇₃₉). Among published and newly developed SIs, DSI_R760–R739 exhibited a significant and strong linear relationship with LAI and showed outstanding performance in LAI assessments. The permissible bandwidths for broad-band DSI_R760–R739 investigated using simulated reflectance were 5 nm for both 760 and 739 nm center wavelengths. The results indicate that the linear regression model based on the narrow-band and broad-band DSI_R760–R739 is a simple but accurate method for timely and nondestructive monitoring of LAI.

Reflectance and vegetation indices obtained from aerial images are often used for monitoring crop fields. In recent years, unmanned aerial vehicles (UAVs) have become popular and aerial images have been collected under various solar radiation conditions. The value of observed reflectance and vegetation indices are considered to be affected by solar radiation conditions, which may lead to inaccurate estimations of crop growth. In this study, in order to evaluate the effect of solar radiation conditions on aerial images, canopy reflectance in paddy fields was simulated by a radiative transfer model, FLiES (Forest Light Environmental Simulator), for various solar radiation conditions and canopy structures. Several parameters including solar zenith angle, proportion of diffuse light for incident sunlight, plant height, coordinates of plants and leaf area density, were tested in FLiES. The simulation results showed that the solar zenith angle did not vary the canopy reflectance under the conditions of the proportion of diffuse light at 1.0, but the variation was greater at lower proportions of diffuse light. The difference in reflectance caused by solar radiation was 0.01 and 0.1 at the maximum for red and near-infrared bands, respectively. The simulation results also showed that the differences in reflectance affect vegetation indices (Simple Ratio (SR), Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index 2 (EVI2)). The variation caused by solar radiation conditions was the least for NDVI and the greatest for SR. However, NDVI was saturated at the least leaf area index (LAI), whereas SR was only slightly saturated. EVI2 was intermediate between SR and NDVI, both in terms of variation and saturation. The simulated reflectance and vegetation indices were similar to those obtained from the aerial images collected in the farmers’ paddy fields. These results suggest that a large proportion of diffuse light (close to 1.0) or a middle range of solar zenith angle (45 to 65 degrees) may be desirable for UAV monitoring. However, to maintain flexibility of time and occasion for UAV monitoring, EVI2 should be used to evaluate crop growth, although calibration based on solar radiation conditions is recommended.

Northeast Thailand is the largest rice cultivation region in Thailand, but the rice yield there is quite low. Soil salinity is one of the major yield restricted factors, is derived from underground rock salt, and is predicted to expand in the future. This study focused on evaluating rice productivity related to salinity conditions in Khon Kaen Province, Northeast Thailand. The field investigations were conducted from 2017 to 2019 in farmer fields in severe, moderate, and slight soil salinity classes determined by the Land Development Department of Thailand. The soil salinity on the basis of the electric conductivity of saturated soil extract (ECe) varied year to year, which seemed to be associated with precipitation. The difference in soil salinity between classes was obvious only in the drought year 2018, and reflected in the rice yield, although severe drought devastated rice yield in some fields. Plenty of rainfall may have alleviated soil salinity and rice yield reduction in other years, causing differences in rice yield that were not significant among soil salinity classes. However, salinity level evaluation by the USDA based on ECe showed that rice yield was damaged depending on the level. This study indicates that ECe-based evaluation is recommended for soil salinity in relation to rice productivity. The spatial and temporal evaluation for rice production may benefit farmers. The results in this study also showed rice production largely varied even in similar salinity levels, implying that salinity damage can be alleviated by farmer management.

Advances in remote sensing technologies have enabled effective drought monitoring globally, even in data-limited areas. However, the negative impact of drought on crop yields still necessitates stakeholders to make informed decisions according to its severity. This research proposes an algorithm to combine a drought monitoring model, based on rainfall, land surface temperature (LST), and normalized difference vegetation index/leaf area index (NDVI/LAI) satellite products, with a crop simulation model to assess drought impact on rice yields in Thailand. Typical crop simulation models can provide yield information, but the requirement for a complicated set of inputs prohibits their potential due to insufficient data. This work utilizes a rice crop simulation model called the Simulation Model for Use with Remote Sensing (SIMRIW–RS), whose inputs can mostly be satisfied by such satellite products. Based on experimental data collected during the 2018/19 crop seasons, this approach can successfully provide a drought monitoring function as well as effectively estimate the rice yield with mean absolute percentage error (MAPE) around 5%. In addition, we show that SIMRIW–RS can reasonably predict the rice yield when historical weather data is available. In effect, this research contributes a methodology to assess the drought impact on rice yields on a farm to regional scale, relevant to crop insurance and adaptation schemes to mitigate climate change.

Drought is increasingly threatening smallholder farmers in Southeast Asia. The crop insurance system is one of the promising countermeasures that was implemented in Indonesia in 2015. Because the damage assessment in the present system is conducted through direct investigations based on appearance, it is not objective and needs a long time to cover large areas. In this study, we investigated a rapid assessment method for paddy fields using a vegetation index (VI) taken by an unmanned aerial vehicle (UAV) with a multispectral camera in 2019 and 2021. Then, two ways of assessment for drought damage were tested: linear regression (LR) based on a visually assessed drought level (DL), and k-means clustering without an assessed DL. As a result, EVI2 could represent the damage level, showing the tendency of the decrease in the value along with the increasing DL. The estimated DL by both methods mostly coincided with the assessed DL, but the concordance rates varied depending on the locations and the number of assessed fields. Differences in the growth stage and rice cultivars also affected the results. This study revealed the feasibility of the UAV-based rapid and objective assessment method. Further data collection and analysis would be required for implementation in the future.