Machine-learning methods to assess the effects of a non-linear damage spectrum taking into account soil moisture on winter wheat yields in Germany

Agricultural production is highly dependent on the weather. The mechanisms of action are complex and interwoven, making it difficult to identify relevant management and adaptation options. The present study uses random forests to investigate such highly non-linear systems for predicting yield anomalies in winter wheat at district levels in Germany. In order to take into account sub-seasonality, monthly features are used that explicitly take soil moisture into account in addition to extreme meteorological events. Clustering is used to show spatially different damage potentials, such as a higher susceptibility to drought damage from May to July in eastern Germany compared to the rest of the country. In addition, relevant heat effects are not detected if the clusters are not sufficiently defined. The variable with the highest importance is soil moisture in March, where higher soil moisture has a detrimental effect on crop yields. In general, soil moisture explains more yield variations than the meteorological variables. The approach has proven to be suitable for explaining historical extreme yield anomalies for years with exceptionally high losses (2003, 2018) and gains (2014) and the spatial distribution of these anomalies. The highest test R-squared (R2) is about 0.68. Furthermore, the sensitivity of yield variations to soil moisture and extreme meteorological conditions, as shown by the visualization of average marginal effects, contributes to the promotion of targeted decision support systems.

Prediction of Drought-Induced Components and Evaluation of Drought Damage of Tea Plants Based on Hyperspectral Imaging

Effective evaluation of physiological and biochemical indexes and drought degree of tea plant is an important technology to determine the drought resistance ability of tea plants. At present, the traditional detection method of tea drought stress is mainly based on physiological and biochemical detection, which is not only destructive to tea plants, but also time-consuming and laborious. In this study, through simulating drought treatment of tea plant, hyperspectral camera was used to obtain spectral data of tea leaves, and three machine learning models, namely, support vector machine (SVM), random forest (RF), and partial least-squares (PLS) regression, were used to model malondialdehyde (MDA), electrolyte leakage (EL), maximum efficiency of photosystem II (Fv/Fm), soluble saccharide (SS), and drought damage degree (DDD) of tea leaves. The results showed that the competitive adaptive reweighted sampling (CARS)-PLS model of MDA had the best effect among the four physiological and biochemical indexes (Rcal = 0.96, Rp = 0.92, RPD = 3.51). Uninformative variable elimination (UVE)-SVM model was the best in DDD (Rcal = 0.97, Rp = 0.95, RPD = 4.28). Therefore, through the establishment of machine learning model using hyperspectral imaging technology, we can monitor the drought degree of tea seedlings under drought stress. This method is not only non-destructive, but also fast and accurate, which is expected to be widely used in tea garden water regime monitoring.

Machine learning methods to assess the effects of a non-linear damage spectrum taking into account soil moisture on winter wheat yields in Germany

Agricultural production is highly dependent on the weather. The mechanisms of action are complex and interwoven, making it difficult to identify relevant management and adaptation options. The present study uses random forests to investigate such highly non-linear systems for predicting yield anomalies in winter wheat at district level in Germany. In order to take into account sub-seasonality, monthly features are used that explicitly take soil moisture into account in addition to extreme meteorological events. Clustering is used to show spatially different damage potentials, such as a higher susceptibility to drought damage from April to July in eastern Germany compared to the rest of the country. The variable that explains most differences is soil moisture in March, where higher soil moisture has a detrimental effect on crop yields. In general, soil moisture explains more yield variations than the meteorological variables, while the top 25 cm of soil moisture is a better yield predictor than the total soil column. The approach has proven to be suitable to explain historical extreme yield anomalies for years with exceptionally high losses (2003, 2018) and gains (2014) and the spatial distribution of these anomalies. The highest test R-square is about 0.70. Furthermore, the sensitivity of yield variations to soil moisture and extreme meteorological conditions, as shown by the visualisation of average marginal effects, contributes to the promotion of targeted decision support systems.

A Study on Establishment of Drought Index for Agricultural Disaster Prediction

This study aims to establish a drought index for disaster prediction in Gyeongsangnam-do, where the most agricultural drought damage occurred from 1965 to 2018. The drought index was analyzed for each duration (3, 6, 9, 12 months) targeting the SPI. Damage characteristics of the duration of agricultural drought were calculated. SPI for each duration of agricultural drought damage period in Gyeongsangnam-do was at least -2.0 or less, and the maximum was -1.0 or more, and weak and moderate drought were analyzed. However, due to the heavy rain effect during the rainy season, the average SPI12 was -1.06, and the impact of agricultural drought was negligible. It was analyzed that the correlation between the damage period of agricultural drought and the SPI by duration was high. However, there is not much difference in SPI for each duration to determine the occurrence of damage. In this study, the criterion for disaster prediction of agricultural drought was calculated as representative drought index by year as the minimum drought index of SPI for each duration of damage occurrence period of past agricultural drought. The Standard of drought index for disaster prediction was set to -1.64, the average of the SPI for each duration of year in which damage occurred in the past.

Assessment of Quantitative Standards for Mega-Drought Using Data on Drought Damages

Among natural disasters, droughts can affect a large area for a prolonged period of time. If a drought happens, an appropriate response requires a lot of time and manpower from beginning to end, and continuous management is necessary for further prevention. Using data on drought damages from 1900 to 2018 in 148 countries in six continents around the world, this study was able to set quantitative standards for mega-droughts. According to data on the status of annual drought damages, the frequency of drought damages (1900–2018) and the subsequent damage costs (1965–2018) are increasing, while human losses (1900–2018) are decreasing. Additionally, Africa had the highest frequency of drought damages, while Africa and Asia were ranked at the top of the list in terms of human losses and damage costs, respectively. Droughts persisted for continuous periods ranging from 1 to 17 years, and the total number of cases involving drought damage was estimated to be 600 in total, with total human losses of 11,731,294 people and total accumulated damage costs of $17,367,007,000. This study provided quantitative standards for the frequency of drought damages, human losses, and damage costs for mega-droughts in consideration of continuous drought periods. This study set the quantitative standards for a mega-drought as follows: (1) if drought damages continue to occur in a country for more than seven years, (2) if human losses continue to occur in a country for more than seven years, and (3) if mean annual damage costs of $17,000,000 continue to occur in a country for more than seven years.