Suitability of satellite remote sensing data for yield estimation in northeast Germany

Abstract

Information provided by satellite data is becoming increasingly important in the field of agriculture. Estimating biomass, nitrogen content or crop yield can improve farm management and optimize precision agriculture applications. A vast amount of data is made available both as map material and from space. However, it is up to the user to select the appropriate data for a particular problem. Without the appropriate knowledge, this may even entail an economic risk. This study therefore investigates the direct relationship between satellite data from six different optical sensors as well as different soil and relief parameters and yield data from cereal and canola recorded by the thresher in the field. A time series of 13 years is considered, with 947 yield data sets consisting of dense point data sets and 755 satellite images. To answer the question of how well the relationship between remote sensing data and yield is, the correlation coefficient r per field is calculated and interpreted in terms of crop type, phenology, and sensor characteristics. The correlation value r is particularly high when a field and its crop are spatially heterogeneous and when the correct phenological time of the crop is reached at the time of satellite imaging. Satellite images with higher resolution, such as RapidEye and Sentinel-2 performed better in comparison with lower resolution sensors of the Landsat series. The additional Red Edge spectral band also has advantage, especially for cereal yield estimation. The study concludes that there are high correlation values between yield data and satellite data, but several conditions must be met which are presented and discussed here.

Predicting Crop Yield Using Deep Learning and Remote Sensing

“…When compared to other methods of data acquisition, such as field surveying, remote sensing has the capacity to create a sufficient amount of data. It is the technique of seeing and identifying locations on Earth by utilizing sensors to measure the radiation that is emitted and reflected [67][68][69][70][71][72]. An important justification for using optical remote sensing to obtain agricultural data is the measurement of vegetation indices.…”

Collecting and Disseminating Ground Truth Data for Generation of Agriculture Statistics

“…In addition, agriculture remains one of the most researched EO data applications, with exponential growth occurring since 2013 [3]. Some of the most notable agricultural applications include crop areas and monitoring [4][5][6][7][8], yield estimates [9][10][11], irrigation mapping [12][13][14], and crop loss assessment [15][16].…”

Wheat Yield Prediction based on Sentinel-2, Regression and Machine Learning Models in Hamedan, Iran

“…The used aerial data collection system was not recommended due to the short flight time, huge post-processing, etc. In [16], satellite images with higher resolution, such as RapidEye and Sentinel-2 performed better in comparison with lower resolution sensors of the Landsat series for yield (i.e. cereal and canola) prediction.…”