Artificial intelligence analysis of hyperspectral remote sensing data for management of water, weed, and nitrogen stresses in corn fields
Bibliographic record
Abstract
This study investigated the possibility of using ground-based remotely sensed hyperspectral observations with a special emphasis on detection of water, weed and nitrogen stresses contributing towards in-season decision support for precision crop management (PCM). A three factor split-split-plot experiment, with four randomized blocks as replicates, was established during the growing seasons of 2003 and 2004. Corn (Zea mays L.) hybrid DKC42-22 was grown because this hybrid is a good performer on light soils in Quebec. There were twelve 12 x 12m plots in a block (one replication per treatment per block) and the total number of plots was 48. Water stress was the main factor in the experiment. A drip irrigation system was laid out and each block was split into irrigated and non-irrigated halves. The second main factor of the experiment was weeds with two levels i.e. full weed control and no weed control. Weed treatments were assigned randomly by further splitting the irrigated and non-irrigated sub-blocks into two halves. Each of the weed treatments was furthermore split into three equal sub-sub-plots for nitrogen treatments (third factor of the experiment). Nitrogen was applied at three levels i.e. 50, 150 and 250 kg N ha-1 (Quebec norm is between 120-160 kg N ha-1). The hyperspectral data were recorded (spectral resolution = 1 nm) mid-day (between 1000 and 1400 hours) with a FieldSpec FR spectroradiometer over a spectral range of 400-2500 run at three growth stages namely: early growth, tasseling and full maturity, in each of the growing season. There are two major original contributions in this thesis: First is the development of a hyperspectral data analysis procedure for separating visible (400-700 nm), near-infrared (700-1300 nm) and mid-infrared (1300-2500 nm) regions of the spectrum for use in discriminant analysis procedure. In addition, of all the spectral band-widths analyzed, seven waveband-aggregates were identified using STEPDISC procedure, which were the most effective for classifying combined water, weed, and nitrogen stress. The second contribution is the successful classification of hyperspectral observations acquired over an agricultural field, using three innovative artificial intelligence approaches; support vector machines (SVM), genetic algorithms (GA) and decision tree (DT) algorithms. These AI approaches were used to evaluate a combined effect of water, weed and nitrogen stresses in corn and of all the three AI approaches used, SVM produced the best results (overall accuracy ranging from 88% to 100%). The general conclusion is that the conventional statistical and artificial intelligence techniques used in this study are all useful for quickly mapping combined affects of irrigation, weed and nitrogen stresses (with overall accuracies ranging from 76% to 100%). These approaches have strong potential and are of great benefit to those investigating the in-season impact of irrigation, weed and nitrogen management for corn crop production and other environment related challenges.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".