Nd RMSE was involving 0.105 and 0.191. Similarly, NDVI can't substantially boost theNd RMSE was

Nd RMSE was involving 0.105 and 0.191. Similarly, NDVI can’t substantially boost the
Nd RMSE was between 0.105 and 0.191. Similarly, NDVI can not drastically enhance the Carboxypeptidase M Proteins Recombinant Proteins accuracy of mercury components. For the PLSR prediction models of As and Hg components, both are E3 Ligases Proteins web logarithmically calculated by spectral aspects as input variables with larger model accuracy than spectral bands. The target heavy metal prediction model established by spectral factors LnB1 LnB4 and NDVI had the highest accuracy.Table four. The outcomes of partial least square regression (PLSR) between heavy metal concentrations and spectrum indicators. Modeling Aspects B1 4 B1 4 NDVI LnB1 nB4 LnB1 nB4 NDVI B1 four LnB1 nB4 B1 four B1 four NDVI LnB1 nB4 LnB1 nB4 NDVI B1 four LnB1 nB4 Modeling Set R 0.431 0.432 0.460 0.462 0.446 0.257 0.263 0.259 0.268 0.260 RMSE 1.976 1.976 1.945 1.943 1.961 0.062 0.062 0.062 0.066 0.062 Verification Set R 0.502 0.498 0.524 0.526 0.536 0.155 0.149 0.155 0.161 0.152 RMSE 2.045 two.048 2.009 two.007 1.999 0.105 0.125 0.191 0.105 0.AsHgAs shown in Table 5, based around the BP model, for the As modeling set, R ranged from 0.482 to 0.530, and RMSE was 1.860 1.909; for the verification set, R was 0.467 0.532, and RMSE was 1.999 to 2.094. For the Hg element modeling set, R was among 0.263 and 0.318, and RMSE was in between 0.061 and 0.062; the verification set was in between 0.149 and 0.186, and RMSE was among 0.105 and 0.288. Compared together with the five PLSR models, theLand 2021, ten, x FOR PEER REVIEWLand 2021, 10,9 of9 ofand RMSE was among 0.105 and 0.288. Compared together with the 5 PLSR models, the correlation among the BP model in the target heavy metal content was correspondingly improved, as well as the accuracy was reasonably higher. heavy metal content was correspondingly correlation in between the BP model from the target The larger the choice coefficient plus the enhanced, and the accuracy was reasonably higher. smaller sized the root mean square error, the much more stable and accurate the model is. It may be concluded that the model together with the highest Table five. The outcomes of backBP model established by the B1 B4 spectral factor, R = 0.530; the accuracy of As was the propagation neural network (BPNN) in between heavy metal concentrations and spectrum indicators. model with all the highest accuracy of Hg was the BP model based on B1 B4 and NDVI spectral characteristic, R = 0.318. For the As element, the relative error of modeling was 0.201, Modeling and for the Hg element, theFactors error was 0.498. The Set model Verification Set can PLSR and BP model Modeling relative R RMSE R RMSE establish the target metal content and spectral reflection element to predict the metal content of the study region. It canB1 four be shown in the evaluation parameters in the model two.048 the that 0.530 1.860 0.507 B1 four capability 0.513 1.865 0.532 1.999 modeling and prediction NDVIof the BP model was higher, and it had a great interpretaLnB1 nB4 0.519 1.874 0.467 two.097 tionAs ability on the target soil heavy metals.LnB1 nB4 NDVI 0.482 1.870 0.499 2.054 B1 4 LnB1 nB4 neural network (BPNN) in between heavy metal concentra0.497 1.909 0.525 two.006 Table 5. The outcomes of back propagation B1 4 0.273 0.062 0.149 0.105 tions and spectrum indicators. B1 four NDVI 0.318 0.062 0.177 0.105 Modeling 0.061 Set Verification Set Hg LnB1 nB4 0.263 0.163 0.105 Modeling Components LnB1 nB4 NDVI 0.269 0.062 0.156 0.288 R RMSE R RMSE B1 four LnB1 nB4 0.292 0.061 0.186 0.105 B1 four 0.530 1.860 0.507 2.B1 4 NDVI 0.513 1.865 0.532 1.999 Because the bigger the selection coefficient and also the smaller the root imply square error, the LnB1 nB4 0.519 1.874 0.467 two.097.