Abstract:
The vegetation of a river basin is affected by various climate factors, such as precipitation and land surface temperature (LST). This study explores the best machine learning model for the prediction of normalized difference vegetation index (NDVI) with LST and precipitation as input parameters. The study also determines the correlation between NDVI, LST, and precipitation of the Mahanadi basin from 2003 to 2021. Monthly precipitation data was extracted from the Center for Hydrometeorology and Remote Sensing (CHRS) portal. The Moderate Resolution Imaging Spectroradiometer (MODIS) products were used to derive the LST and NDVI using Google Earth Engine (GEE). Four different machine learning models were used to predict the NDVI of the Mahanadi basin: linear regression (LR), random forest (RF), support vector regression (SVR), and k-nearest neighbors (KNN). The coefficient of determination (R2), root mean square error (RMSE), mean square error (MSE), mean absolute error (MAE), and explained variance score (EVS) were calculated to evaluate the performance of the models. The results show that the RF model has the highest R2 value in both the training and testing sets among these models, indicating that it is the most optimal among these models for predicting NDVI. The SVR model has the lowest RMSE value in the training set, but the KNN model has the lowest RMSE value in the testing set. The results also show that there is a positive correlation between precipitation and NDVI, a negative correlation between precipitation and LST, and between NDVI and LST. This study provides insights into the relationship between NDVI, LST, and precipitation, and the best machine-learning model for predicting NDVI. The findings of this study can be used to improve the management of river basins and to predict the effects of climate change on vegetation.
摘要:
流域植被受各种气候因素的影响,如降水和地表温度(LST)。本研究探索了以LST和降水量为输入参数的归一化差异植被指数(NDVI)预测的最佳机器学习模型。该研究还确定了NDVI之间的相关性,LST,和2003年至2021年马哈纳迪盆地的降水。从水文气象与遥感中心(CHRS)门户提取了月降水数据。使用中分辨率成像光谱仪(MODIS)产品使用GoogleEarthEngine(GEE)导出LST和NDVI。使用四种不同的机器学习模型来预测马哈纳迪盆地的NDVI:线性回归(LR),随机森林(RF),支持向量回归(SVR),和k-最近邻(KNN)。决定系数(R2),均方根误差(RMSE),均方误差(MSE),平均绝对误差(MAE),并计算解释方差得分(EVS)来评估模型的性能。结果表明,在这些模型中,RF模型在训练集和测试集中都具有最高的R2值,表明它是预测NDVI的最佳模型。SVR模型在训练集中具有最低的RMSE值,但KNN模型在测试集中具有最低的RMSE值。结果还表明,降水与NDVI之间存在正相关关系,降水量和LST之间呈负相关,在NDVI和LST之间。这项研究提供了对NDVI之间关系的见解,LST,和降水,以及预测NDVI的最佳机器学习模型。这项研究的结果可用于改善流域管理并预测气候变化对植被的影响。