Votre recherche
Résultats 5 ressources
-
The 2001–2012 MODIS MCD12Q1 land cover data and MOD17A3 NPP data were used to calculate changes in land cover in China and annual changes in net primary productivity (NPP) during a 12-year period and to quantitatively analyze the effects of land cover change on the NPP of China’s terrestrial ecosystems. The results revealed that during the study period, no changes in land cover type occurred in 7447.31 thousand km2 of China, while the area of vegetation cover increased by 160.97 thousand km2 in the rest of the country. Forest cover increased to 20.91%, which was mainly due to the conversion of large areas of savanna (345.19 thousand km2) and cropland (178.96 thousand km2) to forest. During the 12-year study period, the annual mean NPP of China was 2.70 PgC and increased by 0.25 PgC, from 2.50 to 2.75 PgC. Of this change, 0.21 PgC occurred in areas where there was no land cover change, while 0.04 PgC occurred in areas where there was land cover change. The contributions of forest and cropland to NPP exhibited increasing trends, while the contributions of shrubland and grassland to NPP decreased. Among these land cover types, the contributions of forest and cropland to the national NPP were the greatest, accounting for 40.97% and 27.95%, respectively, of the annual total NPP. There was no significant correlation between changes in forest area and changes in total annual NPP (R2 < 0.1), while the correlation coefficient for changes in cropland area and total annual NPP was 0.48. Additionally, the area of cropland converted to other land cover types was negatively correlated with the changes in NPP, and the loss of cropland caused a reduction in the national NPP.
-
Abstract Digital leaf physiognomy (DLP) is considered as one of the most promising methods for estimating past climate. However, current models built using the DLP data set still lack precision, especially for mean annual precipitation (MAP). To improve predictive power, we developed five machine learning (ML) models for mean annual temperature (MAT) and MAP respectively, and then tested the precision of these models and some of their averaging compared with that obtained from other models. The precision of all models was assessed using a repeated stratified 10‐fold cross‐validation. For MAT, three combinations of models ( R 2 = .77) presented moderate improvements in precision over the multiple linear regression (MLR) model ( R 2 = .68). For log e (MAP), the averaging of the support vector machine (SVM) and boosting models improved the R 2 from .19 to .63 compared with that of the MLR model. For MAP, the R 2 of this model combination was 0.49, which was much better than that of the artificial neural network (ANN) model ( R 2 = .21). Even the bagging model, which had the lowest R 2 (.37) for log e (MAP), demonstrated better precision ( R 2 = .27) for MAP. Our palaeoclimate estimates for nine fossil floras were also more accurate, because they were in better agreement with independent paleoclimate evidence. Our study confirms that our ML models and their averaging can improve paleoclimatic reconstructions, providing a better understanding of the relationship between climate and leaf physiognomy.
-
Abstract Digital leaf physiognomy (DLP) is considered as one of the most promising methods for estimating past climate. However, current models built using the DLP data set still lack precision, especially for mean annual precipitation (MAP). To improve predictive power, we developed five machine learning (ML) models for mean annual temperature (MAT) and MAP respectively, and then tested the precision of these models and some of their averaging compared with that obtained from other models. The precision of all models was assessed using a repeated stratified 10‐fold cross‐validation. For MAT, three combinations of models ( R 2 = .77) presented moderate improvements in precision over the multiple linear regression (MLR) model ( R 2 = .68). For log e (MAP), the averaging of the support vector machine (SVM) and boosting models improved the R 2 from .19 to .63 compared with that of the MLR model. For MAP, the R 2 of this model combination was 0.49, which was much better than that of the artificial neural network (ANN) model ( R 2 = .21). Even the bagging model, which had the lowest R 2 (.37) for log e (MAP), demonstrated better precision ( R 2 = .27) for MAP. Our palaeoclimate estimates for nine fossil floras were also more accurate, because they were in better agreement with independent paleoclimate evidence. Our study confirms that our ML models and their averaging can improve paleoclimatic reconstructions, providing a better understanding of the relationship between climate and leaf physiognomy.