Mineral content estimation for salt lakes on the Tibetan plateau based on the genetic algorithm-based feature selection method using Sentinel-2 imagery: A case study of the Bieruoze Co and Guopu Co lakes

Salt lakes on the Tibetan Plateau (TP) are rich in lithium (Li), boron (B) and other mineral resources, accurate assessment of content spatial distribution brine those salt is important to guide development utilization their resources. There few studies estimating TP due lack situ investigation data. This study introduced an intelligent prediction model combining a feature selection algorithm with machine learning using Sentinel-2 satellite data estimate Li, B, TDS contents Bieruoze Co Guopu TP. First, enrich spectral information, four mathematical transformations (reciprocal, logarithmic, reciprocal logarithm, first-order derivative) were applied original bands. Then, was performed genetic (GA) select optimal input variables for model. Finally, models constructed by partial least squares regression (PLSR), multiple linear (MLR), random forest (RF). The results showed that: 1) transformation provided information estimation. 2) performance estimation optimization method GA better than that based all Based optimization, MAPE GA-RF B testing set reduced 77.52%, 28.54% 36.79%, respectively. 3) Compared GA-MLR GA-PLSR models, estimated Li ( R 2 =0.99, RMSE=1.15 mg L -1 , MAPE=3.00%), =0.97, RMSE=10.65 MAPE=2.73%), =0.93, RMSE=0.60 g MAPE=1.82%) obtained performance. combination GA-based RF has excellent applicability monitoring minerals imagery