Method for predictive soil mapping based on solar radiation in large flat area

A method for predictive soil mapping based on solar radiation in large flat area, the method comprising:

• capturing a response process of earth's surface heat:
• after an observation day or period is selected, a curve graph depicting a dynamic response of the surface heat to solar radiation captured by using a remote sensor of a moderate-resolution imaging spectrometer;
• constructing environmental covariates:
• quantitative analysis is performed for the dynamic response curve graph obtained in the capturing step by a mathematical method, and characteristic parameters are extracted and taken as environmental covariates;
• and
• establishing a machine learning model for predictive soil attribute mapping,
• wherein constructing the environmental covariates comprises:
• parameterizing the dynamic response curve graph of the surface to solar radiation in two aspects of magnitude and shape;
• wherein in the aspect of magnitude, it is represented by a lowest surface temperature “bottom” and a highest surface temperature “plateau”, a value of the lowest surface temperature “bottom” is an average value of surface temperatures at 1:30 am on one day and 1:30 am on next day;
• a maximum surface temperature “plateau” is an average value of surface temperatures at 10:30 am and 13:30 pm on the one day,
• wherein in the aspect of morphology, it is represented by a gradient angle index “angleslp”, and the gradient angle indexes at five time points of 1:30, 10:30, 13:30, 22:30 on the one day and 1:30 on the next day are calculated respectively, wherein the gradient angle index angleslp10:30 at 10:30 is calculated as follows:
• β
• 10
• :
• 30
• =
• cos
• -
• 1
• ⁡
• [
• a
• 2
• +
• b
• 2
• -
• c
• 2
• 2
• ⁢
• ab
• ]
• angleslp
• 10
• :
• 30
• =
• β
• 10
• :
• 30
• *
• slope
• in the formula, a, b and c are Euclidean distances between two adjacent time points of three time points of 1:30, 10:30 and 13:30 on the one day in the surface temperature changing curve graph, β10:30 is a radian angle at 10:30, “slope” is a slope of a line segment c, and a time distance between the two adjacent time points is 1;
• by the same calculation method, the gradient angle indexes, angleslp13:30 and angleslp22:30, at 13:30 and 22:30 are calculated respectively;
• performing statistics on the five environmental covariates calculated in the parameterizing step:
• the lowest surface temperature “bottom”, the highest surface temperature “plateau”, the gradient angle index angleslp10:30 at 10:30, the gradient angle index angleslp13:30 at 13:30, and the gradient angle index angleslp22:30 at 22:30;
• and
• wherein, establishing the machine learning model for the predictive soil attribute mapping comprises:
• establishing a relationship model between soil attributes and the environmental covariates by using a random forest algorithm;
• applying the established relationship model to all mapping areas:
• an environmental covariate value at each geographical location in the mapping areas is substituted into the established relationship model, to obtain a predicted value of the soil attributes at the geographical location;
• and
• obtaining a spatial distribution map of the soil attributes across all mapping areas after traversing all locations.