2.3.2. Ecosystem Service Assessment
Based on the classification of ESs and the actual condition in Fuzhou, seven typical ESs were selected: (i) supply services: tea production and water yield; (ii) regulation services: soil retention, NPP and climate regulation; and (iii) ecosystem disservices: soil erosion and carbon emissions. The evaluation methods of each ecosystem service are as follows:
(1) Tea Production
It is proved that the leaf area index (LAI) has a good correlation with crop yield, and it is often used to characterize and predict crop yield [
42,
43]. Therefore, the LAI and tea production data of each county provided by the authoritative statistics department were used to draw the tea yield map of Fuzhou. In this study, tea production mainly refers to the ability of tea gardens to produce fresh tea leaves. First, we interpreted a remote sensing image of Fuzhou in 2020, acquired by Google Earth, and identified the distribution of tea gardens. Since the tea growing area is stable, it is assumed not change much within 10 years. Thus, this study took the interpreted tea gardens in 2020 as the tea gardens distribution for both 2010 and 2020, based on which tea production was calculated. Then, to map the tea production based on LAI, the tea planting distribution and tea yield of each county, we first clipped LAI according to each county of Fuzhou City. Second, the LAI attribute table of each county was exported and the sum of the LAI of each county was calculated. Finally, tea production in each pixel was calculated with Formula (1) for each county separately and then united together. The calculation method is shown in the formula:
where
is the tea yield corresponding to each grid (30 m × 30 m),
is the LAI of each grid,
is the total LAI corresponding to Fuzhou tea gardens and
is the total annual tea yield of Fuzhou.
(2) Water Yield
Water yield (WY) refers to the surface water yield in a certain area [
44]. The calculation of surface water yield is based on a simplified hydrological cycle model that ignores the influence of groundwater and is determined by many parameters, such as rainfall, evapotranspiration, soil depth and available water for plants. The water supply service is evaluated by the “Water Yield” module of InVEST3.12.0, and the calculation method is shown in Formula (2):
where
is the water yield of land-use grid
,
is the annual actual evapotranspiration of grid units and
is the annual precipitation of grid unit
.
In the water balance formula, the hypothesis formula of the Budyko hydrothermal coupling balance proposed by Fuh [
45] and Zhang [
46] is adopted to calculate the evapotranspiration of vegetation of land use/cover type
:
where
is the potential evapotranspiration and
is a non-physical parameter that characterizes the natural climatic-soil properties.
is defined as:
where,
is the reference evapotranspiration from grid unit
and
is the plant (vegetation) evapotranspiration coefficient associated with the LULC on grid unit
(
Table 2).
Using the expression proposed by Donohue [
47] in the InVEST model, ω is defined as:
where,
is the volumetric (mm) plant available water content. It is estimated as the product of the plant available water capacity (
PAWC) and the minimum of root restricting layer depth and vegetation rooting depth:
Root restricting layer depth is the soil depth at which root penetration is inhibited because of physical or chemical characteristics. Vegetation rooting depth is often given as the depth at which 95% of a vegetation type’s root biomass occurs.
PAWC is the plant available water capacity.
is an empirical constant, which captures the local precipitation pattern and additional hydrogeological characteristics and is positively correlated with
N, the number of rain events per year. According to previous studies, the
Z value is determined as 11.5 [
48]. The 1.25 term is the minimum value of
, which is the value for bare soil (when root depth is 0).
The evapotranspiration coefficient and plant root depth parameters were set according to the InVEST model guidance manual, FAO’s Handbook for Calculating Crop Evapotranspiration—Crop Water Demand and previous studies [
48,
49]. The category is defined as 1 or 0, depending on whether there is vegetation cover. When the vegetation cover is 0, the model will skip the index of plant root depth, so it is set as 1.
(3) Soil Retention and Soil Erosion
Soil erosion (SE) is the displacement of the upper layer of soil, one form of soil degradation. This natural process is caused by the dynamic activity of erosive agents, that is, water, ice, snow, air(wind), plants, animals, and humans [
36]. The soil retention (SR) function represents when ecosystems (such as farmland, forest, etc.) reduce the erosion energy of rainwater through the canopy, litter, roots, and other levels and increase soil erosion resistance to reduce soil erosion, reduce soil loss and maintain soil [
50]. This parameter is evaluated by the sediment transport ratio (SDR) module of InVEST3.12.0, which is mainly based on the universal soil loss equation (USLE), and uses various datasets, such as topography, climate, vegetation and management practices to calculate the annual average soil loss and soil retention of each land-type grid. The calculation method is shown in Formula (7):
Calculation of potential soil erosion based on geomorphology and climatic conditions:
Actual soil erosion considers vegetation cover and soil and water conservation measures, which represent the soil erosion selected for this study:
Soil retention is the difference between
RKLS and
ULSE:
where
represents the soil retention,
represents the potential soil erosion,
represents the actual soil erosion,
represents the rainfall erosivity,
represents the soil erodibility factor,
represents the slope length factor,
represents the vegetation cover factor and management factor and
represents the soil and water conservation measure factor.
The
R factor is generally calculated by the Wischmeier [
51] formula based on monthly average precipitation and annual average precipitation:
where
is the average monthly precipitation (mm) and
is the average annual precipitation (mm).
The
K factor is calculated mainly through the EPIC model [
52], and the formula is as follows:
where
is the content of sand grain (%),
is the content of silty sand (%),
is the content of clay particles (%) and
is the content of organic carbon (%).
The LS factor includes two aspects: slope length factor L and slope factor S.
The calculation of the
L factor refers to the formula proposed by Wischmeier and Smith [
53], and the specific calculation formula is as follows:
where
is the length of the horizontal slope,
is the index of slope length and
is the slope.
The calculation of the
S factor refers to the formula proposed by McCool [
54], and the specific calculation formula is as follows:
where
is the slope.
The calculation of the
C factor is mainly based on
NDVI data, and the specific calculation formula is as follows [
55]:
where
b is the vegetation coverage.
The determination of the P factor (Equation (8)) is mainly based on previous studies [
56,
57]. The P factor values of different land-use types are listed in
Table 3.
(4) NPP
NPP represents the carbon sequestration and oxygen release function by which natural ecosystems absorb CO
2 in the atmosphere to synthesize organic matter during photosynthesis, sequester carbon in plants or soil and release oxygen [
58].
The improved CASA model was used to calculate the NPP of vegetation [
59], and the calculation method is shown in the formula:
where
is the total organic matter accumulation of plants in month
at pixel
[g C/(m
2 month)],
is the effective photosynthetic radiation absorbed in month
at pixel
[MJ/(m
2 month)],
is the actual light energy utilization of plants in month
at pixel
,
is the total radiation of the sun in month
at pixel
[MJ/(m
2 month)],
is the ratio of effective photosynthetic radiation absorbed by plants in month
at pixel
, 0.5 is the ratio of effective photosynthetic radiation to total solar radiation,
and
are the influence coefficients of low temperature and high temperature stress,
is the influence coefficient of water stress and
is the maximum utilization rate of light energy under ideal conditions (%).
(5) Climate Regulation
Climate regulation (CR) is the function by which natural ecosystems absorb solar energy through the transpiration of vegetation and evaporation of the water surface to regulate the summer temperature and improve the suitability of human settlements [
60]. The climate regulation service in this study is calculated by IUEMS. Taking the total energy consumed by ecosystem transpiration and evaporation as the function quantity for regulating climate, the calculation method is shown in the formula:
where
is the total energy consumed by ecosystem transpiration (kWh/a),
is the energy consumed by ecosystem vegetation transpiration (kWh/a),
is the capacity consumed by ecosystem water surface evaporation (kWh/a),
is the heat consumed by transpiration per unit area of the class
ecosystem (KJ·m
−2·d
−1),
is the number of days in a year when the highest temperature is greater than 26℃,
is a dimensionless number equal to 3.0 (dimensionless),
is the land-use type of ecosystem (
= 1,2,3…8) (farmland, forests, grasslands, wetland, water body, sea areas, artificial surfaces and unused land),
is the water surface evaporation (m
3) and
is the latent heat of evaporation, that is, the heat required to evaporate 1 g of water (J/g).
(6) Carbon Emissions
The carbon emission (CE) disservice characterized by land-use type is the difference between the carbon source and carbon sink [
61]. In this study, artificial surfaces, cultivated land and tea gardens regarded as carbon sources, forests, grasslands, water bodies, sea areas and unused land are identified as carbon sinks. In this study, the LULC carbon emission calculation method was used to calculate carbon emissions, which can be divide into direct carbon emissions and indirect carbon emissions [
62]. The carbon emissions (carbon absorption) of cultivated land, tea gardens, forests, grasslands, wetland, water body, sea areas and unused land were calculated by the direct carbon emission coefficient method:
where
is the carbon emissions (carbon absorption) of the
land-use type (
= 1, 2, 3 …8) (farmland, tea gardens, forests, grasslands, wetland, water body, sea areas and unused land),
is the area of the
land-use type and
is the carbon emission (carbon absorption) coefficient of the
land-use type.
Based on previous studies and the actual characteristics of Fuzhou, the carbon emission coefficient of each land-use type was determined from the literature and is presented in
Table 4 [
63,
64].
Artificial surface refers to the type of surface cover formed by human activities and is covered by asphalt, concrete, sand, brick, glass and other building materials, including urban and other residential areas, industrial and mining facilities and transportation facilities [
65]. Artificial surfaces represent a large number of activities of human energy consumption, and we assume in this study that carbon emissions from fossil energy consumption are concentrated on artificial surfaces. Artificial surfaces’ carbon emissions are estimated indirectly, mainly to calculate the carbon emissions generated when coal, coke, crude oil, gasoline, kerosene, diesel oil, fuel oil, natural gas and electricity are consumed in the construction process. The calculation formula is as follows:
In this formula,
is the carbon emissions of the artificial surface,
is the consumption of the
energy,
is the standard coal conversion coefficient of the
energy, and
is the carbon emission coefficient of the
energy. The fossil energy consumption and population data were mainly obtained from the Fuzhou Statistical Yearbook. The conversion coefficient of fossil energy into standard coal and the carbon emission coefficient were mainly derived from the China Energy Statistics Yearbook and IPCC Guidelines for National Greenhouse Gas Inventory [
66]. The carbon emission coefficient of each fossil energy source is listed in
Table 5.
2.3.4. Spatial Clustering Analysis
Cluster analysis classifies a batch of sample data according to different characteristics and the degree of correlation in quality [
68]. In this study, the K-means clustering method was used to reclassify the ESs types. K-means clustering is a nonhierarchical clustering method that is widely used in ecological land surface classification [
69]. Based on the data points of Fuzhou fishing nets, the datasets of ESs were extracted, the optimal K was calculated using the R software and cluster analysis and spatial visualization were realized in SPSS and ArcGIS software.
The method first determines the number of categories to be clustered. Then, the computer preliminarily determines the original center points of each type according to the center of the data structure, calculates the distance from each record to these center points in succession, divides it into k types according to the principle of the closest distance, recalculates the center points of each type and repeats the above process according to the new center position until the preset iteration times are reached. The formula is as follows:
K centroids are randomly selected from n sample data as the initial clustering centers. The centroid is recorded as:
Optimization objectives are defined as:
The loop is started and the distance from each sample point to that of the centroid is calculated. The sample to which the centroid is closest to is assigned and k clusters are obtained:
For each cluster, the average distance of all sample points is divided into the cluster as the new centroid is calculated:
Until j converges, all clusters do not change.
(1) Determining the number of clusters (k)
In this study, the elbow rule was adopted to determine the optimal K value by finding the inflection point where the loss value decreases smoothly. The cluster evaluation index used by elbow SSE (sum of squares of errors) is the square of the sum of distances from all sample points in the dataset to their cluster centers, and the formula is as follows:
where
is the
i cluster;
is the sample points in
,
is the mean of all samples in
and
is the clustering error of all samples, which represents the clustering effect.
is the centroid of the i cluster.
Based on the above formula, the optimal k is calculated in the R software, and finally, the cluster number k = 5 is selected to form the elbow map and is shown in
Figure 3.
(2) Determining the initial clustering center
The initial cluster center is randomly selected by the SPSS software, and the final cluster center is obtained after iterative calculation.
(3) Setting the iteration threshold and calculation
After the iteration threshold is set to 100, the calculation begins. After 97 iterations, the final calculation results converge. The final clustering results and the clustering information of each cluster member are obtained.
(4) Obtaining the final clustering center