STUDY ON THE RELATIONSHIP BETWEEN PHYSICAL STRUCTURAL INTEGRALITY AND ECOSYSTEM SERVICES VALUE IN THE RIPARIAN ZONE

Riparian zone is crucial to the health of streams and their surrounding environment. A healthy riparian zone can provide food, habitats, protecting water quality and many other ecological functions and environmental benefits. Evaluating riparian quality is essential to achieve and maintain good stream health, as well as to guarantee the ecological functions that riparian areas provide. In this study, we addressed the consistency of characterizing integrality of ecosystem of a riparian zone in Northeast China with physical structural integrality (PSI) and ecosystem service value (ESV), and explored the relationship between the PSI and ESV. The procedures included (1) evaluation of PSI of the riparian zone based on remote sensing; (2) calculation of the riparian ESV based on basic evaluation units (BEUs); (3) exploration of statistical relationships between the PSI and the ESV by the performance of linear regression. The study concluded that the trend of PSI was the same as the ESV, and they were consistent in describing the quantitative trend of the riparian zone’s ecosystem integrity. There was statistically significant correlation (R =0.66, P < 0.01 level) between PSI and ESV. * Corresponding author: yqwang@uri.edu; fbl2012@126.com The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLII-3/W10, 2020 International Conference on Geomatics in the Big Data Era (ICGBD), 15–17 November 2019, Guilin, Guangxi, China This contribution has been peer-reviewed. https://doi.org/10.5194/isprs-archives-XLII-3-W10-839-2020 | © Authors 2020. CC BY 4.0 License. 839


INTRODUCTION
Riparian zones are ecosystems located along the bank of rivers, streams, or other water networks. Usually riparian zones are narrow strips of land that line the borders of water source, which are an ecological transition zone of material, energy and information exchange between land and water ecosystems (USDI Bureau of Land Management, 1998;Tang et al., 2014).When riparian zones are tapped and exploited, the market value or the direct use value is captured while other intangible ecological benefits ignored. Excessive exploitation and utilization of riparian zones will inevitably damage and weaken ecosystem service functions. Land use has been considered an indicator to evaluate riparian quality, which plays a decisive role in the maintenance of ecosystem services function (Fernandes et al., 2011;Miserendino et al., 2011;Fernández et al., 2014;Lu and He, 2014). Therefore land-use types and patterns in a riparian zone have been prone to affect the structure and function of the ecosystem. Quantification of ecosystem service value (ESV) of a riparian zone based on land use data has been considered an efficient approach to characterize its integrality of ecosystem (Li et al., 2008;Kindu et al., 2016;Li et al., 2016).
Several methodologies for assessing riparian quality and PSI existed and formed different evaluating indicator systems (Dixon et al., 2005;Munné et al., 2003;Jansen et al., 2005;Barquın et al., 2011;González et al., 2011). Riparian PSI and ESV both characterize the integrality of ecosystem, leading to two questions: (1) when riparian integrality of ecosystem is characterized by PSI and ESV, are their results consistent? (2) What is the statistical relationship between PSI and ESV?
To answer aforementioned questions, we did the followings: (1) used remote sensing method to evaluate the PSI of the riparian zone based on 520 basic evaluation units (BEUs); (2) calculated the coefficient of ESV per unit area and the ESV based on 520 BEUs in the riparian zone; (3) contrastively analyzed the trend of PSI and ESV in four measurement sections along the riparian zone; and (4) explored the statistical relationship between PSI and ESV, and further analyzed the ability of using PSI to model ESV by linear regression model.

Study area
This study focused on the 360 km riparian zone of the Second Songhua River from Fengman Reservoir to Sancha estuary. The Second Songhua River is the largest tributary and source of Songhua River, which originates in Changbai Mountain and flows through major cities and counties of the Jilin Province in Northeast China (Figure 1). The elevation of the river basin is between 54 to 2,667 meters above mean sea level. The climate of river basin is temperate continental climate with four clearly distinct seasons. The mean annual rainfall in the area is about 600-800mm.
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLII-3/W10, 2020 International Conference on Geomatics in the Big Data Era (ICGBD), 15-17 November 2019, Guilin, Guangxi, China

PSI and BEUs
The PSI is one of the important components in river health
The research area was divided into four measurement sections ( Figure 1) and each section was partitioned into small serial units, named basic evaluation units (BEUs). The river was retrieved as a single feature GIS polygon and split from mouth to source for all the main channels. Then, the polygon covering the riparian zone was dissected using lines perpendicular to the river centerlines. This process generated 520 BEUs in the study area ( Figure 1).

Riparian zone's PSI was calculated and evaluated by indicators,
including riparian stability, river connectivity and natural wetland conservation ratio, and within each indicator was composed by some sub-indicators (Table 1). We cross-checked the evaluating results based on remote sensing with field measurements. The results were assigned into one of the five existing quality classes following the technical protocol, i.e., bad good (0.6<PSI=<0.8) and better (0.8<PSI=<1.0).

Sub-indicators Indicators
Field measurement Remote sensing per unit area (kg · ha -1 ).

Growth Curve model
The ESV that human can accept is closely related to the personal willingness to pay and social-economic development level. That is, the willingness to buy ESV will change along with the economic development. So the PGC model is used to modify the value coefficients of ESV. ( Where, E is the economic value of food service per unit area after calibration with the PGC model, e is the natural logarithm, t is

Evaluated PSI of the riparian zone
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLII-3/W10, 2020 International Conference on Geomatics in the Big Data Era (ICGBD), 15-17 November 2019, Guilin, Guangxi, China the socio-economic development indicator, En is the Engel coefficient and E a is the economic value of food service per unit area.

Calculation of ESV per unit area
The ESV per unit area in the riparian zone was calculated by the equation (3), using the equivalent factor of ecosystem services values in Jilin Province and the economic value of food production of farmland ecosystem services after calibration.
Where, i is ecological function types, j is the land cover types; E is the economic value of food production of farmland ecosystem after correction by the PGC model ($ · ha -1 ); e ij is the equivalent factor of the ecosystem service function i of a land cover type j in the Table 5. E ij is the economic value per unit area of ecosystem service function i of an ecosystem j.


Where, S j is the area of land cover type j (ha), E ij is the value per unit area of ecosystem services i of land cover type j ($ · ha -1 ), i is the type of ecosystem services and j is the land cover type.

The statistical relationship between PSI and ESV
Because PSI and ESV were not dimensionally homogeneous, they measured at different scales that did not contribute equally to the analysis. In the study, variables were standardized by the equation (5) to make sure all variables contribute evenly to a scale.
We modelled the ESV using linear regression. In the model, the standardized ESV (Z-ESV) was as a dependent variable, and the standardized PSI (Z-PSI) was as an independent variable. Before performing linear regression, the Z-PSI were arcsine-square root transformed. The same transformation was also applied to the dependent variable (Z-ESV) to achieve normal distribution and improve homoscedasticity. Then, we selected the five mathematical models (such as linear, logarithmic, quadratic, cubic, exponential) to carry out curve estimation referring to the scatter diagram.

Evaluating result of PSI
The evaluation result was shown in Figure 2. To check the accuracy of the result, we used the evaluation result based on field measurements (PSI_FM) to verify the result based on remote sensing observations (PSI_RS) (Figure 2(a)).

Evaluating results of 14 measurement sites
The evaluation results based on remote sensing was that there were a little differences in some measurement sites, and the differences concentrated on the section A and D. But these differences were within 0.2, and would not affect the evaluation of an entire sections. Field measurements of three sites in the section A were higher than that the remote sensing results. The results calculated by remote sensing were higher than that by field measurement in the rest of measurement sites (Figure 2(a)).

Evaluating results of four measurement sections
The evaluating results of four measurement sections based on two methods were that the result concentrated on 0.45 ~ 0.85; the section D was still the highest and the section A was the lowest; according to the quality classes following the technical protocol, This contribution has been peer-reviewed. https://doi.org/10.5194/isprs-archives-XLII-3-W10-839-2020 | © Authors 2020. CC BY 4.0 License. section D were higher than that of other sections, and the results in the section A was the lowest in the four sections. Besides, there was an observed fluctuation in the section B (Figure 2(b) and 2(c)).  Note: 1$ (USD) = 6.21Yuan (CNY) in 2014 Table 3. The coefficients of ESV to each land cover types per unit area ($ · ha -1 ).

The coefficients of ESV per unit area
The coefficients of ESV per unit area were calculated by the Equation (1) ~ (3). The value coefficients and area of different land cover types was shown in Table 3.

ESV based on BEUs
In order to be convenient for the next to explore the relationship The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLII-3/W10, 2020 International Conference on Geomatics in the Big Data Era (ICGBD), 15-17 November 2019, Guilin, Guangxi, China between the PSI and ESV, we calculated the ESV for each individual unit of the 520 BEUs (Figure 3 (b) and 3 (c)), and selected 14 BEUs (Figure 3 (a)) from 520 BEUs, which had consistent one-to-one match the 14 measurement sites in the

The trend between PSI and ESV
The ecological integrality of the riparian zone can be

The statistical relationship between PSI and ESV
In order to further explore the statistical relationship between PSI and ESV, we tested five types of mathematical models, including linear, logarithmic, quadratic, cubic and exponential. Each of these models was utilized to carry out curve estimation referring to the scatter diagram. In the models, the dependent variable (Z-ESV) and the independent variable (Z-PSI) were both arcsinesquare root transformed. The models performed better in revealing the relationship were linear, Quadratic and cubic curve models. The value of R 2 did not appear to be much different between those three models (Table 4).
After curve estimation, the linear model was selected to model

CONCLUSION AND DISCUSSION
The evaluation results concluded that the ecosystem of section A and B were both in moderate condition (0.4 < PSI < 0.6), and the