Home Articles Integrating remotely sensed data with an ecosystem model to estimate net primary...

Integrating remotely sensed data with an ecosystem model to estimate net primary productivity in East Asia

Wenjing Zhao, Masayuki Tamura
Social and Environmental System Division,
National Institute for Environmental Studies
16-2 Onogawa, Tsukuba, Ibaraki 305-0053, Japan
Tel: +81-298-50-2589 Fax: +81-298-58-2645
E-mail: [email protected], [email protected]

Abstract:
This paper describes a method of integrating remotely sensed data with an ecosystem model to estimate NPP in East Asia. Principles of forest biogeochemical cycles (FOREST-BGC) are used for simulating biological processes affecting NPP, such as photosynthesis, respiration, and transpiration. Input requirements for the ecosystem process model are: (1) land cover types and leaf area index from remote sensing data; (2) daily meteorological data such as maximum and minimum air temperature, incoming short-wave radiation, precipitation, and humidity; and (3) water holding capacity of soil. By incorporating the above data sources concerning all major environmental variables affecting plant growth and development, a map showing the distribution of annual NPP in East Asia in 1998 has been produced.

1. Introduction
Net primary Productivity (NPP) is a key component of the terrestrial carbon cycle, and defined as the net amount of new carbon absorbed by plants per unit space and time (Liu et al., 1999). Estimation of NPP accurately at regional or global scales is very important in studies of global climate. Liu et al. (1997) developed a Boreal Ecosystem Productivity Simulator (BEPS) based on the Forest BioGeochemical Cycles (FOREST-BGC) model (Running et al. 1988). Although the BEPS model has improved the original FOREST-BGC model for a regional scale by adding remote sensing inputs, robustness of the BEPS model should be verified for other area (or other land cover types) such as East Asia region, because it was just developed and validated for boreal ecosystem. Therefore, the primary objective of this study are: (1) to develop a model for estimating NPP in East Asia base on BEPS model, (2) to show the NPP distribution in East Asia region.

2. Study Area
The study region of East Asia encompasses a 108 km2 (11040 pixels×9000 lines) area bounded by 66°N-9°S latitude and 78°-170°E longitude (Figure 1). The climatic zone of this area ranges from subfrigid zone in the north to tropical zone in the south. Land cover types includes boreal forest, grassland, crops, needle forest, broadleaf forest, and tropical rainforest.

3. Model
The model is based on BEPS model (Liu et al., 1997), which consists of three parts: remotely sensed and meteorological inputs, ecosystem process model, and NPP output (Figure 2). The BEPS model uses principles of the Forest BioGeochemical Cycles (FOREST-BGC) model (Running et al., 1988) for quantifying the biophysical processes governing ecosystems productivity. The BEPS model modified original model in the following aspects: (1) implement of a more advanced photosynthesis model with a new temporal and spatial scaling scheme (Chen et al., 1999); (2) inclusion of an advanced canopy radiation model to describe the specific boreal canopy architecture; and (3) adjustments of biophysical and biochemical

Figure 1. Location of study area bounded by 66°N-9°S latitude and 78°-170°E longitude.
parameters for the main boreal land cover types. The BEPS model are then modified when we use it in East Asia because its limitation of specific region for boreal ecosystem. They are: (1) more flexible land cover types map (Myneni et al., 1997) for global application is adapted; (2) more robust algorithm of estimating Leaf Area Index (LAI) from remotely sensed data (Myneni et al., 1997) for global application is adapted; (3) adjustments of biophysical and biochemical parameters according to the land cover types in East Asia. The remotely sensed inputs are land cover types and LAI, and the meteorological inputs are temperature, precipitation, humidity, and radiation. The temporal interval is daily for meteorological data, 10 days for LAI, and annual for land cover. The model computes daily NPP pixel by pixel (1km2) assuming vegetation and environment conditions were uniform within each pixel, and then accumulates daily NPP to annual NPP.

4. Input Requirements for the Model

4.1 Land Cover Types
Theoretically, the relationship between NDVI (Normalized Difference Vegetation Index) and LAI depends on the land cover types. However, Loveland et al. (1991) pointed out that traditional land cover classifications based on botanical, ecological or functional metrics may be unsuitable for LAI estimation, because these classifications are not necessarily based on NDVI-LAI considerations. For instance, if several canopies have a similar or a nearly similar NDNI-LAI relationship, information on such land covers is redundant for the estimation of LAI (Myneni et al., 1997). Therefore, Myneni et al. (1997) developed a new land cover classification for global application to estimate LAI from NDVI. They classified global land covers into six types depending on their canopy structure. The structural attributes of these land covers were parameterized in terms of variables that the radiative transfer models admit. The six land cover types are: (1) grasses and cereal crops; (2) shrubs; (3) broadleaf crops; (4) savanna; (5) broadleaf forests; and (6) needle forests. In this study, we incorporate these land cover types into our model to resolve the limitation of specific region in BEPS model.

Figure 2. Framework of estimation NPP showing the major modeling steps, the input requirements, and the data’s spatial resolutions and temporal intervals prior to simulation. (Modified from Liu et al., 1997)
4.2 Leaf Area Index
LAI is a key parameter to integrate remotely sensed data with an ecosystem model. The LAI was estimated from NOAA/AVHRR 10-day composite NDVI by using the NDVI-LAI algorithm developed by Myneni et al. (1997). Similar LAI images were calculated in this study for each 10- (or 11-) day period in 1998 by using the 10- (or 11-) day NDVI composites. Atmospheric corrections were performed to NOAA/AVHRR channel 1 and 2 using 6S code before using them to estimate NDVI. The NDVI composites were produced from single-day co-registered images by using the maximum NDVI criterion to obtain cloud-free pixels.

4.3 Daily Meteorological Data
The meteorological data required by the inputs of ecosystem model include daily maximum and minimum air temperatures, incoming short-wave radiation, precipitation, and specific humidity. These data were obtained from National Center for Atmospheric Research (NCAR), USA. The gridded data were 6-hourly forecasts by the National Meteorological Center (NMC) of NCAR, using their global spectral model (the MRF model). For the maximum and minimum air temperatures, the maximum and minimum of the four 6-hourly readings were used as the daily maximum and minimum air temperatures. For the incoming short-wave radiation and precipitation, the sums of the four 6-hourly readings were used as the daily total. For the specific humidity, the average of the four 6-hourly readings was used as the daily specific humidity. The gridded data were bilinearly interpolated for each pixel of 1km2 to match the remote sensing data because their resolutions are so coarse (approximately 1-degree intervals).

4.4 Soil Data
Soil data (Soil Water Holding Capacity) was downloaded from a free access web site of Center for Global Environmental Research (CGER) of National Institute for Environmental Studies (NIES) ). This data set shows the global distribution of soil water holding capacity, at field capacity for the top soil (0-30cm), with a 1-degree latitude/longitude spatial resolution, and was derived from information on soil type and texture (FAO Soil Map of the World). This data was also bilinearly interpolated for each pixel of 1km2 to match the remote sensing data.
5. Results and Discussion

5.1 Annual NPP
Annual NPP in1998 was estimated using the above model, and the distribution map of NPP over the East Asia region in 1998 is shown in Figure 3. Except ocean, lake, and barren pixels (NPP of these areas are set to 0 according to land cover map), the mean NPP over the study region was calculated to be 511 g C m-2 yr -1 in 1998. The mean NPP of grasslands and cereal crops were the smallest (443 g C m-2 yr -1), followed by that of needle forests (467 g C m-2 yr -1), shrubs (504 g C m-2 yr -1), savannas (538 g C m-2 yr -1), and broadleaf crops (579 g C m-2 yr -1). The mean NPP of broadleaf forests (671 g C m-2 yr -1) were the largest. It is very difficult to validate our modeled NPP because of the lack of field measurement data. However, we can obtain some NPP data from NPP Database, which provided by Oak Ridge National Laboratory Distributed Active Archive Center in the web site ). The NPP database contains documented field measurements of biomass and estimated NPP for terrestrial sites worldwide, compiled from published literature and other extant data sources. From this web site, we found that some field measurements of NPP in some Country of East Asia such as Russia, India, Japan, Malaysia, Indonesia, and Thailand. The NPP values range from 26 g C m-2 yr -1 to 2793 g C m-2 yr -1, and very closed our modeled result, which ranges from 6 g C m-2 yr -1 to 2210 g C m-2 yr -1, even though the filed measurements are rather old (from 1947-1981).

5.2 Seasonal Variations of NPP
Figure 4 shows the seasonal NPP distributions in East Asia region in 1998. In the north area, the NPP became very low in period 1 (Jan. – Mar.) and period 4 (Oct. – Dec.) because the average daily air temperature is almost lower than 5°C in this area during these periods, and then make plant growth stop. However, in the south area, the NPP in period 2 (Apr. – Jun.) and period 3 (Jul. – Sep.) was lower than periods 1 and 4. This is because the average daily air temperature is always higher than 5°C through a year, and periods 2 and 3 are rainy seasons, then have not enough solar radiation for plant photosynthesis in this area. The peak of NPP occurred in period 3 in the north area and in period 4 in the south area, respectively.

6. Conclusion
The method of integrating remotely sensed data with an ecosystem model to estimate NPP in East Asia was described. By incorporating the remote sensing data, daily meteorological data and soil data concerning all major environmental variables affecting plant growth and development, the distribution map of annual NPP in East Asia region in 1998 was produced. The mean NPP of grasslands and cereal crops were the smallest (443 g C m-2 yr -1), followed by that of needle forests (467 g C m-2 yr -1), shrubs (504 g C m-2 yr -1), savannas (538 g C m-2 yr -1), and broadleaf crops (579 g C m-2 yr -1). The mean NPP of broadleaf forests (671 g C m-2 yr -1) were the largest. The results also showed the difference of seasonal variation of NPP between the north and south areas.

7. Acknowledgment
We thank Dr. Liu J. and M. J. Chen of Canada Center for Remote Sensing for providing the BEPS code and very useful suggestions to use it. We also thank Dr. R. B. Myneni of Boston University and Dr. R. R. Nemani of University of Montana for providing NDVI-LAI algorithm and global land cover map, soil NDVI map and soil type map.

References

  • Chen J. M., J. Liu, J. Cihlar, and M. L. Goulden, 1999. Daily canopy photosynthesis model through temporal and spatial scaling for remote sensing applications. Ecological Modelling, 124, 99-119.
  • Liu J., J. M. Chen, J. Cihlar, and W. M. Park, 1997. A process-based boreal ecosystem productivity simulator using remote sensing inputs. Remote Sensing of Environment, 62, 158-175.
  • Liu J., J. M. Chen, J. Cihlar, and W. Chen, 1999. Net primary productivity distribution in the BOREAS region from a process model using satellite and surface data. Journal of Geophysical Research, Vol. 104, No. D22, 27735-27754.
  • Loveland T. R., J. W. Merchant, D. O. Ohlen, and J. F. Brown, 1991. Development of a land cover characteristic data base for the conterminous U. S.. Photgramm. Eng. Remote Sens., Vol. 57, 1453-1463.
  • Myneni R. B., R. R. Nemani, and S. W. Running, 1997. Estimation of global leaf area index and absorbed par using radiative transfer models. IEEE Transactions on Geoscience and Remote Sensing, Vol. 35, No. 6, 1380-1393.
  • Running S. W., and J. C. Coughlan, 1988. A general model of forest ecosystem processes for regional applications, I, Hydrological balance, canopy gas exchange and primary production processes. Ecological Modelling, 42, 125-154.