Pasture Productivity Changes In Mongolia

Erdenetuya Magsar,
Remote Sensing specialist
National Remote Sensing Center, Namhem
Juulchny Str – 5, Meteorological Bldg.
Ulaanbaatar 210646, Mongolia
Tel: 976-11-329984(o), 976-99140523(mobile),
Fax: 976-11-329968
E-mail: [email protected]

Introduction
Mongolia, with its nomadic animal husbandry or livestock farming is one of the most dependent countries on nature and climate condition. Today the number of livestock reached 26 millions a head and the pastureland occupies more than 80 per cent of the territory, which is main source for animal grazing for whole year, as forage.

Geographical position and associated climatic influences can be a negative environmental condition that affects sustainable use of land resources, especially pastoral livestock production. Vegetation condition of the country is sensitively changes upon climate change and human impacts. Within last 60 years data the annual air temperature has increased in 1.66 degrees in average and the total precipitation amount had almost no change.

Different kinds of natural hazard such as drought, flood, heavy winter (zud), land degradation, desertification, strong wind/sand storm, and steppe and forest fires, which are related with global warming in someway, have major influences on pasture condition. Increased interest to use natural resources is also affecting ecosystem condition of pastureland of the country.

The main goal of this work is to assess vegetation cover change within last 20 years with pasture condition, estimated by NOAA/NDVI 10 days composite 8 km resolution data from January 1982 to September 2001, provided from NOAA/NASA Pathfinder data set. Also we tried to integrate NDVI data with observed and simulated biomass data.

Data And Methods

Satellite data
To monitor the pasture vegetation coverage of whole territory of Mongolia used the normalized difference vegetation index (NDVI) data derived from NOAA/AVHRR data. The NDVI dataset is unique in that it is global, multi seasonal, multi annual, and multi spectral. These features are useful for environmental studies, particularly for the pasture monitoring of the country, where are large pasture and pastoral animal husbandry, which dependent on natural conditions. In this study we have used 10 days composite NOAA/NDVI 8 km resolution data from 1982 to 2001.

Pasture analyze data
Pasture plant biomass data measured at the meteorological/agro-meteorological stations and biomass data simulated by CENTURY model, are used to estimate pasture productivity for whole country. We have analyzed 40 years data of more than 60 stations and defined average biomass of different ecosystems.

In this study we have tried to establish integration of remotely sensed data with ground observation data and calculate their correlation.

The main purpose of the study was:

• to estimate climate change trend within last 60 years
• to analyse the start and end of growing season by observation data
• to use modeling for pasture plant productivity
• to monitor pasture vegetation growth by satellite data
• to analyse pasture observation data
• to integrate simulated data, derived from model with pasture observation and satellite data

In this study we have used the pasture productivity (plant biomass) data observed and simulated by ecosystem model Century and integrated with NOAA/NDVI data.

Pasture Productivity
However, the pasture of Mongolia has special characteristics as, wide spreading throughout different ecosystems with high numbers of plant species; it deeply depends on climatic situation. There are 2600 species, but only about 600 species eaten by livestock. Some research showed that the total pasture capacity of the country is 50 to 60 million sheep unit. Today, the current total livestock reached to 72 million sheep unit and pasture area reduced by 5 – 6 millions ha because of urbanization, mining industries, and tourism. The average livestock density is 54 animals per 100 ha in the country and in different natural zones.

Plant productivity of the Mongolian arid and semi arid pasture has high variability as precipitation. Mean of peak standing biomass varies from 100 to 1000 kg/ha, decrease from the North to the South.

Pasture biomass measurement is crucial to estimate forage resource. As defined Dr. Tserendash S, the peak biological biomass is 1050-1500 kg/ha in the high mountains, 1150-1940 kg/ha in the forest steppe, 650-1300 kg/ha in the steppe and 290-380 kg/ha in the desert steppe. Other scientists estimated livestock available biomass is 500 -600 kg/ha in the forest steppe, 200-400 kg/ha in the steppe and 100-200 kg/ha in the desert steppe.

According to the study, the average peak standing biomass is 590 kg/ha in the forest steppe, 300 kg/ha in the steppe, 220 kg/ha in the desert steppe and 170 kg/ha in the Altai Mountains and the desert (Table 1).

Table 1.

Pasture biomass, 100 kg/ha

Fig. 1. Peak biomass, simulated by
Century model, g/m2

The result of Century model gives simulated pasture biomass values at 899 grid points over Mongolia. From above figure we can distinguish summer peak biomass from 1 ts/ha to 25 ts/ha in different natural zones. In fact this figure could directly present the natural zones of the country. Climate change has had an effect on not only peak standing biomass but also spring biomass. Livestock survival capability is strongly depends on spring weather and forage resource for winter and spring.

The average spring potential biomass by the ecosystem model Century was estimated as 27-50 g/m2 in the forest steppe, 15-33 g/m2 in the steppe, 5-13 g/m2 in the Altai mountains and 3-6 g/m2 in the Gobi desert.

Biomass in April and May was decreased in the forest steppe and the steppe directly caused by precipitation changes for those months. The total pasture carrying capacity* was estimated as 44.5 million sheep unit based on average observed biomass. Based on the decreasing trend of peak biomass by 20-30 %, the total pasture capacity was calculated as 32.6 million sheep unit. Past 40 years the total pasture carrying capacity was drop down by 27 % because of biomass decrease.

Monitoring Of Pasture Condition From Space
Among several types of satellite data available for monitoring of global scale and NOAA AVHRR data has been selected primarily because of its high frequency, wide coverage for one pass and low cost compared to high resolution satellite data. NDVI value derived from NOAA/AVHRR has becoming the main tool to estimate and monitor vegetation dynamics for whole territory of Mongolia over long duration.

According to the 10 days NDVI composite images we can determine the temporal and spatial vegetation changes i.e. onset, growing pick and senescence time of vegetation growth. From the general dynamics of pasture condition we could say that better grazing time for the pasture animals begins from earlier May to latest October till suffering full snow coverage and the best grazing time is too short as, from July to August.

Vegetation dynamics
As mentioned, NDVI is most suitable data for global vegetation monitoring; we have analyzed the seasonal and long term vegetation dynamics. However, Mongolia has short summer season, as well as short duration of growing period, from June to early September. In general, the peak of vegetation growth fits in August. Depending on the climate situation of certain region, its pasture vegetation condition changes within the years and also within 10 days. The long term dynamic of August of 1982 – 2001 is showing in Fig. 2.

Fig. 2. Long term NDVI dynamics (August, 1982 – 2001)
In general, 1994 was quite wet year and 2001 was more drought year within above 20 years of study.

From NDVI dynamics we could estimate only vegetation condition of certain period. Also we could determine the vegetation condition by comparison with long term NDVI data. For example, by calculation of Vegetation condition index (VCI) could be distinguished the areas, where vegetation condition is good, where is normal and/or bad. By NDVI difference calculation we could present areas, where current NDVI is much better, better, normal and lower or much lower than long term average NDVI of corresponding period.

Natural zones
There are several types of ecosystem, like high mountain, mountain taiga, forest steppe, steppe, desert steppe, and desert, which are completely different from each other.

Consistent with the diverse vegetation and soil, the pasture distribution is different in ecosystems. Pasture and hay land occupies 76.5 % of the total area and 99 % of the agricultural used land of the country. However, 94896.6 thousand ha of the total pasture is distributed in the steppe, 9367.8 thousand ha is in the mountain forest, 7376.1 thousand ha – in the high mountains, 28340.5 thousand ha – in the desert and 7185.5 thousand ha is in the meadows. We have analyzed the Natural zone map of Mongolia and tried to classify this map using long term NDVI data and the result was highly correlated (Fig. 3) with original map (correlation coefficient is 0.9619).

a b
Fig. 3. Natural zones of Mongolia

a. National Atlas of Mongolia, b. NOAA/NDVI classification
This result could approve that NDVI data could be illustrate the natural zones more precisely. The mean NDVI values for each natural zone derived from time series analyzed NDVI data of 1982-2001 and in Table 2 showed NDVI values of each decade from June to August.

Table 2.

Mean NDVI of June-August for each natural zone

By the changes of NDVI value of each zone we could estimate their vegetation condition. Further NDVI analysis based above

NDVI changes
By time series analysis of NDVI data, from 1982 to 2001 and calculated long term mean values for each month and for all year over whole territory of Mongolia within various natural zones. With comparison of above mean NDVI of each zone we could assess in which month and of which year its vegetation condition was lower or better than long term average.

To correlate them with peak standing biomass data we have analyzed NDVI deviation from the mean in July and August. And there are 2 different trends, and from 1982 to 1994-1995 NDVI deviation increased 0.33-1.1 values a year (green color trend) in different zones and then it decreased stronger than its increase and it reached – 0.87 to -2.78 values a year (blue color trend) up to 2001 (Fig. 4). We need to distinguish above statistical results that, these are corresponding to some climate variability or pasture degradation, related with increase of livestock numbers or they could present only natural changes of pasture or human impacts.

Fig. 4. NDVI deviation trends in different zones (July and August, 1982 – 2001)

In order to estimate NDVI changes, we have used Two Years Difference method of respective 10 days within 1982-2001. The results showed that, which place had decrease of NDVI compare to previous year and how many years it has repeated within these 20 years. In Fig. 5 have showing the spatial distribution of NDVI changes.

Fig. 5. NDVI changes between
May – August, 1982 – 2001 (3rd 10 day)
In this figure the green color corresponds to no change or increase of NDVI from previous year within 20 years, color yellow to dark red colors showing the frequency of NDVI decrease. Yellow – lower year repetition, red – higher frequency, respectively.

During 20 years study in each 10 day within May to September occurred maximum 9 years had NDVI decreases over Mongolia. Especially, in May in southern and eastern regions, in June, July and in September (not included in Fig. 5) partly in all regions, and in August central and eastern regions had the negative influence on vegetation growth. In 24.4 – 32.7% of all territory occurred one year decrease of NDVI.

From time series, most of changes have been occurred in last decade of June and July, and all September becoming worse than previous years. In Fig. 7. showed 1-3, 3-5, 5-7, and 7-9 years frequency of NDVI decreases in each 10 days within 20 years. Over the Kherlen river basin NDVI decreasing frequency occurred mostly in August and September. The maximum of NDVI decrease was in 9 years over forest steppe and steppe zones.

Fig. 7. Frequency of NDVI decreases in each
10 days of 1982-2001.
NDVI and biomass
In order to integrate NDVI dynamics of each zone with biomass data we have analyzed 20 years NDVI dynamics separated into two decades as, from 1982 to 1991 and from 1992 to 2001. In first period the trend of NDVI in July was increased from 0.31 to 1.39, in August this value changed from -0.72 to 0.87 values per year and in second period it was from 0.96 to -1.61 and from -0.39 to -3.12 values, respectively in July and August within all ecosystems.

In this study we integrated both of observed and simulated biomass data with NDVI values over all natural zones. As mentioned, the biomass measurement is done by traditional method, 3 times a month. Simulated biomass data was calculated at 899 grid points. The mean NDVI values for each natural zone derived from time series analyzed NDVI data of 1982-2001 and biomass observation data were analyzed. Data dynamics are illustrated in Fig. 8.

Fig. 8. Mean NDVI and pasture biomass dynamics in different zones.
In general, the NDVI and pasture biomass are following the similar dynamics. The averaged by different zones, correlation coefficients between NDVI and simulated biomass was higher than correlation between NDVI and measured biomass both inside fence and in open pasture and these values reached higher in July than in August. The comparison of correlation between NDVI and both fenced and open measured biomass showed that, the correlation NDVI and open measured biomass was less than first one in some zones. But the data quility and data range could influnce to this correlation.

The values of correlation coefficients ranged from 0.45 to 0.61. The comparison of correlation between NDVI and both fenced and open measured biomass showed that, the correlation NDVI and open measured biomass was less than first one in some zones. But the data quality and data range could influence to this correlation.

Conclusion
Pasture vegetation condition highly depends on both of natural factors and human activities. The natural factors are consequent increase of air temperature (1.66oC per 60 years), insufficient amount, and various temporal and spatial distribution of annual precipitation over the territory and increase of occurrence of natural disasters. Within negative human impacts on pasture vegetation, included increase of livestock number, heavy overgrazing of pasture and mining and etc. All above factors are cause reduction of vegetation biomass in different natural zones of Mongolia. According to the ground observation data within past 40 years the total pasture carrying capacity was decreased by 27 % as well as biomass decrease.

By NDVI data of NOAA satellite data with low resolution and high frequency, could be estimated not only the pasture vegetation condition and also it could be related with both of climate and biomass data. Within 1982-2001 in each 10 day NDVI value was decreased differently in each place and occurred 9 years maximum decreases at the same place.

The traditional measurement of biomass was done since 1940s and used biomass data simulated by CENTURY model and both biomass data were correlated with NDVI data. The correlations between vegetation biomass, simulated by above model and NDVI values gives higher results than correlation between NDVI and observed vegetation biomass data, collected by agro-meteorologists at certain meteorological stations of different zones. This correlation coefficient in July reached the maximum value in each zone.

References:

1. Batima P, Dagvadorj D, “Climate Change and its Impacts in Mongolia”, Ulaanbaatar (UB), 2000.
2. Ellis J. 1992. Resent advances in Arid Land Ecology. In: Sustainable Crop-Livestock Systems for the Bolivan Highlands. C.Valdivial, Ed. Univ. of Missouri Press. pp. 1-14.
3. Erdenetuya M, “Pasture monitoring from space”, Annual reports, UB, 2000, 2001.
4. Erdenetuya M, Adyasuren Ts, Bolortsetseg B, “Vegetation Cover Monitoring in Mongolian Plateau using Remote Sensing Technology”, Proceedings of 2nd International Workshop on Terristerial Change in Mongolia, Yokohama, Japan, 2003, pp. 70-77.
5. Jambaajamts B, “Climate of Mongolia”, UB, Mongolia, 1989, pp. 169-172.
6. Natsagdorj L, Batima P, “Climate Change of Mongolia”, UB, 2002.
7. Ojima D.S, W.J. Parton, D.S.Shimel, J.M.O.Scurlock, T.G.F. Kittel. 1993. Modeling the effects of climatic and CO2 changes on grassland storage of soil C, in Water, Air, and Soil Pollution 70:643-657.